diff --git a/build_tools/android/run_benchmarks.py b/build_tools/android/run_benchmarks.py
index 53caac1..f4bddf1 100755
--- a/build_tools/android/run_benchmarks.py
+++ b/build_tools/android/run_benchmarks.py
@@ -228,16 +228,15 @@
                                                    model_benchmark_dir)
     print(f"--> benchmark: {benchmark_info} <--")
     android_relative_dir = os.path.relpath(model_benchmark_dir, model_root_dir)
-    android_flagfile_path = adb_push_to_tmp_dir(os.path.join(
-        model_benchmark_dir, MODEL_FLAGFILE_NAME),
-                                                android_relative_dir,
-                                                verbose=verbose)
+    adb_push_to_tmp_dir(os.path.join(model_benchmark_dir, MODEL_FLAGFILE_NAME),
+                        android_relative_dir,
+                        verbose=verbose)
 
     cmd = [
         "taskset",
         benchmark_info.deduce_taskset(),
         android_tool_path,
-        f"--flagfile={android_flagfile_path}",
+        f"--flagfile={MODEL_FLAGFILE_NAME}",
         f"--benchmark_repetitions={BENCHMARK_REPETITIONS}",
         "--benchmark_format=json",
     ]
@@ -317,6 +316,10 @@
     raise ValueError(f"Unrecognized GPU name: '{device_info.gpu_name}'; "
                      "need to update the map")
 
+  # Clear the benchmark directory on the Android device first just in case
+  # there are leftovers from manual or failed runs.
+  adb_execute_in_dir(["rm", "-rf", "*"], relative_dir="", verbose=args.verbose)
+
   benchmarks = filter_python_model_benchmark_suite(device_info, args.build_dir,
                                                    args.verbose)
   run_results = run_python_model_benchmark_suite(device_info,
diff --git a/colab/edge_detection.ipynb b/colab/edge_detection.ipynb
index 2952ff0..9efae40 100644
--- a/colab/edge_detection.ipynb
+++ b/colab/edge_detection.ipynb
@@ -53,28 +53,28 @@
         "colab": {
           "base_uri": "https://localhost:8080/"
         },
-        "outputId": "7c11098d-9c4a-49f9-ed44-391006c8fd2b"
+        "outputId": "c5b534f2-accf-4dfd-88e2-e2511070a94f"
       },
       "source": [
         "!python -m pip install iree-compiler-snapshot iree-runtime-snapshot iree-tools-tf-snapshot -f https://github.com/google/iree/releases"
       ],
-      "execution_count": 2,
+      "execution_count": 1,
       "outputs": [
         {
           "output_type": "stream",
           "text": [
             "Looking in links: https://github.com/google/iree/releases\n",
             "Collecting iree-compiler-snapshot\n",
-            "\u001b[?25l  Downloading https://github.com/google/iree/releases/download/snapshot-20210107.21/iree_compiler_snapshot-20210107.21-py3-none-manylinux2014_x86_64.whl (27.8MB)\n",
-            "\u001b[K     |████████████████████████████████| 27.9MB 168kB/s \n",
+            "\u001b[?25l  Downloading https://github.com/google/iree/releases/download/snapshot-20210530.309/iree_compiler_snapshot-20210530.309-py3-none-manylinux2014_x86_64.whl (33.2MB)\n",
+            "\u001b[K     |████████████████████████████████| 33.2MB 121kB/s \n",
             "\u001b[?25hCollecting iree-runtime-snapshot\n",
-            "\u001b[?25l  Downloading https://github.com/google/iree/releases/download/snapshot-20210107.21/iree_runtime_snapshot-20210107.21-cp36-cp36m-manylinux2014_x86_64.whl (1.0MB)\n",
-            "\u001b[K     |████████████████████████████████| 1.0MB 46.3MB/s \n",
+            "\u001b[?25l  Downloading https://github.com/google/iree/releases/download/snapshot-20210530.309/iree_runtime_snapshot-20210530.309-cp37-cp37m-manylinux2014_x86_64.whl (522kB)\n",
+            "\u001b[K     |████████████████████████████████| 532kB 49.9MB/s \n",
             "\u001b[?25hCollecting iree-tools-tf-snapshot\n",
-            "\u001b[?25l  Downloading https://github.com/google/iree/releases/download/snapshot-20210107.21/iree_tools_tf_snapshot-20210107.21-py3-none-manylinux2014_x86_64.whl (41.4MB)\n",
-            "\u001b[K     |████████████████████████████████| 41.4MB 99kB/s \n",
+            "\u001b[?25l  Downloading https://github.com/google/iree/releases/download/snapshot-20210530.309/iree_tools_tf_snapshot-20210530.309-py3-none-manylinux2014_x86_64.whl (52.1MB)\n",
+            "\u001b[K     |████████████████████████████████| 52.1MB 90kB/s \n",
             "\u001b[?25hInstalling collected packages: iree-compiler-snapshot, iree-runtime-snapshot, iree-tools-tf-snapshot\n",
-            "Successfully installed iree-compiler-snapshot-20210107.21 iree-runtime-snapshot-20210107.21 iree-tools-tf-snapshot-20210107.21\n"
+            "Successfully installed iree-compiler-snapshot-20210530.309 iree-runtime-snapshot-20210530.309 iree-tools-tf-snapshot-20210530.309\n"
           ],
           "name": "stdout"
         }
@@ -100,7 +100,7 @@
         "from iree.compiler import compile_str\n",
         "from iree.compiler import tf as tfc"
       ],
-      "execution_count": 3,
+      "execution_count": 2,
       "outputs": []
     },
     {
@@ -112,10 +112,10 @@
         "#@title Setup Artifacts Directory\n",
         "\n",
         "# Used in the low-level compilation section.\n",
-        "ARITFACTS_DIR = os.path.join(tempfile.gettempdir(), \"iree\", \"colab_artifacts\")\n",
-        "os.makedirs(ARITFACTS_DIR, exist_ok=True)"
+        "ARTIFACTS_DIR = os.path.join(tempfile.gettempdir(), \"iree\", \"colab_artifacts\")\n",
+        "os.makedirs(ARTIFACTS_DIR, exist_ok=True)"
       ],
-      "execution_count": 4,
+      "execution_count": 3,
       "outputs": []
     },
     {
@@ -144,7 +144,7 @@
         "\n",
         "tf_module = EdgeDetectionModule()"
       ],
-      "execution_count": 5,
+      "execution_count": 4,
       "outputs": []
     },
     {
@@ -156,7 +156,7 @@
           "height": 281
         },
         "id": "Q2no8DO_s125",
-        "outputId": "b166f401-5bf1-4a91-ebab-8177635f836a"
+        "outputId": "fbbc7e5d-8826-4ae2-a15f-810b75eac54a"
       },
       "source": [
         "#@title Load a test image of a [labrador](https://commons.wikimedia.org/wiki/File:YellowLabradorLooking_new.jpg) and run the module with TF\n",
@@ -192,7 +192,7 @@
         "# Plot the input and output images\n",
         "show_images(image, tf_edges)"
       ],
-      "execution_count": 6,
+      "execution_count": 5,
       "outputs": [
         {
           "output_type": "stream",
@@ -238,7 +238,7 @@
         "backend_choice = backend_choice.split(\" \")[0]\n",
         "backend = module_utils.BackendInfo(backend_choice)"
       ],
-      "execution_count": 7,
+      "execution_count": 6,
       "outputs": []
     },
     {
@@ -246,10 +246,10 @@
       "metadata": {
         "colab": {
           "base_uri": "https://localhost:8080/",
-          "height": 332
+          "height": 366
         },
         "id": "2OPBr0eGroE0",
-        "outputId": "d0626857-a2fe-4dd3-b157-2cdd7982f990"
+        "outputId": "0e155db6-8a34-4a18-e216-2a1b618ea9f1"
       },
       "source": [
         "#@title Compile and Run the EdgeDetectionModule with IREE.\n",
@@ -259,29 +259,31 @@
         "iree_edges = module.edge_detect_sobel_operator(image)\n",
         "show_images(image, iree_edges)"
       ],
-      "execution_count": 8,
+      "execution_count": 7,
       "outputs": [
         {
           "output_type": "stream",
           "text": [
-            "INFO:tensorflow:Assets written to: /tmp/tmplv_9_sai.sm/assets\n"
+            "INFO:tensorflow:Assets written to: /tmp/tmp173tk864.sm/assets\n"
           ],
           "name": "stdout"
         },
         {
           "output_type": "stream",
           "text": [
-            "2021-01-07 18:28:56.664173: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:32] Reading SavedModel from: /tmp/tmplv_9_sai.sm\n",
-            "2021-01-07 18:28:56.664738: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:55] Reading SavedModel debug info (if present) from: /tmp/tmplv_9_sai.sm\n",
-            "Created IREE driver vmvx: <iree.runtime.binding.HalDriver object at 0x7fe5a1372810>\n",
-            "SystemContext driver=<iree.runtime.binding.HalDriver object at 0x7fe5a1372810>\n"
+            "2021-06-04 16:34:58.905213: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:37] Reading SavedModel from: /tmp/tmp173tk864.sm\n",
+            "2021-06-04 16:34:58.905584: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:60] Reading SavedModel debug info (if present) from: /tmp/tmp173tk864.sm\n",
+            "2021-06-04 16:34:58.926014: I external/org_tensorflow/tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  SSE3 SSE4.1 SSE4.2 AVX AVX2 FMA\n",
+            "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n",
+            "Created IREE driver vmvx: <iree.runtime.binding.HalDriver object at 0x7f6cc5e38ef0>\n",
+            "SystemContext driver=<iree.runtime.binding.HalDriver object at 0x7f6cc5e38ef0>\n"
           ],
           "name": "stderr"
         },
         {
           "output_type": "display_data",
           "data": {
-            "image/png": "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\n",
+            "image/png": "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\n",
             "text/plain": [
               "<Figure size 432x288 with 2 Axes>"
             ]
@@ -316,7 +318,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "6YGqN2uqP_7P",
-        "outputId": "905ae463-918a-483d-8d11-3dd463443ee0"
+        "outputId": "3f1d6133-4f17-4c8d-c7cc-cd8e016eee94"
       },
       "source": [
         "#@title Construct a module containing the edge detection function\n",
@@ -332,39 +334,39 @@
         "compiler_module = tfc.compile_module(EdgeDetectionModule(), import_only=True)\n",
         "print(\"Edge Detection MLIR: \", compiler_module.decode('utf-8'))\n",
         "\n",
-        "edge_detection_mlir_path = os.path.join(ARITFACTS_DIR, \"edge_detection.mlir\")\n",
+        "edge_detection_mlir_path = os.path.join(ARTIFACTS_DIR, \"edge_detection.mlir\")\n",
         "with open(edge_detection_mlir_path, \"wt\") as output_file:\n",
         "  output_file.write(compiler_module.decode('utf-8'))\n",
         "print(f\"Wrote MLIR to path '{edge_detection_mlir_path}'\")"
       ],
-      "execution_count": 9,
+      "execution_count": 8,
       "outputs": [
         {
           "output_type": "stream",
           "text": [
-            "INFO:tensorflow:Assets written to: /tmp/tmp4m3azah4.sm/assets\n",
-            "Edge Detection MLIR:  module  {\n",
-            "  func @edge_detect_sobel_operator(%arg0: tensor<1x128x128x1xf32> {tf._user_specified_name = \"image\"}) -> tensor<1x128x128x1xf32> attributes {iree.module.export, iree.reflection = {abi = \"sip\", abiv = 1 : i32, sip = \"I8!S5!k0_0R3!_0\"}, tf._input_shapes = [#tf.shape<1x128x128x1>]} {\n",
+            "INFO:tensorflow:Assets written to: /tmp/tmp247st3qi.sm/assets\n",
+            "Edge Detection MLIR:  #loc0 = loc(unknown)\n",
+            "module  {\n",
+            "  func @edge_detect_sobel_operator(%arg0: tensor<1x128x128x1xf32> {tf._user_specified_name = \"image\"} loc(unknown)) -> tensor<1x128x128x1xf32> attributes {iree.module.export, iree.reflection = {abi = \"sip\", abiv = 1 : i32, sip = \"I8!S5!k0_0R3!_0\"}, tf._construction_context = \"kEagerRuntime\", tf._input_shapes = [#tf.shape<1x128x128x1>]} {\n",
             "    %0 = mhlo.constant dense<[[[[-1.000000e+00]], [[0.000000e+00]], [[1.000000e+00]]], [[[-2.000000e+00]], [[0.000000e+00]], [[2.000000e+00]]], [[[-1.000000e+00]], [[0.000000e+00]], [[1.000000e+00]]]]> : tensor<3x3x1x1xf32> loc(#loc1)\n",
             "    %1 = mhlo.constant dense<[[[[1.000000e+00]], [[2.000000e+00]], [[1.000000e+00]]], [[[0.000000e+00]], [[0.000000e+00]], [[0.000000e+00]]], [[[-1.000000e+00]], [[-2.000000e+00]], [[-1.000000e+00]]]]> : tensor<3x3x1x1xf32> loc(#loc2)\n",
-            "    %2 = \"mhlo.convolution\"(%arg0, %0) {batch_group_count = 1 : i64, dimension_numbers = {input_batch_dimension = 0 : i64, input_feature_dimension = 3 : i64, input_spatial_dimensions = dense<[1, 2]> : tensor<2xi64>, kernel_input_feature_dimension = 2 : i64, kernel_output_feature_dimension = 3 : i64, kernel_spatial_dimensions = dense<[0, 1]> : tensor<2xi64>, output_batch_dimension = 0 : i64, output_feature_dimension = 3 : i64, output_spatial_dimensions = dense<[1, 2]> : tensor<2xi64>}, feature_group_count = 1 : i64, padding = dense<1> : tensor<2x2xi64>, rhs_dilation = dense<1> : tensor<2xi64>, window_strides = dense<1> : tensor<2xi64>} : (tensor<1x128x128x1xf32>, tensor<3x3x1x1xf32>) -> tensor<1x128x128x1xf32> loc(#loc3)\n",
+            "    %2 = mhlo.convolution(%arg0, %0) dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f], window = {stride = [1, 1], pad = [[1, 1], [1, 1]], rhs_dilate = [1, 1]} {batch_group_count = 1 : i64, feature_group_count = 1 : i64} : (tensor<1x128x128x1xf32>, tensor<3x3x1x1xf32>) -> tensor<1x128x128x1xf32> loc(#loc3)\n",
             "    %3 = mhlo.multiply %2, %2 : tensor<1x128x128x1xf32> loc(#loc4)\n",
-            "    %4 = \"mhlo.convolution\"(%arg0, %1) {batch_group_count = 1 : i64, dimension_numbers = {input_batch_dimension = 0 : i64, input_feature_dimension = 3 : i64, input_spatial_dimensions = dense<[1, 2]> : tensor<2xi64>, kernel_input_feature_dimension = 2 : i64, kernel_output_feature_dimension = 3 : i64, kernel_spatial_dimensions = dense<[0, 1]> : tensor<2xi64>, output_batch_dimension = 0 : i64, output_feature_dimension = 3 : i64, output_spatial_dimensions = dense<[1, 2]> : tensor<2xi64>}, feature_group_count = 1 : i64, padding = dense<1> : tensor<2x2xi64>, rhs_dilation = dense<1> : tensor<2xi64>, window_strides = dense<1> : tensor<2xi64>} : (tensor<1x128x128x1xf32>, tensor<3x3x1x1xf32>) -> tensor<1x128x128x1xf32> loc(#loc5)\n",
+            "    %4 = mhlo.convolution(%arg0, %1) dim_numbers = [b, 0, 1, f]x[0, 1, i, o]->[b, 0, 1, f], window = {stride = [1, 1], pad = [[1, 1], [1, 1]], rhs_dilate = [1, 1]} {batch_group_count = 1 : i64, feature_group_count = 1 : i64} : (tensor<1x128x128x1xf32>, tensor<3x3x1x1xf32>) -> tensor<1x128x128x1xf32> loc(#loc5)\n",
             "    %5 = mhlo.multiply %4, %4 : tensor<1x128x128x1xf32> loc(#loc6)\n",
             "    %6 = mhlo.add %3, %5 : tensor<1x128x128x1xf32> loc(#loc7)\n",
             "    %7 = \"mhlo.sqrt\"(%6) : (tensor<1x128x128x1xf32>) -> tensor<1x128x128x1xf32> loc(#loc8)\n",
             "    return %7 : tensor<1x128x128x1xf32> loc(#loc0)\n",
             "  } loc(#loc0)\n",
             "} loc(#loc0)\n",
-            "#loc0 = loc(unknown)\n",
-            "#loc1 = loc(callsite(\"Const@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/constant_op.py\":287:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/constant_op.py\":265:0 at callsite(\"<ipython-input-5-814f817fc65c>\":7:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc2 = loc(callsite(\"Const_1@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/constant_op.py\":287:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/constant_op.py\":265:0 at callsite(\"<ipython-input-5-814f817fc65c>\":11:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc3 = loc(callsite(\"Conv2D@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/nn_ops.py\":2170:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/util/dispatch.py\":201:0 at callsite(\"<ipython-input-5-814f817fc65c>\":15:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc4 = loc(callsite(\"mul@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py\":1496:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py\":1164:0 at callsite(\"<ipython-input-5-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc5 = loc(callsite(\"Conv2D_1@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/nn_ops.py\":2170:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/util/dispatch.py\":201:0 at callsite(\"<ipython-input-5-814f817fc65c>\":16:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc6 = loc(callsite(\"mul_1@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py\":1496:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py\":1164:0 at callsite(\"<ipython-input-5-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc7 = loc(callsite(\"add@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/util/dispatch.py\":201:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py\":1164:0 at callsite(\"<ipython-input-5-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
-            "#loc8 = loc(callsite(\"Sqrt@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/ops/math_ops.py\":4903:0) at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/util/dispatch.py\":201:0 at callsite(\"<ipython-input-5-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":973:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3887:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/def_function.py\":634:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/framework/func_graph.py\":990:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3206:0 at callsite(\"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":3361:0 at \"/usr/local/lib/python3.6/dist-packages/tensorflow/python/eager/function.py\":2969:0))))))))))\n",
+            "#loc1 = loc(callsite(\"Const@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/constant_op.py\":287:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/constant_op.py\":265:0 at callsite(\"<ipython-input-4-814f817fc65c>\":7:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc2 = loc(callsite(\"Const_1@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/constant_op.py\":287:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/constant_op.py\":265:0 at callsite(\"<ipython-input-4-814f817fc65c>\":11:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc3 = loc(callsite(\"Conv2D@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/nn_ops.py\":2169:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/dispatch.py\":206:0 at callsite(\"<ipython-input-4-814f817fc65c>\":15:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc4 = loc(callsite(\"mul@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":1575:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":1234:0 at callsite(\"<ipython-input-4-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc5 = loc(callsite(\"Conv2D_1@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/nn_ops.py\":2169:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/dispatch.py\":206:0 at callsite(\"<ipython-input-4-814f817fc65c>\":16:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc6 = loc(callsite(\"mul_1@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":1575:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":1234:0 at callsite(\"<ipython-input-4-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc7 = loc(callsite(\"add@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/dispatch.py\":206:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":1234:0 at callsite(\"<ipython-input-4-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc8 = loc(callsite(\"Sqrt@__inference_edge_detect_sobel_operator_111\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":5086:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/util/dispatch.py\":206:0 at callsite(\"<ipython-input-4-814f817fc65c>\":17:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
             "\n",
             "\n",
             "Wrote MLIR to path '/tmp/iree/colab_artifacts/edge_detection.mlir'\n"
@@ -374,8 +376,10 @@
         {
           "output_type": "stream",
           "text": [
-            "2021-01-07 18:28:57.128635: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:32] Reading SavedModel from: /tmp/tmp4m3azah4.sm\n",
-            "2021-01-07 18:28:57.129303: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:55] Reading SavedModel debug info (if present) from: /tmp/tmp4m3azah4.sm\n"
+            "2021-06-04 16:34:59.529912: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:37] Reading SavedModel from: /tmp/tmp247st3qi.sm\n",
+            "2021-06-04 16:34:59.530488: I external/org_tensorflow/tensorflow/cc/saved_model/bundle_v2.cc:60] Reading SavedModel debug info (if present) from: /tmp/tmp247st3qi.sm\n",
+            "2021-06-04 16:34:59.552411: I external/org_tensorflow/tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations:  SSE3 SSE4.1 SSE4.2 AVX AVX2 FMA\n",
+            "To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.\n"
           ],
           "name": "stderr"
         }
@@ -388,7 +392,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "Ytvb5Gx_EFJl",
-        "outputId": "3673ec15-2bbf-4680-d401-a2a27d904597"
+        "outputId": "d6e19c31-86b8-42ba-d797-3a9dd5e66ea1"
       },
       "source": [
         "#@title Compile and prepare to test the edge detection module\n",
@@ -401,13 +405,13 @@
         "ctx = ireert.SystemContext(config=config)\n",
         "ctx.add_vm_module(vm_module)"
       ],
-      "execution_count": 10,
+      "execution_count": 9,
       "outputs": [
         {
           "output_type": "stream",
           "text": [
-            "Created IREE driver vmvx: <iree.runtime.binding.HalDriver object at 0x7fe5a997a7a0>\n",
-            "SystemContext driver=<iree.runtime.binding.HalDriver object at 0x7fe5a997a7a0>\n"
+            "Created IREE driver vmvx: <iree.runtime.binding.HalDriver object at 0x7f6d1b6f75b0>\n",
+            "SystemContext driver=<iree.runtime.binding.HalDriver object at 0x7f6d1b6f75b0>\n"
           ],
           "name": "stderr"
         }
@@ -421,7 +425,7 @@
           "height": 247
         },
         "id": "OUUXxol7wl-f",
-        "outputId": "55174ed1-4681-4ddf-cb66-8b6a5dcfe308"
+        "outputId": "9398b352-9c21-475c-c3b4-30ea7d49ca44"
       },
       "source": [
         "edge_detect_sobel_operator_f = ctx.modules.module[\"edge_detect_sobel_operator\"]\n",
@@ -430,12 +434,12 @@
         "\n",
         "show_images(image, low_level_iree_edges)"
       ],
-      "execution_count": 11,
+      "execution_count": 10,
       "outputs": [
         {
           "output_type": "display_data",
           "data": {
-            "image/png": "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\n",
+            "image/png": "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\n",
             "text/plain": [
               "<Figure size 432x288 with 2 Axes>"
             ]
@@ -448,4 +452,4 @@
       ]
     }
   ]
-}
+}
\ No newline at end of file
diff --git a/colab/tensorflow_hub_import.ipynb b/colab/tensorflow_hub_import.ipynb
index 417ddad..a4d7209 100644
--- a/colab/tensorflow_hub_import.ipynb
+++ b/colab/tensorflow_hub_import.ipynb
@@ -70,7 +70,7 @@
         "%%capture\n",
         "!python -m pip install iree-compiler-snapshot iree-runtime-snapshot iree-tools-tf-snapshot -f https://github.com/google/iree/releases"
       ],
-      "execution_count": 2,
+      "execution_count": 1,
       "outputs": []
     },
     {
@@ -80,7 +80,7 @@
         "colab": {
           "base_uri": "https://localhost:8080/"
         },
-        "outputId": "0016f7c2-c0f5-4339-d852-81019c4b8398"
+        "outputId": "15da7042-578c-4cc9-8c54-e1e22d4fc6f1"
       },
       "source": [
         "import os\n",
@@ -98,12 +98,12 @@
         "os.makedirs(ARTIFACTS_DIR, exist_ok=True)\n",
         "print(f\"Using artifacts directory '{ARTIFACTS_DIR}'\")"
       ],
-      "execution_count": 3,
+      "execution_count": 2,
       "outputs": [
         {
           "output_type": "stream",
           "text": [
-            "TensorFlow version:  2.4.1\n",
+            "TensorFlow version:  2.5.0\n",
             "Using artifacts directory '/tmp/iree/colab_artifacts'\n"
           ],
           "name": "stdout"
@@ -130,7 +130,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "7fd0vmnloZo9",
-        "outputId": "b5751219-0e88-4aa4-c4e2-e8645309e9e0"
+        "outputId": "71ac2d17-d7ea-4982-c3d6-f709e394a37e"
       },
       "source": [
         "#@title Download the pretrained model\n",
@@ -141,7 +141,7 @@
         "model_path = hub.resolve(HUB_PATH)\n",
         "print(f\"Downloaded model from tfhub to path: '{model_path}'\")"
       ],
-      "execution_count": 4,
+      "execution_count": 3,
       "outputs": [
         {
           "output_type": "stream",
@@ -175,7 +175,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "qiO66oEYQmsd",
-        "outputId": "b04179cf-f9dd-4624-a4df-db7190ec8e10"
+        "outputId": "23dabcab-5e97-488f-a133-7614666f6609"
       },
       "source": [
         "#@title Check for serving signatures\n",
@@ -192,7 +192,7 @@
         "print(\"Checking for signature_defs using saved_model_cli:\\n\")\n",
         "!saved_model_cli show --dir {model_path} --tag_set serve --signature_def serving_default"
       ],
-      "execution_count": 5,
+      "execution_count": 4,
       "outputs": [
         {
           "output_type": "stream",
@@ -230,7 +230,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "OlDG2OuqOBGC",
-        "outputId": "61c77c94-3bc2-426c-c90c-78926a3d2cfa"
+        "outputId": "0e3be224-a373-441f-cb8f-aaf2b19e0d26"
       },
       "source": [
         "#@title Look up input signatures to use when exporting\n",
@@ -242,7 +242,7 @@
         "!saved_model_cli show --dir {model_path} --all \\\n",
         "    2> /dev/null | grep \"inputs: TensorSpec\" | tail -n 1"
       ],
-      "execution_count": 6,
+      "execution_count": 5,
       "outputs": [
         {
           "output_type": "stream",
@@ -260,7 +260,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "gnb4HhMmkgiT",
-        "outputId": "5278de21-0c74-4927-b8cb-21ca73fd43ed"
+        "outputId": "9a223ee5-cc52-4580-be88-5073a75dd97a"
       },
       "source": [
         "#@title Re-export the model using the known signature\n",
@@ -290,7 +290,7 @@
         "print(\"Checking for signature_defs using saved_model_cli:\\n\")\n",
         "!saved_model_cli show --dir {resaved_model_path} --tag_set serve --signature_def serving_default"
       ],
-      "execution_count": 7,
+      "execution_count": 6,
       "outputs": [
         {
           "output_type": "stream",
@@ -336,7 +336,7 @@
           "base_uri": "https://localhost:8080/"
         },
         "id": "GLkjlHE5mdmg",
-        "outputId": "4e25eabb-0355-4c77-f4c0-6c7a9f9ea8e3"
+        "outputId": "d174d83d-350e-490e-f5e8-d4419f0c633a"
       },
       "source": [
         "#@title Import from SavedModel\n",
@@ -370,7 +370,7 @@
         "print(f\"Saved tf_input to        '{tf_input}'\")\n",
         "print(f\"Saved iree_input to      '{iree_input}'\")"
       ],
-      "execution_count": 8,
+      "execution_count": 7,
       "outputs": [
         {
           "output_type": "stream",
@@ -388,34 +388,33 @@
       "metadata": {
         "colab": {
           "base_uri": "https://localhost:8080/",
-          "height": 119
+          "height": 102
         },
         "id": "IEJAzOb5qASI",
-        "outputId": "8d7c9b19-8891-43b2-de20-ebeeeafcff70"
+        "outputId": "3515b4ef-fd6a-4f14-ecd1-0676e246a089"
       },
       "source": [
         "#@title Download compilation artifacts\n",
         "\n",
-        "ARTIFACTS_ZIP = \"/tmp/mobilenet_colab_arficats.zip\"\n",
+        "ARTIFACTS_ZIP = \"/tmp/mobilenet_colab_artifacts.zip\"\n",
         "\n",
         "print(f\"Zipping '{ARTIFACTS_DIR}' to '{ARTIFACTS_ZIP}' for download...\")\n",
-        "!zip -r {ARTIFACTS_ZIP} {ARTIFACTS_DIR}\n",
+        "!cd {ARTIFACTS_DIR} && zip -r {ARTIFACTS_ZIP} .\n",
         "\n",
         "# Note: you can also download files using the file explorer on the left\n",
         "from google.colab import files\n",
         "print(\"Downloading the artifacts zip file...\")\n",
         "files.download(ARTIFACTS_ZIP)"
       ],
-      "execution_count": 9,
+      "execution_count": 8,
       "outputs": [
         {
           "output_type": "stream",
           "text": [
-            "Zipping '/tmp/iree/colab_artifacts' to '/tmp/mobilenet_colab_arficats.zip' for download...\n",
-            "  adding: tmp/iree/colab_artifacts/ (stored 0%)\n",
-            "  adding: tmp/iree/colab_artifacts/mobilenet_v2_tf_input.mlir (deflated 47%)\n",
-            "  adding: tmp/iree/colab_artifacts/mobilenet_v2.vmfb (deflated 8%)\n",
-            "  adding: tmp/iree/colab_artifacts/mobilenet_v2_iree_input.mlir (deflated 47%)\n",
+            "Zipping '/tmp/iree/colab_artifacts' to '/tmp/mobilenet_colab_artifacts.zip' for download...\n",
+            "  adding: mobilenet_v2_iree_input.mlir (deflated 47%)\n",
+            "  adding: mobilenet_v2.vmfb (deflated 8%)\n",
+            "  adding: mobilenet_v2_tf_input.mlir (deflated 47%)\n",
             "Downloading the artifacts zip file...\n"
           ],
           "name": "stdout"
@@ -478,7 +477,7 @@
           "output_type": "display_data",
           "data": {
             "application/javascript": [
-              "download(\"download_23ce7642-800c-4e5e-afdb-7c6541fc9cd9\", \"mobilenet_colab_arficats.zip\", 43802015)"
+              "download(\"download_64f7173f-1d8f-42f2-b184-0c2feb0700f4\", \"mobilenet_colab_artifacts.zip\", 43803643)"
             ],
             "text/plain": [
               "<IPython.core.display.Javascript object>"
@@ -491,4 +490,4 @@
       ]
     }
   ]
-}
+}
\ No newline at end of file
diff --git a/integrations/tensorflow/e2e/math/BUILD b/integrations/tensorflow/e2e/math/BUILD
index af7c0f1..544a014 100644
--- a/integrations/tensorflow/e2e/math/BUILD
+++ b/integrations/tensorflow/e2e/math/BUILD
@@ -292,7 +292,6 @@
     "asinh",
     "atan",
     "atan2",
-    "atanh",
     "bessel_i0",
     "bessel_i0e",
     "bessel_i1",
diff --git a/integrations/tensorflow/e2e/math/CMakeLists.txt b/integrations/tensorflow/e2e/math/CMakeLists.txt
index 419129a..7827871 100644
--- a/integrations/tensorflow/e2e/math/CMakeLists.txt
+++ b/integrations/tensorflow/e2e/math/CMakeLists.txt
@@ -141,7 +141,6 @@
     ",,asinh,,,iree_vulkan"
     ",,atan,,,iree_vulkan"
     ",,atan2,,,iree_vulkan"
-    ",,atanh,,,iree_vulkan"
     ",,bessel_i0,,,iree_vulkan"
     ",,bessel_i0e,,,iree_vulkan"
     ",,bessel_i1,,,iree_vulkan"
diff --git a/integrations/tensorflow/iree_tf_compiler/MHLO/BUILD b/integrations/tensorflow/iree_tf_compiler/MHLO/BUILD
index 986b490..4dff91d 100644
--- a/integrations/tensorflow/iree_tf_compiler/MHLO/BUILD
+++ b/integrations/tensorflow/iree_tf_compiler/MHLO/BUILD
@@ -37,7 +37,6 @@
         "@iree//iree/compiler/Dialect/Flow/IR",
         "@iree//iree/compiler/Dialect/Flow/Transforms",
         "@iree//iree/compiler/Dialect/IREE/IR",
-        "@iree//iree/compiler/Dialect/Shape/Conversion",
         "@iree//iree/compiler/Dialect/Shape/Transforms",
         "@llvm-project//llvm:Support",
         "@llvm-project//mlir:IR",
diff --git a/integrations/tensorflow/iree_tf_compiler/MHLO/Passes.cpp b/integrations/tensorflow/iree_tf_compiler/MHLO/Passes.cpp
index 2719f23..50def9d 100644
--- a/integrations/tensorflow/iree_tf_compiler/MHLO/Passes.cpp
+++ b/integrations/tensorflow/iree_tf_compiler/MHLO/Passes.cpp
@@ -16,7 +16,6 @@
 
 #include "iree/compiler/Conversion/Passes.h"
 #include "iree/compiler/Dialect/Flow/Transforms/Passes.h"
-#include "iree/compiler/Dialect/Shape/Conversion/Passes.h"
 #include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
 #include "mlir-hlo/Dialect/mhlo/transforms/passes.h"
 #include "mlir/Conversion/SCFToStandard/SCFToStandard.h"
diff --git a/integrations/tensorflow/iree_tf_compiler/TF/BUILD b/integrations/tensorflow/iree_tf_compiler/TF/BUILD
index fc63547..f241293 100644
--- a/integrations/tensorflow/iree_tf_compiler/TF/BUILD
+++ b/integrations/tensorflow/iree_tf_compiler/TF/BUILD
@@ -42,7 +42,6 @@
         "@iree//iree/compiler/Dialect/Flow/IR",
         "@iree//iree/compiler/Dialect/Flow/Transforms",
         "@iree//iree/compiler/Dialect/IREE/IR",
-        "@iree//iree/compiler/Dialect/Shape/Conversion",
         "@iree//iree/compiler/Dialect/Shape/Transforms",
         "@llvm-project//llvm:Support",
         "@llvm-project//mlir:IR",
diff --git a/integrations/tensorflow/iree_tf_compiler/TF/Passes.cpp b/integrations/tensorflow/iree_tf_compiler/TF/Passes.cpp
index 147ea19..af3996c 100644
--- a/integrations/tensorflow/iree_tf_compiler/TF/Passes.cpp
+++ b/integrations/tensorflow/iree_tf_compiler/TF/Passes.cpp
@@ -7,7 +7,6 @@
 #include "iree_tf_compiler/TF/Passes.h"
 
 #include "iree/compiler/Dialect/Flow/Transforms/Passes.h"
-#include "iree/compiler/Dialect/Shape/Conversion/Passes.h"
 #include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
 #include "iree_tf_compiler/MHLO/Passes.h"
 #include "iree_tf_compiler/dialect/tf_strings/ir/dialect.h"
diff --git a/integrations/tensorflow/iree_tf_compiler/TFL/BUILD b/integrations/tensorflow/iree_tf_compiler/TFL/BUILD
index 0976ab8..1395442 100644
--- a/integrations/tensorflow/iree_tf_compiler/TFL/BUILD
+++ b/integrations/tensorflow/iree_tf_compiler/TFL/BUILD
@@ -27,7 +27,6 @@
     deps = [
         "@iree//iree/compiler/Dialect/Flow/IR",
         "@iree//iree/compiler/Dialect/IREE/IR",
-        "@iree//iree/compiler/Dialect/Shape/Conversion",
         "@iree//iree/compiler/Dialect/Shape/IR",
         "@iree//iree/compiler/Dialect/Shape/Transforms",
         "@llvm-project//llvm:Support",
diff --git a/integrations/tensorflow/iree_tf_compiler/TFL/Passes.cpp b/integrations/tensorflow/iree_tf_compiler/TFL/Passes.cpp
index 03aa183..eaf7a7a 100644
--- a/integrations/tensorflow/iree_tf_compiler/TFL/Passes.cpp
+++ b/integrations/tensorflow/iree_tf_compiler/TFL/Passes.cpp
@@ -6,7 +6,6 @@
 
 #include "iree_tf_compiler/TFL/Passes.h"
 
-#include "iree/compiler/Dialect/Shape/Conversion/Passes.h"
 #include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
 #include "mlir/Dialect/Shape/Transforms/Passes.h"
 #include "mlir/Pass/PassManager.h"
diff --git a/iree/compiler/Bindings/Native/Transforms/BUILD b/iree/compiler/Bindings/Native/Transforms/BUILD
index b6f46e3..d2f45e4 100644
--- a/iree/compiler/Bindings/Native/Transforms/BUILD
+++ b/iree/compiler/Bindings/Native/Transforms/BUILD
@@ -23,7 +23,6 @@
         "//iree/compiler/Dialect/Flow/IR",
         "//iree/compiler/Dialect/HAL/IR",
         "//iree/compiler/Dialect/IREE/IR",
-        "//iree/compiler/Dialect/Shape/Conversion",
         "//iree/compiler/Dialect/Shape/IR",
         "//iree/compiler/Dialect/Shape/Transforms",
         "//iree/compiler/Dialect/Shape/Utils:TypeConversion",
diff --git a/iree/compiler/Bindings/Native/Transforms/CMakeLists.txt b/iree/compiler/Bindings/Native/Transforms/CMakeLists.txt
index 93c4447..f1086e1 100644
--- a/iree/compiler/Bindings/Native/Transforms/CMakeLists.txt
+++ b/iree/compiler/Bindings/Native/Transforms/CMakeLists.txt
@@ -33,7 +33,6 @@
     iree::compiler::Dialect::Flow::IR
     iree::compiler::Dialect::HAL::IR
     iree::compiler::Dialect::IREE::IR
-    iree::compiler::Dialect::Shape::Conversion
     iree::compiler::Dialect::Shape::IR
     iree::compiler::Dialect::Shape::Transforms
     iree::compiler::Dialect::Shape::Utils::TypeConversion
diff --git a/iree/compiler/Bindings/SIP/Transforms/BUILD b/iree/compiler/Bindings/SIP/Transforms/BUILD
index d64b591..6005b88 100644
--- a/iree/compiler/Bindings/SIP/Transforms/BUILD
+++ b/iree/compiler/Bindings/SIP/Transforms/BUILD
@@ -24,7 +24,6 @@
         "//iree/compiler/Dialect/Flow/IR",
         "//iree/compiler/Dialect/Flow/Transforms",
         "//iree/compiler/Dialect/IREE/IR",
-        "//iree/compiler/Dialect/Shape/Conversion",
         "//iree/compiler/Dialect/Shape/IR",
         "//iree/compiler/Dialect/Shape/Transforms",
         "//iree/compiler/Dialect/Shape/Utils:TypeConversion",
diff --git a/iree/compiler/Bindings/SIP/Transforms/CMakeLists.txt b/iree/compiler/Bindings/SIP/Transforms/CMakeLists.txt
index 4e10c94..c668a79 100644
--- a/iree/compiler/Bindings/SIP/Transforms/CMakeLists.txt
+++ b/iree/compiler/Bindings/SIP/Transforms/CMakeLists.txt
@@ -33,7 +33,6 @@
     iree::compiler::Dialect::Flow::IR
     iree::compiler::Dialect::Flow::Transforms
     iree::compiler::Dialect::IREE::IR
-    iree::compiler::Dialect::Shape::Conversion
     iree::compiler::Dialect::Shape::IR
     iree::compiler::Dialect::Shape::Transforms
     iree::compiler::Dialect::Shape::Utils::TypeConversion
diff --git a/iree/compiler/Bindings/TFLite/Transforms/BUILD b/iree/compiler/Bindings/TFLite/Transforms/BUILD
index e9e0c07..952494d 100644
--- a/iree/compiler/Bindings/TFLite/Transforms/BUILD
+++ b/iree/compiler/Bindings/TFLite/Transforms/BUILD
@@ -23,7 +23,6 @@
     deps = [
         "//iree/compiler/Dialect/Flow/IR",
         "//iree/compiler/Dialect/IREE/IR",
-        "//iree/compiler/Dialect/Shape/Conversion",
         "//iree/compiler/Dialect/Shape/IR",
         "//iree/compiler/Dialect/Shape/Transforms",
         "//iree/compiler/Dialect/Shape/Utils:TypeConversion",
diff --git a/iree/compiler/Bindings/TFLite/Transforms/CMakeLists.txt b/iree/compiler/Bindings/TFLite/Transforms/CMakeLists.txt
index d04063d..7e070a2 100644
--- a/iree/compiler/Bindings/TFLite/Transforms/CMakeLists.txt
+++ b/iree/compiler/Bindings/TFLite/Transforms/CMakeLists.txt
@@ -32,7 +32,6 @@
     MLIRTransforms
     iree::compiler::Dialect::Flow::IR
     iree::compiler::Dialect::IREE::IR
-    iree::compiler::Dialect::Shape::Conversion
     iree::compiler::Dialect::Shape::IR
     iree::compiler::Dialect::Shape::Transforms
     iree::compiler::Dialect::Shape::Utils::TypeConversion
diff --git a/iree/compiler/Conversion/HLOToLinalg/BUILD b/iree/compiler/Conversion/HLOToLinalg/BUILD
index 7359ed3..7a4a159 100644
--- a/iree/compiler/Conversion/HLOToLinalg/BUILD
+++ b/iree/compiler/Conversion/HLOToLinalg/BUILD
@@ -41,5 +41,6 @@
         "@mlir-hlo//:chlo_legalize_to_hlo",
         "@mlir-hlo//:hlo",
         "@mlir-hlo//:legalize_to_linalg",
+        "@mlir-hlo//:materialize_broadcasts",
     ],
 )
diff --git a/iree/compiler/Conversion/HLOToLinalg/HLOToLinalgOnTensors.cpp b/iree/compiler/Conversion/HLOToLinalg/HLOToLinalgOnTensors.cpp
index e629d04..d834cae 100644
--- a/iree/compiler/Conversion/HLOToLinalg/HLOToLinalgOnTensors.cpp
+++ b/iree/compiler/Conversion/HLOToLinalg/HLOToLinalgOnTensors.cpp
@@ -221,33 +221,21 @@
     MLIRContext *context = &getContext();
 
     auto typeConverter = mhlo::createHloToLinalgSignedIntegerConverter();
-    if (directHloClientLowering) {
-      // NOTE: not using corresponding setupHLOToFlowPatterns because the entire
-      // HLO dialects are marked illegal by this pass.
-      // TODO: Collapse/rework all of these patterns once the consolidation
-      // lands. There is little reason to have these so spread out.
-      populateHLOToFlowPatterns(context, patterns);
-      chlo::PopulateDecomposeChloPatterns(context, &patterns);
-      populateHLOBroadcastingToLinalgPatterns(context, *typeConverter,
-                                              patterns);
-      populateHLOToLinalgOnTensorsConversionPatterns(context, *typeConverter,
-                                                     patterns);
-      populateHLOComplexToRealPatterns(context, *typeConverter, patterns);
-    } else {
-      // Legacy: assumes that HLO client -> HLO has been run previously.
-      populateHLOToLinalgOnTensorsConversionPatterns(context, *typeConverter,
-                                                     patterns);
-    }
+    // NOTE: not using corresponding setupHLOToFlowPatterns because the entire
+    // HLO dialects are marked illegal by this pass.
+    // TODO: Collapse/rework all of these patterns once the consolidation
+    // lands. There is little reason to have these so spread out.
+    populateHLOToFlowPatterns(context, patterns);
+    chlo::PopulateDecomposeChloPatterns(context, &patterns);
+    populateHLOBroadcastingToLinalgPatterns(context, *typeConverter, patterns);
+    populateHLOToLinalgOnTensorsConversionPatterns(context, *typeConverter,
+                                                   patterns);
+    populateHLOComplexToRealPatterns(context, *typeConverter, patterns);
 
     ConversionTarget target(getContext());
     target.addIllegalDialect<chlo::HloClientDialect>();
     target.addIllegalDialect<mhlo::MhloDialect>();
 
-    if (directHloClientLowering) {
-      // Conversions to also legalize the HLO client dialect.
-      target.addIllegalDialect<chlo::HloClientDialect>();
-    }
-
     // Let the rest fall through.
     target.markUnknownOpDynamicallyLegal([](Operation *) { return true; });
 
diff --git a/iree/compiler/Conversion/HLOToLinalg/test/broadcasting.mlir b/iree/compiler/Conversion/HLOToLinalg/test/broadcasting.mlir
index 6ad5dd8..3893e8b 100644
--- a/iree/compiler/Conversion/HLOToLinalg/test/broadcasting.mlir
+++ b/iree/compiler/Conversion/HLOToLinalg/test/broadcasting.mlir
@@ -1,4 +1,4 @@
-// RUN: iree-opt -split-input-file -iree-codegen-hlo-to-linalg-on-tensors='direct-hlo-client-lowering' %s | IreeFileCheck %s
+// RUN: iree-opt -split-input-file -iree-codegen-hlo-to-linalg-on-tensors %s | IreeFileCheck %s
 
 // Check the non-broadcast case for each registered op, then just check a
 // representative op for detailed broadcast semantics. Since the broadcasting
diff --git a/iree/compiler/Conversion/HLOToLinalg/test/missing_legalizations.mlir b/iree/compiler/Conversion/HLOToLinalg/test/missing_legalizations.mlir
index c56a13d..8eff464 100644
--- a/iree/compiler/Conversion/HLOToLinalg/test/missing_legalizations.mlir
+++ b/iree/compiler/Conversion/HLOToLinalg/test/missing_legalizations.mlir
@@ -1,4 +1,4 @@
-// RUN: iree-opt -split-input-file -iree-codegen-hlo-to-linalg-on-tensors='direct-hlo-client-lowering' -verify-diagnostics %s
+// RUN: iree-opt -split-input-file -iree-codegen-hlo-to-linalg-on-tensors -verify-diagnostics %s
 
 // Left and right shift operators are missing HLO->Linalg lowerings.
 func @shift_leftWithoutBroadcast(%arg0: tensor<4xf32>, %arg1: tensor<4xf32>) -> tensor<4xf32> {
diff --git a/iree/compiler/Conversion/LinalgToLLVMGPU/ConvertToLLVM.cpp b/iree/compiler/Conversion/LinalgToLLVMGPU/ConvertToLLVM.cpp
index f1900ee..0223753 100644
--- a/iree/compiler/Conversion/LinalgToLLVMGPU/ConvertToLLVM.cpp
+++ b/iree/compiler/Conversion/LinalgToLLVMGPU/ConvertToLLVM.cpp
@@ -71,6 +71,34 @@
   }
 };
 
+// Convention with the HAL side to pass kernel arguments.
+// The bindings are ordered based on binding index then compressed and mapped to
+// dense set of arguments.
+// This function looks at the symbols and return the mapping between binding
+// index and kernel argument index. For instance if the kernel has bindings 1,
+// 5, 6 it will return the mapping [1, 0], [5, 1], [6, 2]
+static llvm::SmallDenseMap<uint64_t, size_t> getKernelArgMapping(
+    Operation *func) {
+  llvm::SmallDenseMap<uint64_t, size_t> mapBindingArgIndex;
+  llvm::SmallVector<uint64_t> bindingUsed;
+  Operation *symbolTableOp = SymbolTable::getNearestSymbolTable(func);
+  SymbolTable::walkSymbolTables(symbolTableOp, true, [&](Operation *op, bool) {
+    if (auto interface = dyn_cast<IREE::HAL::InterfaceOp>(op)) {
+      interface.walk([&](Operation *symbolOp) {
+        if (auto binding = dyn_cast<IREE::HAL::InterfaceBindingOp>(symbolOp)) {
+          uint64_t bindingIndex = binding.binding().getZExtValue();
+          bindingUsed.push_back(bindingIndex);
+        }
+      });
+    }
+  });
+  std::sort(bindingUsed.begin(), bindingUsed.end());
+  for (auto binding : llvm::enumerate(bindingUsed)) {
+    mapBindingArgIndex[binding.value()] = binding.index();
+  }
+  return mapBindingArgIndex;
+}
+
 class ConvertFunc : public ConvertToLLVMPattern {
  public:
   explicit ConvertFunc(MLIRContext *context, LLVMTypeConverter &converter)
@@ -88,13 +116,16 @@
     assert(fnType.getNumInputs() == 0 && fnType.getNumResults() == 0);
 
     TypeConverter::SignatureConversion signatureConverter(/*numOrigInputs=*/0);
-    SmallVector<Type, 8> llvmInputTypes;
+    llvm::SmallDenseMap<uint64_t, size_t> argMapping =
+        getKernelArgMapping(funcOp);
+    SmallVector<Type, 8> llvmInputTypes(argMapping.size());
     funcOp.walk([&](IREE::HAL::InterfaceBindingSubspanOp input) {
       auto memrefType = input.getType().cast<MemRefType>();
       Type elType = memrefType.getElementType();
       auto llvmType =
           LLVM::LLVMPointerType::get(elType, memrefType.getMemorySpaceAsInt());
-      llvmInputTypes.push_back(llvmType);
+      uint64_t binding = input.queryBindingOp().binding().getZExtValue();
+      llvmInputTypes[argMapping[binding]] = llvmType;
     });
     signatureConverter.addInputs(llvmInputTypes);
 
@@ -142,18 +173,15 @@
     if (!llvmFuncOp) return failure();
     assert(llvmFuncOp.getNumArguments() > 0);
 
+    llvm::SmallDenseMap<uint64_t, size_t> argMapping =
+        getKernelArgMapping(llvmFuncOp);
     Location loc = op->getLoc();
     auto ireeBindingOp = cast<IREE::HAL::InterfaceBindingSubspanOp>(op);
     IREE::HAL::InterfaceBindingSubspanOpAdaptor adaptor(operands);
     MemRefType memrefType =
         ireeBindingOp.getResult().getType().dyn_cast<MemRefType>();
-
-    // Fetch the interface binding op and extract the buffer index from void**.
-    auto symbol = SymbolTable::lookupNearestSymbolFrom(
-        op, op->getAttrOfType<SymbolRefAttr>("binding"));
-    auto interfaceBindingOp = cast<IREE::HAL::InterfaceBindingOp>(symbol);
-    Value llvmBufferBasePtr =
-        llvmFuncOp.getArgument(interfaceBindingOp.binding().getZExtValue());
+    uint64_t binding = ireeBindingOp.queryBindingOp().binding().getZExtValue();
+    Value llvmBufferBasePtr = llvmFuncOp.getArgument(argMapping[binding]);
     if (memrefType.hasStaticShape()) {
       auto desc = MemRefDescriptor::fromStaticShape(
           rewriter, loc, *getTypeConverter(), memrefType, llvmBufferBasePtr);
diff --git a/iree/compiler/Conversion/LinalgToLLVMGPU/test/convert_to_nvvm.mlir b/iree/compiler/Conversion/LinalgToLLVMGPU/test/convert_to_nvvm.mlir
index 9484bb4..e9c87a4 100644
--- a/iree/compiler/Conversion/LinalgToLLVMGPU/test/convert_to_nvvm.mlir
+++ b/iree/compiler/Conversion/LinalgToLLVMGPU/test/convert_to_nvvm.mlir
@@ -4,26 +4,27 @@
 func @abs_ex_dispatch_0() {
   %c0 = constant 0 : index
   %0 = hal.interface.binding.subspan @io::@arg0[%c0] : memref<16xf32>
-  %1 = hal.interface.binding.subspan @io::@arg1[%c0] : memref<16xf32>
+  %1 = hal.interface.binding.subspan @io::@arg1[%c0] : memref<16xi32>
   %2 = hal.interface.binding.subspan @io::@ret0[%c0] : memref<16xf32>
   %3 = "gpu.block_id"() {dimension = "x"} : () -> index
   %4 = "gpu.block_dim"() {dimension = "x"} : () -> index
   %5 = "gpu.thread_id"() {dimension = "x"} : () -> index
   %6 = muli %3, %4 : index
   %7 = addi %6, %5 : index
-  %9 = memref.load %1[%7] : memref<16xf32>
-  %10 = memref.load %2[%7] : memref<16xf32>
-  %11 = addf %9, %10 : f32
-  memref.store %11, %0[%7] : memref<16xf32>
+  %9 = memref.load %0[%7] : memref<16xf32>
+  %10 = memref.load %1[%7] : memref<16xi32>
+  %11 = sitofp %10 : i32 to f32
+  %12 = addf %9, %11 : f32
+  memref.store %12, %2[%7] : memref<16xf32>
   return
 }
 hal.interface @io attributes {sym_visibility = "private"} {
-  hal.interface.binding @arg0, set=0, binding=0, type="StorageBuffer", access="Read"
-  hal.interface.binding @arg1, set=0, binding=1, type="StorageBuffer", access="Read"
-  hal.interface.binding @ret0, set=0, binding=2, type="StorageBuffer", access="Write|Discard"
+  hal.interface.binding @arg0, set=0, binding=4, type="StorageBuffer", access="Read"
+  hal.interface.binding @arg1, set=0, binding=0, type="StorageBuffer", access="Read"
+  hal.interface.binding @ret0, set=0, binding=7, type="StorageBuffer", access="Write|Discard"
 }
 
 // CHECK-LABEL: llvm.func @abs_ex_dispatch_0
-//  CHECK-SAME: (%{{.*}}: !llvm.ptr<f32>, %{{.*}}: !llvm.ptr<f32>, %{{.*}}: !llvm.ptr<f32>)
+//  CHECK-SAME: (%{{.*}}: !llvm.ptr<i32>, %{{.*}}: !llvm.ptr<f32>, %{{.*}}: !llvm.ptr<f32>)
 //      CHECK:    nvvm.read.ptx.sreg.tid.x
 //      CHECK:    llvm.fadd
diff --git a/iree/compiler/Conversion/Passes.td b/iree/compiler/Conversion/Passes.td
index 7483385..06dfc08 100644
--- a/iree/compiler/Conversion/Passes.td
+++ b/iree/compiler/Conversion/Passes.td
@@ -14,12 +14,6 @@
     Pass<"iree-codegen-hlo-to-linalg-on-tensors", "FuncOp"> {
   let summary = "Convert from XLA-HLO ops to Linalg ops on tensors";
   let constructor = "mlir::iree_compiler::createHLOToLinalgOnTensorsPass()";
-
-  let options = [
-    Option<"directHloClientLowering", "direct-hlo-client-lowering",
-        "bool", /*default=*/"true",
-        "Lower directly from the HLO client dialect.">,
-  ];
 }
 
 def DemoteF32ToF16 :
diff --git a/iree/compiler/Dialect/Flow/Transforms/BUILD b/iree/compiler/Dialect/Flow/Transforms/BUILD
index 3a6d023..3359517 100644
--- a/iree/compiler/Dialect/Flow/Transforms/BUILD
+++ b/iree/compiler/Dialect/Flow/Transforms/BUILD
@@ -46,7 +46,6 @@
         "OutlineLargeConstants.cpp",
         "PassDetail.h",
         "Passes.cpp",
-        "PrePostPartitioningConversion.cpp",
         "PromoteI1ToI8Pass.cpp",
         "PromoteTensorLoads.cpp",
         "StripAndSplatConstantVariables.cpp",
@@ -69,7 +68,6 @@
         "//iree/compiler/Dialect/Flow/IR",
         "//iree/compiler/Dialect/HAL/IR",
         "//iree/compiler/Dialect/IREE/IR",
-        "//iree/compiler/Dialect/Shape/Conversion",
         "//iree/compiler/Dialect/Shape/IR",
         "//iree/compiler/Dialect/Shape/Transforms",
         "//iree/compiler/Dialect/Shape/Utils:TypeConversion",
diff --git a/iree/compiler/Dialect/Flow/Transforms/CMakeLists.txt b/iree/compiler/Dialect/Flow/Transforms/CMakeLists.txt
index c576f73..729a12e 100644
--- a/iree/compiler/Dialect/Flow/Transforms/CMakeLists.txt
+++ b/iree/compiler/Dialect/Flow/Transforms/CMakeLists.txt
@@ -42,7 +42,6 @@
     "OutlineLargeConstants.cpp"
     "PassDetail.h"
     "Passes.cpp"
-    "PrePostPartitioningConversion.cpp"
     "PromoteI1ToI8Pass.cpp"
     "PromoteTensorLoads.cpp"
     "StripAndSplatConstantVariables.cpp"
@@ -77,7 +76,6 @@
     iree::compiler::Dialect::Flow::IR
     iree::compiler::Dialect::HAL::IR
     iree::compiler::Dialect::IREE::IR
-    iree::compiler::Dialect::Shape::Conversion
     iree::compiler::Dialect::Shape::IR
     iree::compiler::Dialect::Shape::Transforms
     iree::compiler::Dialect::Shape::Utils::TypeConversion
diff --git a/iree/compiler/Dialect/Flow/Transforms/Passes.cpp b/iree/compiler/Dialect/Flow/Transforms/Passes.cpp
index 27b0319..8130008 100644
--- a/iree/compiler/Dialect/Flow/Transforms/Passes.cpp
+++ b/iree/compiler/Dialect/Flow/Transforms/Passes.cpp
@@ -11,7 +11,6 @@
 #include "iree/compiler/Conversion/Common/Passes.h"
 #include "iree/compiler/Conversion/LinalgToLinalg/Passes.h"
 #include "iree/compiler/Conversion/Passes.h"
-#include "iree/compiler/Dialect/Shape/Conversion/Passes.h"
 #include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
 #include "mlir-hlo/Dialect/mhlo/transforms/passes.h"
 #include "mlir/Conversion/SCFToStandard/SCFToStandard.h"
@@ -122,6 +121,8 @@
   passManager.addPass(mlir::createInlinerPass());
 
   passManager.addNestedPass<FuncOp>(tosa::createTosaToStandard());
+  passManager.addNestedPass<FuncOp>(mlir::createCanonicalizerPass());
+  passManager.addNestedPass<FuncOp>(Flow::createPromoteI1ToI8Pass());
   passManager.addNestedPass<FuncOp>(tosa::createTosaToLinalgOnTensors());
   passManager.addNestedPass<FuncOp>(mlir::createCanonicalizerPass());
 }
diff --git a/iree/compiler/Dialect/Flow/Transforms/Passes.h b/iree/compiler/Dialect/Flow/Transforms/Passes.h
index 3be438d..06c2ecf 100644
--- a/iree/compiler/Dialect/Flow/Transforms/Passes.h
+++ b/iree/compiler/Dialect/Flow/Transforms/Passes.h
@@ -73,11 +73,6 @@
 /// backends.
 std::unique_ptr<OperationPass<FuncOp>> createHLOToHLOPreprocessingPass();
 
-// Runs pre-partitioning conversion passes to convert to the flow dialect.
-// This converts some input ops directly to flow ops when doing so has a
-// benefit. Other ops are left unmodified and will be outlined later on.
-std::unique_ptr<OperationPass<FuncOp>> createPrePartitioningConversionPass();
-
 // Converts standard ops which match to flow.tensor.load (typically causing a
 // read-back).
 // Note that there are typically very specific phase ordering issues with
diff --git a/iree/compiler/Dialect/Flow/Transforms/Passes.td b/iree/compiler/Dialect/Flow/Transforms/Passes.td
index 36d7a31..d9021aa 100644
--- a/iree/compiler/Dialect/Flow/Transforms/Passes.td
+++ b/iree/compiler/Dialect/Flow/Transforms/Passes.td
@@ -94,12 +94,6 @@
   let constructor = "mlir::iree_compiler::IREE::Flow::createPromoteTensorLoadsPass()";
 }
 
-def PrePartitioningConversion :
-    Pass<"iree-flow-pre-partitioning-conversion", "FuncOp"> {
-  let summary = "Dialect conversion prior to partitioning";
-  let constructor = "mlir::iree_compiler::IREE::Flow::createPrePartitioningConversionPass()";
-}
-
 def StripAndSplatConstantVariables :
     Pass<"iree-flow-strip-and-splat-constant-variables", "ModuleOp"> {
   let summary = "Strips constant flow.variables and replaces them with splats.";
diff --git a/iree/compiler/Dialect/Flow/Transforms/PrePostPartitioningConversion.cpp b/iree/compiler/Dialect/Flow/Transforms/PrePostPartitioningConversion.cpp
deleted file mode 100644
index 2b443cf..0000000
--- a/iree/compiler/Dialect/Flow/Transforms/PrePostPartitioningConversion.cpp
+++ /dev/null
@@ -1,97 +0,0 @@
-// Copyright 2019 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include <utility>
-
-#include "iree/compiler/Dialect/Flow/Conversion/HLOToFlow/ConvertHLOToFlow.h"
-#include "iree/compiler/Dialect/Flow/Conversion/StandardToFlow/ConvertStandardToFlow.h"
-#include "iree/compiler/Dialect/Flow/Conversion/TypeConverter.h"
-#include "iree/compiler/Dialect/Flow/IR/FlowDialect.h"
-#include "iree/compiler/Dialect/Flow/Transforms/PassDetail.h"
-#include "iree/compiler/Dialect/Flow/Transforms/Passes.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Patterns.h"
-#include "mlir-hlo/Dialect/mhlo/IR/hlo_ops.h"
-#include "mlir-hlo/Dialect/mhlo/transforms/rewriters.h"
-#include "mlir/Dialect/Linalg/IR/LinalgOps.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/Dialect/Tensor/IR/Tensor.h"
-#include "mlir/IR/Builders.h"
-#include "mlir/IR/BuiltinOps.h"
-#include "mlir/Pass/Pass.h"
-#include "mlir/Support/LLVM.h"
-#include "mlir/Transforms/DialectConversion.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace IREE {
-namespace Flow {
-
-class PrePartitioningConversionPass
-    : public PrePartitioningConversionBase<PrePartitioningConversionPass> {
- public:
-  void getDependentDialects(DialectRegistry &registry) const override {
-    registry.insert<linalg::LinalgDialect, FlowDialect, mhlo::MhloDialect,
-                    StandardOpsDialect, tensor::TensorDialect>();
-  }
-
-  void runOnOperation() override {
-    auto *context = &getContext();
-    ConversionTarget conversionTarget(*context);
-    OwningRewritePatternList conversionPatterns(&getContext());
-
-    conversionTarget.addLegalDialect<IREE::Flow::FlowDialect>();
-
-    // Standard ops always pass through as import code may have produced some
-    // and control flow should have been legalized from HLO to std.
-    // The flow dialect uses std.module and std.func for its structure and they
-    // must be allowed.
-    conversionTarget.addLegalDialect<StandardOpsDialect>();
-    conversionTarget.addLegalOp<FuncOp>();
-
-    // Allow XLA HLO ops - we explicitly mark the ones we don't want below.
-    conversionTarget.addLegalDialect<mhlo::MhloDialect>();
-
-    // Control flow must be converted to standard form via
-    // mhlo::createLegalizeControlFlowPass() prior to conversion.
-    conversionTarget.addIllegalOp<mhlo::IfOp, mhlo::CaseOp, mhlo::WhileOp>();
-
-    // We don't support broadcast_dimensions as part of ops, so materialize
-    // any such attributes to dedicated mhlo.broadcast_in_dim ops.
-    mhlo::SetupMaterializeBroadcastsLegality(context, &conversionTarget);
-    mhlo::PopulateMaterializeBroadcastsPatterns(context, &conversionPatterns);
-
-    Shape::populateShapeToStandardConversionPatterns(conversionPatterns,
-                                                     context);
-    Shape::setupShapeToStandardLegality(conversionTarget);
-
-    // Early conversion of ops that have matches we want to route through.
-    // For example, DynamicUpdateSlice should end up as a stream operation.
-    setupDirectHLOToFlowLegality(context, conversionTarget);
-    populateHLOToFlowPatterns(context, conversionPatterns);
-    setupStandardToFlowTensorLoadLegality(context, conversionTarget);
-
-    conversionTarget.addLegalOp<linalg::GenericOp, linalg::IndexedGenericOp>();
-    conversionTarget
-        .markOpRecursivelyLegal<linalg::GenericOp, linalg::IndexedGenericOp>();
-    populateStandardToFlowTensorLoadPatterns(context, conversionPatterns);
-
-    if (failed(applyPartialConversion(getOperation(), conversionTarget,
-                                      std::move(conversionPatterns)))) {
-      getOperation().emitError()
-          << "module is not in a compatible input format";
-      return signalPassFailure();
-    }
-  }
-};
-
-std::unique_ptr<OperationPass<FuncOp>> createPrePartitioningConversionPass() {
-  return std::make_unique<PrePartitioningConversionPass>();
-}
-
-}  // namespace Flow
-}  // namespace IREE
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Flow/Transforms/test/BUILD b/iree/compiler/Dialect/Flow/Transforms/test/BUILD
index 1144999..aa480b3 100644
--- a/iree/compiler/Dialect/Flow/Transforms/test/BUILD
+++ b/iree/compiler/Dialect/Flow/Transforms/test/BUILD
@@ -33,7 +33,6 @@
             "legalize_input_types.mlir",
             "outline_dispatch_regions.mlir",
             "outline_large_constants.mlir",
-            "pre_partitioning_conversion.mlir",
             "promote_i1_to_i8.mlir",
             "promote_tensor_loads.mlir",
             "strip_and_splat_constant_variables.mlir",
diff --git a/iree/compiler/Dialect/Flow/Transforms/test/CMakeLists.txt b/iree/compiler/Dialect/Flow/Transforms/test/CMakeLists.txt
index e7ec6fb..95af5a9 100644
--- a/iree/compiler/Dialect/Flow/Transforms/test/CMakeLists.txt
+++ b/iree/compiler/Dialect/Flow/Transforms/test/CMakeLists.txt
@@ -30,7 +30,6 @@
     "legalize_input_types.mlir"
     "outline_dispatch_regions.mlir"
     "outline_large_constants.mlir"
-    "pre_partitioning_conversion.mlir"
     "promote_i1_to_i8.mlir"
     "promote_tensor_loads.mlir"
     "strip_and_splat_constant_variables.mlir"
diff --git a/iree/compiler/Dialect/Flow/Transforms/test/pre_partitioning_conversion.mlir b/iree/compiler/Dialect/Flow/Transforms/test/pre_partitioning_conversion.mlir
deleted file mode 100644
index 2623563..0000000
--- a/iree/compiler/Dialect/Flow/Transforms/test/pre_partitioning_conversion.mlir
+++ /dev/null
@@ -1,43 +0,0 @@
-// RUN: iree-opt -split-input-file -iree-flow-pre-partitioning-conversion %s | IreeFileCheck %s
-
-#map = affine_map<(d0) -> (d0)>
-
-// CHECK-LABEL: @linalg_generic
-func @linalg_generic(%arg0 : tensor<1xi32>) {
-  %init = linalg.init_tensor [1] : tensor<1xi32>
-  %generic = linalg.generic {indexing_maps = [#map, #map], iterator_types = ["parallel"]}
-    ins(%arg0 : tensor<1xi32>) outs(%init : tensor<1xi32>) {
-    ^bb0(%arg1: i32, %s: i32):
-      %idx = index_cast %arg1 : i32 to index
-      // CHECK: tensor.extract
-      %extract = tensor.extract %arg0[%idx] : tensor<1xi32>
-      linalg.yield %extract : i32
-    } -> tensor<1xi32>
-  return
-}
-
-// -----
-
-#map = affine_map<(d0) -> (d0)>
-
-// CHECK-LABEL: @linalg_indexed_generic
-func @linalg_indexed_generic(%arg0 : tensor<1xi32>) {
-  %init = linalg.init_tensor [1] : tensor<1xi32>
-  %generic = linalg.indexed_generic {indexing_maps = [#map, #map], iterator_types = ["parallel"]}
-    ins(%arg0 : tensor<1xi32>) outs(%init : tensor<1xi32>) {
-    ^bb0(%arg1: index, %arg2: i32, %s: i32):
-      %idx = index_cast %arg2 : i32 to index
-      // CHECK: tensor.extract
-      %extract = tensor.extract %arg0[%idx] : tensor<1xi32>
-      linalg.yield %extract : i32
-    } -> tensor<1xi32>
-  return
-}
-
-// -----
-
-func @tensor_extract(%arg0 : tensor<1xi32>, %arg1 : index) {
-  // CHECK: flow.tensor.load
-  %extract = tensor.extract %arg0[%arg1] : tensor<1xi32>
-  return
-}
diff --git a/iree/compiler/Dialect/Shape/Conversion/BUILD b/iree/compiler/Dialect/Shape/Conversion/BUILD
deleted file mode 100644
index 8e72535..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/BUILD
+++ /dev/null
@@ -1,29 +0,0 @@
-package(
-    default_visibility = ["//visibility:public"],
-    features = ["layering_check"],
-    licenses = ["notice"],  # Apache 2.0
-)
-
-cc_library(
-    name = "ConvertShapeToShapex",
-    srcs = ["ConvertShapeToShapex.cpp"],
-    deps = [
-        "//iree/compiler/Dialect/Shape/IR",
-        "@llvm-project//mlir:Dialect",
-        "@llvm-project//mlir:IR",
-        "@llvm-project//mlir:Pass",
-        "@llvm-project//mlir:Shape",
-        "@llvm-project//mlir:StandardOps",
-        "@llvm-project//mlir:TensorDialect",
-        "@llvm-project//mlir:Transforms",
-    ],
-)
-
-cc_library(
-    name = "Conversion",
-    hdrs = ["Passes.h"],
-    deps = [
-        ":ConvertShapeToShapex",
-        "@llvm-project//mlir:Pass",
-    ],
-)
diff --git a/iree/compiler/Dialect/Shape/Conversion/CMakeLists.txt b/iree/compiler/Dialect/Shape/Conversion/CMakeLists.txt
deleted file mode 100644
index 6b299e3..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/CMakeLists.txt
+++ /dev/null
@@ -1,41 +0,0 @@
-################################################################################
-# Autogenerated by build_tools/bazel_to_cmake/bazel_to_cmake.py from           #
-# iree/compiler/Dialect/Shape/Conversion/BUILD                                 #
-#                                                                              #
-# Use iree_cmake_extra_content from iree/build_defs.oss.bzl to add arbitrary   #
-# CMake-only content.                                                          #
-#                                                                              #
-# To disable autogeneration for this file entirely, delete this header.        #
-################################################################################
-
-iree_add_all_subdirs()
-
-iree_cc_library(
-  NAME
-    ConvertShapeToShapex
-  SRCS
-    "ConvertShapeToShapex.cpp"
-  DEPS
-    MLIRDialect
-    MLIRIR
-    MLIRPass
-    MLIRShape
-    MLIRStandard
-    MLIRTensor
-    MLIRTransforms
-    iree::compiler::Dialect::Shape::IR
-  PUBLIC
-)
-
-iree_cc_library(
-  NAME
-    Conversion
-  HDRS
-    "Passes.h"
-  DEPS
-    ::ConvertShapeToShapex
-    MLIRPass
-  PUBLIC
-)
-
-### BAZEL_TO_CMAKE_PRESERVES_ALL_CONTENT_BELOW_THIS_LINE ###
diff --git a/iree/compiler/Dialect/Shape/Conversion/ConvertShapeToShapex.cpp b/iree/compiler/Dialect/Shape/Conversion/ConvertShapeToShapex.cpp
deleted file mode 100644
index 227562c..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/ConvertShapeToShapex.cpp
+++ /dev/null
@@ -1,356 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include "iree/compiler/Dialect/Shape/IR/ShapeDialect.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeTypes.h"
-#include "mlir/Dialect/Shape/IR/Shape.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/Dialect/Tensor/IR/Tensor.h"
-#include "mlir/Dialect/Traits.h"
-#include "mlir/IR/BuiltinOps.h"
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/Matchers.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Pass/Pass.h"
-#include "mlir/Pass/PassRegistry.h"
-#include "mlir/Transforms/DialectConversion.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-namespace {
-
-// This conversion is currently quite limited, such as not handling multiple
-// basic blocks in general, due to doing a type conversion that the MLIR core
-// conversion infra doesn't handle well.
-//
-// In particular, we convert `!shape.shape` to `!shapex.ranked_shape<...>`, but
-// the contents of the `...` are context-dependent. Thus, one could say that
-// this pass does a context-dependent type conversion.
-//
-// The current MLIR conversion infra doesn't handle context-dependent type
-// conversions.
-//
-// I can see two solutions:
-//
-// 1. Extend the MLIR conversion infra to better support context-dependent type
-// conversions. One way to do this would be for the conversion infra to convert
-// blocks in RPO and use the type of the converted successor operand in a
-// dominating predecessor as the type for the block argument when converting a
-// block. A similar thing could be done with an RPO traversal of the callgraph.
-// This algorithm wouldn't work in the presence of recursively dead cycles. And
-// of course linkage boundaries cannot have a context-dependent type conversion
-// (by definition).
-//
-// 2. Avoid needing to convert to !shapex.ranked_shape in the first place. This
-// could be accomplished by generalizing !shape.shape to be able to support the
-// use case of !shapex.ranked_shape. One important requirement here is that
-// !shapex.ranked_shape models a partially-specified shape (hardcoded for the
-// ranked case). !shape.shape could be extended to capture partially-specified
-// shapes in the type, such as allowing `!shape.shape<*>` to model an unranked
-// shape (which is the default; no information), `!shape.shape<?x?x5x?>` to
-// model a rank-4 shape with dimension 2 being of extent 5, etc.
-//
-// Once we have this, we could do this lowering from generic !shape.shape to
-// statically-known ranked shapes more progressively and treat it more like a
-// type refinement algorithm.
-//
-// The main risk is that we are trying to shove too much stuff into the
-// !shape.shape type. There's a risk that "progressive lowering" becomes "no
-// clear boundaries" and we end up with code deep into the compiler continuously
-// needing to doublecheck that the !shape.shape's at this point are in fact
-// statically known to be ranked, or silently making that assumption and
-// triggering assertions on verifier-valid IR. Pipelines and legalization
-// targets could make these assertions not fire in practice, but it would
-// be a maintenance burden.
-
-class ConvertConstShapeOp : public OpConversionPattern<shape::ConstShapeOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::ConstShapeOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    SmallVector<int64_t, 4> extents;
-    for (APInt extent : op.shape()) {
-      extents.push_back(extent.getZExtValue());
-    }
-    auto rsType = RankedShapeType::get(extents, rewriter.getContext());
-    rewriter.replaceOpWithNewOp<ConstRankedShapeOp>(op, rsType);
-    return success();
-  }
-};
-
-class ConvertShapeOfOp : public OpConversionPattern<shape::ShapeOfOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::ShapeOfOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    auto tensorType = operands[0].getType().dyn_cast<RankedTensorType>();
-    if (!tensorType) {
-      return failure();
-    }
-    auto resultType =
-        RankedShapeType::get(tensorType.getShape(), rewriter.getContext());
-    // TODO(jpienaar): The following needs to be re-evaluated once the patch
-    // train from 2020/07/23 integrates properly. This is required to make
-    // it forward and backwards compatible. Also, tests need to be added once
-    // upstream integrates (and this can be tested).
-    // rewriter.replaceOpWithNewOp<Shape::GetRankedShapeOp>(op, resultType,
-    //                                                      operands[0]);
-    auto getRanked = rewriter.create<Shape::GetRankedShapeOp>(
-        op.getLoc(), resultType, operands[0]);
-
-    // For FromExtentTensorOp users, just forward the result from GetRanked.
-    SmallPtrSet<Operation *, 2> toDelete;
-    for (auto use : op.getOperation()->getUsers()) {
-      if (isa<FromExtentTensorOp>(use)) {
-        use->replaceAllUsesWith(getRanked);
-        toDelete.insert(use);
-      }
-    }
-    for (Operation *use : toDelete) {
-      rewriter.eraseOp(use);
-    }
-
-    rewriter.replaceOp(op.getOperation(), getRanked.getResult());
-    return success();
-  }
-};
-
-class ConvertTensorExtract : public OpConversionPattern<tensor::ExtractOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      tensor::ExtractOp op, ArrayRef<Value> rawOperands,
-      ConversionPatternRewriter &rewriter) const override {
-    tensor::ExtractOpAdaptor operands(rawOperands);
-    if (!operands.tensor().getType().isa<RankedShapeType>()) {
-      return rewriter.notifyMatchFailure(op, "not acting on a ranked shape");
-    }
-    auto dim = operands.indices().front();
-    auto dimConstOp = dyn_cast_or_null<ConstantIndexOp>(dim.getDefiningOp());
-    if (!dimConstOp) {
-      return rewriter.notifyMatchFailure(op, "extract index not constant");
-    }
-    rewriter.replaceOpWithNewOp<Shape::RankedDimOp>(
-        op, rewriter.getIndexType(), operands.tensor(),
-        dimConstOp.value().cast<IntegerAttr>());
-    return success();
-  }
-};
-
-class ConvertGetExtent : public OpConversionPattern<shape::GetExtentOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::GetExtentOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    rewriter.replaceOpWithNewOp<Shape::RankedDimOp>(
-        op, rewriter.getIndexType(), operands[0],
-        op.getConstantDim().getValue());
-    return success();
-  }
-};
-
-class ConvertFromExtents : public OpConversionPattern<shape::FromExtentsOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::FromExtentsOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    SmallVector<Value, 4> dynOperands;
-    SmallVector<int64_t, 4> extents;
-    for (auto operand : operands) {
-      IntegerAttr indexAttr;
-      if (matchPattern(operand, m_Constant(&indexAttr))) {
-        extents.push_back(indexAttr.getValue().getSExtValue());
-        continue;
-      }
-      dynOperands.push_back(operand);
-      extents.push_back(-1);
-    }
-
-    auto resultType = RankedShapeType::get(extents, rewriter.getContext());
-    auto make = rewriter.create<Shape::MakeRankedShapeOp>(
-        op.getLoc(), resultType, dynOperands);
-
-    rewriter.replaceOp(op, make.getResult());
-
-    return success();
-  }
-};
-
-class ConvertSplitAtOp : public OpConversionPattern<shape::SplitAtOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::SplitAtOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    IntegerAttr indexAttr;
-    if (!matchPattern(op.index(), m_Constant(&indexAttr))) {
-      return rewriter.notifyMatchFailure(op, "requires constant `index`");
-    }
-    auto rank = operands[0].getType().cast<RankedShapeType>().getRank();
-    int64_t index = indexAttr.getInt();
-    if (index < 0) {
-      index += rank;
-    }
-    auto head_indices = llvm::to_vector<4>(llvm::seq<int64_t>(0, index));
-    auto tail_indices = llvm::to_vector<4>(llvm::seq<int64_t>(index, rank));
-    Value head = rewriter.create<GatherExtentsOp>(
-        op.getLoc(), operands[0], rewriter.getI64TensorAttr(head_indices));
-    Value tail = rewriter.create<GatherExtentsOp>(
-        op.getLoc(), operands[0], rewriter.getI64TensorAttr(tail_indices));
-    rewriter.replaceOp(op, {head, tail});
-    return success();
-  }
-};
-
-class ConvertBroadcastOp : public OpConversionPattern<shape::BroadcastOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::BroadcastOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    Value lhs = operands[0];
-    Value rhs = operands[1];
-    auto lhsType = lhs.getType().dyn_cast<RankedShapeType>();
-    auto rhsType = rhs.getType().dyn_cast<RankedShapeType>();
-    if (!lhsType || !rhsType) {
-      return failure();
-    }
-    // Establish invariant that rank(lhs) <= rank(rhs)
-    if (lhsType.getRank() > rhsType.getRank()) {
-      std::swap(lhsType, rhsType);
-      std::swap(lhs, rhs);
-    }
-    SmallVector<int64_t, 6> resultShape;
-    OpTrait::util::getBroadcastedShape(lhsType.getAllDims(),
-                                       rhsType.getAllDims(), resultShape);
-    auto resultType = RankedShapeType::get(resultShape, rewriter.getContext());
-    auto iota = llvm::to_vector<4>(llvm::seq<int64_t>(0, rhsType.getRank()));
-    Value broadcasted = rewriter.replaceOpWithNewOp<RankedBroadcastShapeOp>(
-        op, resultType, lhs, rhs,
-        /*lhs_broadcast_dimensions=*/
-        rewriter.getI64TensorAttr(makeArrayRef(iota).drop_front(
-            rhsType.getRank() - lhsType.getRank())),
-        /*rhs_broadcast_dimensions=*/
-        rewriter.getI64TensorAttr(iota));
-
-    // For FromExtentTensorOp users, just forward the RankedShapeType result.
-    for (Operation *user : op.getResult().getUsers()) {
-      if (isa<Shape::FromExtentTensorOp>(user)) {
-        rewriter.replaceOp(user, ValueRange{broadcasted});
-      }
-    }
-    return success();
-  }
-};
-
-class ConvertConcatOp : public OpConversionPattern<shape::ConcatOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::ConcatOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    auto resultRank = operands[0].getType().cast<RankedShapeType>().getRank() +
-                      operands[1].getType().cast<RankedShapeType>().getRank();
-    auto indices = llvm::to_vector<4>(llvm::seq<int64_t>(0, resultRank));
-    rewriter.replaceOpWithNewOp<Shape::GatherExtentsOp>(
-        op, ValueRange({operands[0], operands[1]}),
-        rewriter.getI64TensorAttr(indices));
-    return success();
-  }
-};
-
-class ConvertToExtentTensorOp
-    : public OpConversionPattern<shape::ToExtentTensorOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::ToExtentTensorOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    rewriter.replaceOpWithNewOp<Shape::ToExtentTensorOp>(op, op.getType(),
-                                                         operands[0]);
-    return success();
-  }
-};
-
-class ConvertFromExtentTensorOp
-    : public OpConversionPattern<shape::FromExtentTensorOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      shape::FromExtentTensorOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    if (operands.front().getType().isa<RankedTensorType>()) {
-      rewriter.replaceOpWithNewOp<Shape::FromExtentTensorOp>(op,
-                                                             operands.front());
-      return success();
-    }
-    if (operands.front().getType().isa<RankedShapeType>()) {
-      rewriter.replaceOp(op, operands.front());
-      return success();
-    }
-    return failure();
-  }
-};
-
-// Currently, upstream shape lowering can use tensor<?xindex> to represent a
-// shape, and will insert tensor_cast ops to convert to specific extent tensor
-// types. However, not all tensor_cast ops are shape-related.
-class ConvertTensorCastOp : public OpConversionPattern<tensor::CastOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      tensor::CastOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    if (!operands[0].getType().isa<RankedShapeType>())
-      return rewriter.notifyMatchFailure(op, "not a shape-related tensor_cast");
-    rewriter.replaceOpWithNewOp<Shape::ToExtentTensorOp>(op, op.getType(),
-                                                         operands[0]);
-    return success();
-  }
-};
-
-class ConvertShapeToShapex
-    : public PassWrapper<ConvertShapeToShapex, OperationPass<ModuleOp>> {
-  void getDependentDialects(DialectRegistry &registry) const override {
-    registry.insert<iree_compiler::ShapeDialect>();
-  }
-
-  void runOnOperation() override {
-    ModuleOp module = getOperation();
-    MLIRContext *context = &getContext();
-
-    // Conversion target definition.
-    ConversionTarget conversionTarget(*context);
-    conversionTarget.addIllegalDialect<shape::ShapeDialect>();
-    conversionTarget.addLegalDialect<iree_compiler::ShapeDialect>();
-
-    // Patterns.
-    OwningRewritePatternList patterns(&getContext());
-    patterns.insert<ConvertConstShapeOp>(context);
-    patterns.insert<ConvertShapeOfOp>(context);
-    patterns.insert<ConvertTensorExtract>(context);
-    patterns.insert<ConvertGetExtent>(context);
-    patterns.insert<ConvertFromExtents>(context);
-    patterns.insert<ConvertFromExtentTensorOp>(context);
-    patterns.insert<ConvertSplitAtOp>(context);
-    patterns.insert<ConvertBroadcastOp>(context);
-    patterns.insert<ConvertConcatOp>(context);
-    patterns.insert<ConvertToExtentTensorOp>(context);
-    patterns.insert<ConvertTensorCastOp>(context);
-
-    if (failed(applyPartialConversion(module, conversionTarget,
-                                      std::move(patterns)))) {
-      return signalPassFailure();
-    }
-  }
-};
-}  // namespace
-
-std::unique_ptr<OperationPass<ModuleOp>> createConvertShapeToShapexPass() {
-  return std::make_unique<ConvertShapeToShapex>();
-}
-
-static PassRegistration<ConvertShapeToShapex> registration(
-    "convert-shape-to-shapex", "Convert `shape` dialect to `shapex` dialect");
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Conversion/Passes.h b/iree/compiler/Dialect/Shape/Conversion/Passes.h
deleted file mode 100644
index 1edf727..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/Passes.h
+++ /dev/null
@@ -1,27 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#ifndef IREE_COMPILER_DIALECT_SHAPE_CONVERSION_PASSES_H_
-#define IREE_COMPILER_DIALECT_SHAPE_CONVERSION_PASSES_H_
-
-#include "mlir/Pass/Pass.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-
-// Convert `shape` dialect to `shapex` dialect.
-std::unique_ptr<OperationPass<ModuleOp>> createConvertShapeToShapexPass();
-
-inline void registerShapeConversionPasses() {
-  createConvertShapeToShapexPass();
-}
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
-
-#endif  // IREE_COMPILER_DIALECT_SHAPE_CONVERSION_PASSES_H_
diff --git a/iree/compiler/Dialect/Shape/Conversion/test/BUILD b/iree/compiler/Dialect/Shape/Conversion/test/BUILD
deleted file mode 100644
index 325d263..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/test/BUILD
+++ /dev/null
@@ -1,26 +0,0 @@
-# Copyright 2020 The IREE Authors
-#
-# Licensed under the Apache License v2.0 with LLVM Exceptions.
-# See https://llvm.org/LICENSE.txt for license information.
-# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-load("//iree:lit_test.bzl", "iree_lit_test_suite")
-load("//build_tools/bazel:enforce_glob.bzl", "enforce_glob")
-
-package(
-    default_visibility = ["//visibility:public"],
-    features = ["layering_check"],
-    licenses = ["notice"],  # Apache 2.0
-)
-
-iree_lit_test_suite(
-    name = "lit",
-    srcs = enforce_glob(
-        ["shape_to_shapex.mlir"],
-        include = ["*.mlir"],
-    ),
-    data = [
-        "//iree/tools:IreeFileCheck",
-        "//iree/tools:iree-opt",
-    ],
-)
diff --git a/iree/compiler/Dialect/Shape/Conversion/test/CMakeLists.txt b/iree/compiler/Dialect/Shape/Conversion/test/CMakeLists.txt
deleted file mode 100644
index d22775f..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/test/CMakeLists.txt
+++ /dev/null
@@ -1,23 +0,0 @@
-################################################################################
-# Autogenerated by build_tools/bazel_to_cmake/bazel_to_cmake.py from           #
-# iree/compiler/Dialect/Shape/Conversion/test/BUILD                            #
-#                                                                              #
-# Use iree_cmake_extra_content from iree/build_defs.oss.bzl to add arbitrary   #
-# CMake-only content.                                                          #
-#                                                                              #
-# To disable autogeneration for this file entirely, delete this header.        #
-################################################################################
-
-iree_add_all_subdirs()
-
-iree_lit_test_suite(
-  NAME
-    lit
-  SRCS
-    "shape_to_shapex.mlir"
-  DATA
-    iree::tools::IreeFileCheck
-    iree::tools::iree-opt
-)
-
-### BAZEL_TO_CMAKE_PRESERVES_ALL_CONTENT_BELOW_THIS_LINE ###
diff --git a/iree/compiler/Dialect/Shape/Conversion/test/shape_to_shapex.mlir b/iree/compiler/Dialect/Shape/Conversion/test/shape_to_shapex.mlir
deleted file mode 100644
index e6dcd01..0000000
--- a/iree/compiler/Dialect/Shape/Conversion/test/shape_to_shapex.mlir
+++ /dev/null
@@ -1,176 +0,0 @@
-// RUN: iree-opt -convert-shape-to-shapex -split-input-file -verify-diagnostics -allow-unregistered-dialect <%s | IreeFileCheck %s
-
-// -----
-// shape.const_shape
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>) {
-  // CHECK: shapex.const_ranked_shape : !shapex.ranked_shape<[1,2,3]>
-  %0 = shape.const_shape [1, 2, 3] : tensor<?xindex>
-  "foo.use"(%0) : (tensor<?xindex>) -> ()
-  return
-}
-
-// -----
-// shape.shape_of
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>) {
-  // CHECK: shapex.get_ranked_shape %arg0 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-  "foo.use"(%0) : (!shape.shape) -> ()
-  return
-}
-
-// -----
-// shape.split_at
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>) {
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-  %index = constant 0 : index
-  // CHECK: %[[RS:.+]] = shapex.get_ranked_shape %arg0
-  // CHECK: %[[HEAD:.+]] = "shapex.gather_extents"(%[[RS]]) {indices = dense<> : tensor<0xi64>} : (!shapex.ranked_shape<[?]>) -> !shapex.ranked_shape<[]>
-  // CHECK: %[[TAIL:.+]] = "shapex.gather_extents"(%[[RS]]) {indices = dense<0> : tensor<1xi64>} : (!shapex.ranked_shape<[?]>) -> !shapex.ranked_shape<[?]>
-  // CHECK: "foo.use"(%[[HEAD]], %[[TAIL]])
-  %head, %tail = "shape.split_at"(%0, %index) : (!shape.shape, index) -> (!shape.shape, !shape.shape)
-  "foo.use"(%head, %tail) : (!shape.shape, !shape.shape) -> ()
-  return
-}
-
-// -----
-// No conversion -- index is dynamic.
-func @f(%arg0: tensor<?xf32>, %index: index) {
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-  // expected-error @+1 {{failed to legalize operation}}
-  %head, %tail = "shape.split_at"(%0, %index) : (!shape.shape, index) -> (!shape.shape, !shape.shape)
-  "foo.use"(%head, %tail) : (!shape.shape, !shape.shape) -> ()
-  return
-}
-
-// -----
-// shape.get_extent
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>) {
-  %c0 = constant 0 : index
-  // CHECK: %[[SHAPE:.+]] = shapex.get_ranked_shape %arg0 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-
-  // CHECK: [[DIM:%.+]] = shapex.ranked_dim %[[SHAPE]][0] : !shapex.ranked_shape<[?]> -> index
-  %result = shape.get_extent %0, %c0 : !shape.shape, index -> !shape.size
-  return
-}
-
-// -----
-// shape.from_extents
-// CHECK-LABEL: func @f
-func @f(%arg0: index) {
-  // CHECK: shapex.make_ranked_shape %arg0, %arg0
-  %result = "shape.from_extents"(%arg0, %arg0) : (index, index) -> !shape.shape
-  return
-}
-
-// -----
-// shape.from_extent_tensor
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<3xindex>) {
-  // CHECK: "shapex.from_extent_tensor"(%arg0) : (tensor<3xindex>) -> !shapex.ranked_shape<[?,?,?]>
-  %result = "shape.from_extent_tensor"(%arg0) : (tensor<3xindex>)
-      -> !shape.shape
-  return
-}
-
-// -----
-// shape.broadcast
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: tensor<?xf32>) {
-  // CHECK: %[[LHSRS:.+]] = shapex.get_ranked_shape %arg0 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  // CHECK: %[[RHSRS:.+]] = shapex.get_ranked_shape %arg1 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-  %1 = "shape.shape_of"(%arg1) : (tensor<?xf32>) -> !shape.shape
-  // CHECK: %[[BROADCASTED:.+]] = "shapex.ranked_broadcast_shape"(%[[LHSRS]], %[[RHSRS]]) {
-  // CHECK-SAME: lhs_broadcast_dimensions = dense<0> : tensor<1xi64>,
-  // CHECK-SAME: rhs_broadcast_dimensions = dense<0> : tensor<1xi64>}
-  // CHECK-SAME: : (!shapex.ranked_shape<[?]>, !shapex.ranked_shape<[?]>) -> !shapex.ranked_shape<[?]>
-  %2 = "shape.broadcast"(%0, %1) : (!shape.shape, !shape.shape) -> !shape.shape
-  // CHECK: "foo.use"(%[[BROADCASTED]])
-  "foo.use"(%2) : (!shape.shape) -> ()
-  return
-}
-
-// -----
-// shape.broadcast
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?x?xf32>, %arg1: tensor<?x?xf32>) {
-  // CHECK: %[[LHSRS:.+]] = shapex.get_ranked_shape %arg0
-  // CHECK: %[[RHSRS:.+]] = shapex.get_ranked_shape %arg1
-  // CHECK: %[[BROADCASTED:.+]] = "shapex.ranked_broadcast_shape"(%[[LHSRS]], %[[RHSRS]])
-  // CHECK-SAME: : (!shapex.ranked_shape<[?,?]>, !shapex.ranked_shape<[?,?]>) -> !shapex.ranked_shape<[?,?]>
-  // CHECK: "use"(%[[BROADCASTED]])
-  %0 = "shape.shape_of"(%arg0) : (tensor<?x?xf32>) -> tensor<2xindex>
-  %1 = "shape.shape_of"(%arg1) : (tensor<?x?xf32>) -> tensor<2xindex>
-  %2 = "shape.broadcast"(%0, %1) : (tensor<2xindex>, tensor<2xindex>) -> tensor<2xindex>
-  "use"(%2) : (tensor<2xindex>) -> ()
-  return
-}
-
-// -----
-// shape.concat
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: tensor<?xf32>) {
-  // CHECK: %[[LHSRS:.+]] = shapex.get_ranked_shape %arg0 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  // CHECK: %[[RHSRS:.+]] = shapex.get_ranked_shape %arg1 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-  %1 = "shape.shape_of"(%arg1) : (tensor<?xf32>) -> !shape.shape
-  // CHECK: %[[CONCATTED:.+]] = "shapex.gather_extents"(%[[LHSRS]], %[[RHSRS]]) {indices = dense<[0, 1]> : tensor<2xi64>}
-  %2 = "shape.concat"(%0, %1) : (!shape.shape, !shape.shape) -> !shape.shape
-  // CHECK: "foo.use"(%[[CONCATTED]])
-  "foo.use"(%2) : (!shape.shape) -> ()
-  return
-}
-
-// -----
-// shape.to_extent_tensor
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: tensor<?xf32>) {
-  // CHECK: %[[RS:.+]] = shapex.get_ranked_shape %arg0 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  %0 = "shape.shape_of"(%arg0) : (tensor<?xf32>) -> !shape.shape
-  // CHECK: %[[EXTENTS:.+]] = "shapex.to_extent_tensor"(%[[RS]])
-  %1 = "shape.to_extent_tensor"(%0) : (!shape.shape) -> tensor<1xindex>
-  // CHECK: "foo.use"(%[[EXTENTS]])
-  "foo.use"(%1) : (tensor<1xindex>) -> ()
-  return
-}
-
-// -----
-// tensor.cast
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: tensor<?xf32>) {
-  // CHECK: %[[LHSRS:.+]] = shapex.get_ranked_shape %arg0 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  // CHECK: %[[RHSRS:.+]] = shapex.get_ranked_shape %arg1 : tensor<?xf32> -> !shapex.ranked_shape<[?]>
-  %0 = shape.shape_of %arg0 : tensor<?xf32> -> tensor<?xindex>
-  %1 = shape.shape_of %arg1 : tensor<?xf32> -> tensor<?xindex>
-  // CHECK: %[[BROADCASTED:.+]] = "shapex.ranked_broadcast_shape"(%[[LHSRS]], %[[RHSRS]]) {
-  // CHECK-SAME: lhs_broadcast_dimensions = dense<0> : tensor<1xi64>,
-  // CHECK-SAME: rhs_broadcast_dimensions = dense<0> : tensor<1xi64>}
-  // CHECK-SAME: : (!shapex.ranked_shape<[?]>, !shapex.ranked_shape<[?]>) -> !shapex.ranked_shape<[?]>
-  %2 = shape.broadcast %0, %1 : tensor<?xindex>, tensor<?xindex> -> tensor<?xindex>
-  // CHECK: %[[EXTENTS:.+]] = "shapex.to_extent_tensor"(%[[BROADCASTED]]) : (!shapex.ranked_shape<[?]>) -> tensor<1xindex>
-  %3 = tensor.cast %2 : tensor<?xindex> to tensor<1xindex>
-  // CHECK: "foo.use"(%[[EXTENTS]])
-  "foo.use"(%3) : (tensor<1xindex>) -> ()
-  return
-}
-
-// -----
-// Handling of existing shapex.from_extent_tensor ops.
-// CHECK-LABEL:   func @f
-func @f(%arg0: tensor<?x3x3x3x3xf32>, %arg1: tensor<?x3x3x3x3xf32>) -> (tensor<?x3x3x3x3xf32>) {
-  %0 = shape.shape_of %arg0 : tensor<?x3x3x3x3xf32> -> tensor<5xindex>
-  %1 = shape.shape_of %arg1 : tensor<?x3x3x3x3xf32> -> tensor<5xindex>
-  %2 = shape.broadcast %0, %1 : tensor<5xindex>, tensor<5xindex> -> tensor<5xindex>
-  // CHECK-NOT: from_extent_tensor
-  %3 = "shapex.from_extent_tensor"(%2) : (tensor<5xindex>) -> !shapex.ranked_shape<[?,?,?,?,?]>
-  %4 = "shapex.ranked_broadcast_in_dim"(%arg0, %3) {broadcast_dimensions = dense<[0, 1, 2, 3, 4]> : tensor<5xi64>} : (tensor<?x3x3x3x3xf32>, !shapex.ranked_shape<[?,?,?,?,?]>) -> tensor<?x3x3x3x3xf32>
-  %5 = "shapex.from_extent_tensor"(%2) : (tensor<5xindex>) -> !shapex.ranked_shape<[?,?,?,?,?]>
-  %6 = "shapex.ranked_broadcast_in_dim"(%arg1, %5) {broadcast_dimensions = dense<[0, 1, 2, 3, 4]> : tensor<5xi64>} : (tensor<?x3x3x3x3xf32>, !shapex.ranked_shape<[?,?,?,?,?]>) -> tensor<?x3x3x3x3xf32>
-  %7 = mhlo.add %4, %6 : tensor<?x3x3x3x3xf32>
-  return %7 : tensor<?x3x3x3x3xf32>
-}
diff --git a/iree/compiler/Dialect/Shape/Plugins/BUILD b/iree/compiler/Dialect/Shape/Plugins/BUILD
deleted file mode 100644
index 236a474..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/BUILD
+++ /dev/null
@@ -1,11 +0,0 @@
-# Copyright 2020 The IREE Authors
-#
-# Licensed under the Apache License v2.0 with LLVM Exceptions.
-# See https://llvm.org/LICENSE.txt for license information.
-# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-package(
-    default_visibility = ["//visibility:public"],
-    features = ["layering_check"],
-    licenses = ["notice"],  # Apache 2.0
-)
diff --git a/iree/compiler/Dialect/Shape/Plugins/CMakeLists.txt b/iree/compiler/Dialect/Shape/Plugins/CMakeLists.txt
deleted file mode 100644
index 20acde1..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/CMakeLists.txt
+++ /dev/null
@@ -1,13 +0,0 @@
-################################################################################
-# Autogenerated by build_tools/bazel_to_cmake/bazel_to_cmake.py from           #
-# iree/compiler/Dialect/Shape/Plugins/BUILD                                    #
-#                                                                              #
-# Use iree_cmake_extra_content from iree/build_defs.oss.bzl to add arbitrary   #
-# CMake-only content.                                                          #
-#                                                                              #
-# To disable autogeneration for this file entirely, delete this header.        #
-################################################################################
-
-iree_add_all_subdirs()
-
-### BAZEL_TO_CMAKE_PRESERVES_ALL_CONTENT_BELOW_THIS_LINE ###
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/BUILD b/iree/compiler/Dialect/Shape/Plugins/XLA/BUILD
deleted file mode 100644
index f2e8a25..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/BUILD
+++ /dev/null
@@ -1,27 +0,0 @@
-# Copyright 2020 The IREE Authors
-#
-# Licensed under the Apache License v2.0 with LLVM Exceptions.
-# See https://llvm.org/LICENSE.txt for license information.
-# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-package(
-    default_visibility = ["//visibility:public"],
-    features = ["layering_check"],
-    licenses = ["notice"],  # Apache 2.0
-)
-
-cc_library(
-    name = "XlaHloShapeBuilder",
-    srcs = [
-        "XlaHloShapeBuilder.cpp",
-    ],
-    hdrs = [
-        "XlaHloShapeBuilder.h",
-    ],
-    deps = [
-        "//iree/compiler/Dialect/Shape/IR",
-        "@llvm-project//llvm:Support",
-        "@llvm-project//mlir:IR",
-        "@mlir-hlo//:hlo",
-    ],
-)
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/CMakeLists.txt b/iree/compiler/Dialect/Shape/Plugins/XLA/CMakeLists.txt
deleted file mode 100644
index 1107476..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/CMakeLists.txt
+++ /dev/null
@@ -1,28 +0,0 @@
-################################################################################
-# Autogenerated by build_tools/bazel_to_cmake/bazel_to_cmake.py from           #
-# iree/compiler/Dialect/Shape/Plugins/XLA/BUILD                                #
-#                                                                              #
-# Use iree_cmake_extra_content from iree/build_defs.oss.bzl to add arbitrary   #
-# CMake-only content.                                                          #
-#                                                                              #
-# To disable autogeneration for this file entirely, delete this header.        #
-################################################################################
-
-iree_add_all_subdirs()
-
-iree_cc_library(
-  NAME
-    XlaHloShapeBuilder
-  HDRS
-    "XlaHloShapeBuilder.h"
-  SRCS
-    "XlaHloShapeBuilder.cpp"
-  DEPS
-    LLVMSupport
-    MLIRIR
-    iree::compiler::Dialect::Shape::IR
-    tensorflow::mlir_hlo
-  PUBLIC
-)
-
-### BAZEL_TO_CMAKE_PRESERVES_ALL_CONTENT_BELOW_THIS_LINE ###
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.cpp b/iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.cpp
deleted file mode 100644
index 5f12d30..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.cpp
+++ /dev/null
@@ -1,473 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include "iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.h"
-
-#include "iree/compiler/Dialect/Shape/IR/Builders.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeInterface.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "llvm/ADT/BitVector.h"
-#include "llvm/ADT/Optional.h"
-#include "mlir-hlo/Dialect/mhlo/IR/hlo_ops.h"
-#include "mlir/IR/BuiltinTypes.h"
-#include "mlir/IR/Value.h"
-
-using llvm::None;
-using llvm::Optional;
-using llvm::SmallVector;
-
-using namespace mlir::iree_compiler::Shape;
-
-namespace mlir {
-namespace mhlo {
-namespace {
-
-template <typename HloOp>
-Value rewriteXlaNaryElementwiseOpShape(RankedShapeType resultShape, HloOp op,
-                                       OpBuilder &builder) {
-  if (!op) return nullptr;
-  SmallVector<Value, 4> inputOperands(op.getOperands());
-  return buildCastInputsToResultShape(op.getLoc(), resultShape, inputOperands,
-                                      builder);
-}
-
-Value rewriteXlaDotOpShape(RankedShapeType resultRs, DotOp dotOp,
-                           OpBuilder &builder) {
-  auto lhsType = dotOp.lhs().getType().dyn_cast<RankedTensorType>();
-  auto rhsType = dotOp.rhs().getType().dyn_cast<RankedTensorType>();
-  auto resultType = dotOp.getResult().getType().dyn_cast<RankedTensorType>();
-  if (!lhsType || !rhsType || !resultType) return nullptr;
-
-  // Shape transfer function:
-  //  [n] dot [n] -> scalar
-  //  [m x k] dot [k] -> [m]
-  //  [m x k] dot [k x n] -> [m x n]
-  auto lhsRank = lhsType.getRank();
-  auto rhsRank = rhsType.getRank();
-  auto resultRank = resultType.getRank();
-  auto loc = dotOp.getLoc();
-  if (lhsRank == 1 && rhsRank == 1 && resultRank == 0) {
-    auto scalarShape = RankedShapeType::getChecked({}, loc);
-    return builder.create<ConstRankedShapeOp>(dotOp.getLoc(), scalarShape);
-  } else if (lhsRank == 2 && rhsRank == 1 && resultRank == 1) {
-    SmallVector<Value, 1> dynamicDims;
-    if (resultRs.isDimDynamic(0)) {
-      auto lhsGetShape = builder.create<GetRankedShapeOp>(
-          loc, RankedShapeType::get(lhsType.getShape(), builder.getContext()),
-          dotOp.lhs());
-      auto getMDim =
-          builder.create<RankedDimOp>(loc, builder.getIndexType(), lhsGetShape,
-                                      builder.getI64IntegerAttr(0));
-      dynamicDims.push_back(getMDim);
-    }
-    return builder.create<MakeRankedShapeOp>(loc, resultRs, dynamicDims);
-  } else if (lhsRank == 2 && rhsRank == 2 && resultRank == 2) {
-    SmallVector<Value, 2> dynamicDims;
-    if (resultRs.isDimDynamic(0)) {
-      auto lhsGetShape = builder.create<GetRankedShapeOp>(
-          loc, RankedShapeType::get(lhsType.getShape(), builder.getContext()),
-          dotOp.lhs());
-      auto getMDim =
-          builder.create<RankedDimOp>(loc, builder.getIndexType(), lhsGetShape,
-                                      builder.getI64IntegerAttr(0));
-      dynamicDims.push_back(getMDim);
-    }
-    if (resultRs.isDimDynamic(1)) {
-      auto rhsGetShape = builder.create<GetRankedShapeOp>(
-          loc, RankedShapeType::get(rhsType.getShape(), builder.getContext()),
-          dotOp.rhs());
-      auto getNDim =
-          builder.create<RankedDimOp>(loc, builder.getIndexType(), rhsGetShape,
-                                      builder.getI64IntegerAttr(1));
-      dynamicDims.push_back(getNDim);
-    }
-    return builder.create<MakeRankedShapeOp>(loc, resultRs, dynamicDims);
-  } else {
-    return nullptr;
-  }
-}
-
-Value rewriteReduce(RankedShapeType resultShape, ReduceOp reduceOp,
-                    OpBuilder &builder) {
-  Location loc = reduceOp.getLoc();
-
-  // Get a common operand shape.
-  Value operandShape;
-  SmallVector<Value, 4> operandShapes;
-  RankedShapeType operandRs;
-  for (auto operand : reduceOp.inputs()) {
-    auto shape = builder.create<GetRankedShapeOp>(loc, operand);
-    operandRs = shape.getRankedShape();
-    operandShapes.push_back(shape);
-  }
-  assert(!operandShapes.empty());
-  if (operandShapes.size() == 1) {
-    // Single operand.
-    operandShape = operandShapes.front();
-  } else {
-    // Multiple operands must be compatible.
-    operandShape =
-        builder.create<CastCompatibleShapeOp>(loc, operandRs, operandShapes);
-  }
-
-  // Map reduction dims onto operand dimensions.
-  SmallVector<bool, 4> isDimReduced;
-  isDimReduced.resize(operandRs.getRank());
-  for (const auto &apIntValue : reduceOp.dimensions().getIntValues()) {
-    auto intValue = apIntValue.getZExtValue();
-    assert(intValue < isDimReduced.size());
-    isDimReduced[intValue] = true;
-  }
-
-  // Map operand -> result dynamic dims.
-  assert(resultShape.getRank() ==
-         (operandRs.getRank() - reduceOp.dimensions().getNumElements()));
-  SmallVector<Value, 4> resultDims;
-  for (unsigned operandDimIndex = 0, e = isDimReduced.size();
-       operandDimIndex < e; ++operandDimIndex) {
-    unsigned resultDimIndex = resultDims.size();
-    // Skip reduced operand indices and non-dynamic result indices.
-    if (isDimReduced[operandDimIndex] ||
-        !resultShape.isDimDynamic(resultDimIndex))
-      continue;
-    resultDims.push_back(
-        builder.create<RankedDimOp>(loc, operandShape, operandDimIndex));
-  }
-
-  return builder.create<MakeRankedShapeOp>(loc, resultShape, resultDims);
-}
-
-// NOTE: This op is an HLO interloper and is just here until a corresponding
-// HLO is created. As such, it is included in this file even though not HLO
-// currently.
-Value rewriteShapexRankedBroadcastInDim(RankedShapeType resultShape,
-                                        RankedBroadcastInDimOp bidOp,
-                                        OpBuilder &builder) {
-  if (!bidOp) return nullptr;
-  return bidOp.result_shape();
-}
-
-Value rewriteShapexIota(RankedShapeType resultShape,
-                        iree_compiler::Shape::IotaOp iotaOp,
-                        OpBuilder &builder) {
-  if (!iotaOp) return nullptr;
-  return iotaOp.result_shape();
-}
-
-Value rewriteTranspose(RankedShapeType resultShape, TransposeOp transposeOp,
-                       OpBuilder &builder) {
-  if (!transposeOp) return nullptr;
-  auto loc = transposeOp.getLoc();
-
-  auto operandType =
-      transposeOp.operand().getType().dyn_cast<RankedTensorType>();
-  if (!operandType) return nullptr;
-
-  auto operandShapeValue = builder.create<GetRankedShapeOp>(
-      loc, RankedShapeType::get(operandType.getShape(), builder.getContext()),
-      transposeOp.operand());
-
-  SmallVector<int64_t, 4> perm;
-  for (const auto &permValue : transposeOp.permutation().getIntValues()) {
-    perm.push_back(permValue.getSExtValue());
-  }
-  assert(perm.size() == resultShape.getRank());
-
-  // Map the dynamic dims.
-  SmallVector<Value, 4> dynamicDims;
-  for (int64_t i = 0, e = resultShape.getRank(); i < e; ++i) {
-    if (!resultShape.isDimDynamic(i)) continue;
-    int64_t operandDim = perm[i];
-    auto dimValue = builder.create<RankedDimOp>(
-        loc, builder.getIndexType(), operandShapeValue,
-        builder.getI64IntegerAttr(operandDim));
-    dynamicDims.push_back(dimValue);
-  }
-
-  return builder.create<MakeRankedShapeOp>(loc, resultShape, dynamicDims);
-}
-
-// Returns a value of type `!shapex.ranked_shape` for the input value.
-static Value getRankedShapeAsValue(Value v, OpBuilder &builder, Location loc) {
-  assert(v.getType().isa<TensorType>());
-  auto type = v.getType().dyn_cast<RankedTensorType>();
-  if (!type) {
-    return nullptr;
-  }
-  return builder.create<GetRankedShapeOp>(
-      loc, RankedShapeType::get(type.getShape(), builder.getContext()), v);
-}
-
-// Returns a value representing the extent of dimension `dim`.
-static Value getExtent(Value v, int64_t dim, OpBuilder &builder, Location loc) {
-  return builder.create<RankedDimOp>(loc, v, dim);
-}
-
-Value rewriteDotGeneral(RankedShapeType resultShape, DotGeneralOp op,
-                        OpBuilder &builder) {
-  Location loc = op.getLoc();
-  auto lhsShape = getRankedShapeAsValue(op.lhs(), builder, loc);
-  auto rhsShape = getRankedShapeAsValue(op.rhs(), builder, loc);
-  if (!lhsShape || !rhsShape) {
-    return nullptr;
-  }
-  auto getFreeDims = [&](ArrayRef<int64_t> batchDims,
-                         ArrayRef<int64_t> contractingDims, int64_t rank) {
-    llvm::BitVector freeDims(rank, true);
-    for (int64_t dim : batchDims) {
-      freeDims.reset(dim);
-    }
-    for (int64_t dim : contractingDims) {
-      freeDims.reset(dim);
-    }
-    SmallVector<int64_t, 4> result;
-    for (auto bitIndex : freeDims.set_bits()) {
-      result.push_back(bitIndex);
-    }
-    return result;
-  };
-  auto lhsRankedShape = lhsShape.getType().cast<RankedShapeType>();
-  auto rhsRankedShape = rhsShape.getType().cast<RankedShapeType>();
-  auto dotDimensions = op.dot_dimension_numbers();
-  auto lhsFreeDims = getFreeDims(
-      llvm::to_vector<4>(
-          dotDimensions.lhs_batching_dimensions().getValues<int64_t>()),
-      llvm::to_vector<4>(
-          dotDimensions.lhs_contracting_dimensions().getValues<int64_t>()),
-      lhsRankedShape.getRank());
-  auto rhsFreeDims = getFreeDims(
-      llvm::to_vector<4>(
-          dotDimensions.rhs_batching_dimensions().getValues<int64_t>()),
-      llvm::to_vector<4>(
-          dotDimensions.rhs_contracting_dimensions().getValues<int64_t>()),
-      rhsRankedShape.getRank());
-
-  SmallVector<Value, 6> outputExtents;
-  for (int64_t dim :
-       dotDimensions.lhs_batching_dimensions().getValues<int64_t>()) {
-    // TODO(silvasean): Add a version of MakeRankedShapeOp that takes
-    // all dimensions. Having to constantly check if a dim is dynamic
-    // upon construction is a waste of time, more testing burden, etc.
-    // We can easily canonicalize to the more constrained one.
-    if (lhsRankedShape.isDimDynamic(dim)) {
-      outputExtents.push_back(getExtent(lhsShape, dim, builder, loc));
-    }
-  }
-  for (int64_t dim : lhsFreeDims) {
-    if (lhsRankedShape.isDimDynamic(dim)) {
-      outputExtents.push_back(getExtent(lhsShape, dim, builder, loc));
-    }
-  }
-  for (int64_t dim : rhsFreeDims) {
-    if (rhsRankedShape.isDimDynamic(dim)) {
-      outputExtents.push_back(getExtent(rhsShape, dim, builder, loc));
-    }
-  }
-  return builder.create<MakeRankedShapeOp>(loc, resultShape, outputExtents);
-}
-
-Value rewriteDynamicReshape(RankedShapeType resultShape, DynamicReshapeOp op,
-                            OpBuilder &builder) {
-  return builder.create<FromExtentTensorOp>(op.getLoc(), resultShape,
-                                            op.output_shape());
-}
-
-Value rewriteSelectOp(RankedShapeType resultShape, SelectOp selectOp,
-                      OpBuilder &builder) {
-  // TODO: broadcast case
-  if (!selectOp) return nullptr;
-  auto loc = selectOp.getLoc();
-
-  auto trueType = selectOp.on_true().getType().dyn_cast<RankedTensorType>();
-  auto falseType = selectOp.on_false().getType().dyn_cast<RankedTensorType>();
-  auto predType = selectOp.pred().getType().dyn_cast<RankedTensorType>();
-  if (!trueType || !falseType || !predType) return nullptr;
-
-  assert(trueType.getRank() == resultShape.getRank() &&
-         trueType.getRank() == falseType.getRank() &&
-         trueType.getRank() == predType.getRank());
-
-  auto operandShapeValue = builder.create<GetRankedShapeOp>(
-      loc, RankedShapeType::get(trueType.getShape(), builder.getContext()),
-      selectOp.on_true());
-
-  // Map the dynamic dims.
-  SmallVector<Value, 4> dynamicDims;
-  for (int64_t i = 0, e = resultShape.getRank(); i < e; ++i) {
-    if (!resultShape.isDimDynamic(i)) continue;
-    auto dimValue = builder.create<RankedDimOp>(loc, builder.getIndexType(),
-                                                operandShapeValue,
-                                                builder.getI64IntegerAttr(i));
-    dynamicDims.push_back(dimValue);
-  }
-
-  return builder.create<MakeRankedShapeOp>(loc, resultShape, dynamicDims);
-}
-
-Value rewriteTorchIndexSelect(RankedShapeType resultShape,
-                              TorchIndexSelectOp torchIndexSelectOp,
-                              OpBuilder &builder) {
-  if (!torchIndexSelectOp) return nullptr;
-  auto loc = torchIndexSelectOp.getLoc();
-
-  int64_t resultShapeRank = resultShape.getRank();
-  auto paramsType =
-      torchIndexSelectOp.input().getType().dyn_cast<RankedTensorType>();
-  auto indicesType =
-      torchIndexSelectOp.index().getType().dyn_cast<RankedTensorType>();
-  if (!paramsType || !indicesType) {
-    return nullptr;
-  }
-
-  auto axis = torchIndexSelectOp.dim();
-  auto batchDim = torchIndexSelectOp.batch_dims();
-  int64_t paramsRank = paramsType.getRank();
-  int64_t indicesRank = indicesType.getRank();
-
-  std::vector<int64_t> shape(paramsType.getShape());
-  int64_t axisValue = axis;
-  int64_t batchDimValue = batchDim;
-
-  // For neg axis values, we wrap around params,
-  // e.g. axis = -1 => params[:-1]
-  if (axisValue < 0) {
-    axisValue += paramsRank;
-  }
-  if (batchDimValue < 0) {
-    batchDimValue += indicesRank;
-  }
-
-  // params must be at least rank axis + 1
-  if (paramsRank < axisValue + 1) {
-    return nullptr;
-  }
-
-  auto paramsShapeValue = builder.create<GetRankedShapeOp>(
-      loc, RankedShapeType::get(paramsType.getShape(), builder.getContext()),
-      torchIndexSelectOp.input());
-  auto indicesShapeValue = builder.create<GetRankedShapeOp>(
-      loc, RankedShapeType::get(indicesType.getShape(), builder.getContext()),
-      torchIndexSelectOp.index());
-
-  SmallVector<Value, 4> dynamicDims;
-#define GENERATE_RANKED_DIM_OP(value, index)                                   \
-  do {                                                                         \
-    auto dimValue = builder.create<RankedDimOp>(                               \
-        loc, builder.getIndexType(), value, builder.getI64IntegerAttr(index)); \
-    dynamicDims.push_back(dimValue);                                           \
-  } while (0)
-
-  if (indicesRank == 0) {
-    // Scalar indices (output is rank(params) - 1).
-    if (resultShapeRank != paramsRank - 1) {
-      return nullptr;
-    }
-
-    // params.shape[:axis] + params.shape[axis+1:]
-    for (int64_t i = 0; i < paramsRank; ++i) {
-      if ((i == axisValue) || (i < axisValue && !resultShape.isDimDynamic(i)) ||
-          (i > axisValue && !resultShape.isDimDynamic(i - 1)))
-        continue;
-      GENERATE_RANKED_DIM_OP(paramsShapeValue, i);
-    }
-  } else if (indicesRank == 1) {
-    // Vector indices (output is rank(params)).
-    // Copy indices.shape into params.shape[axis]
-    if (resultShapeRank != paramsRank) {
-      return nullptr;
-    }
-
-    // params.shape[:axis] + indices.shape[batch_dims:]
-    //   + params.shape[indicesRank-batchDim+axisValue:]
-    int resultShapeIndex = 0;
-    // params.shape[:axis]
-    for (int64_t i = 0; i < axisValue; ++i) {
-      if (!resultShape.isDimDynamic(resultShapeIndex++)) continue;
-      GENERATE_RANKED_DIM_OP(paramsShapeValue, i);
-    }
-    // indices.shape[:batchDim]
-    for (int64_t i = batchDimValue;
-         i < indicesRank && resultShapeIndex < resultShapeRank; ++i) {
-      if (!resultShape.isDimDynamic(resultShapeIndex++)) continue;
-      GENERATE_RANKED_DIM_OP(indicesShapeValue, i);
-    }
-    // params.shape[indicesRank-batchDim+axisValue:]
-    // resultShapeIndex == indicesRank-batchDim+axisValue
-    for (int64_t i = resultShapeIndex; i < resultShapeRank; ++i) {
-      if (!resultShape.isDimDynamic(resultShapeIndex++)) continue;
-      GENERATE_RANKED_DIM_OP(paramsShapeValue, i);
-    }
-  } else {
-    // params.shape[:axis] + indices.shape[batch_dims:] + params.shape[axis +
-    // 1:]
-    // The expected rank is (paramsRank-1) + (indicesRank-batchDim)
-    auto expectedRank = paramsRank - 1 + indicesRank - batchDimValue;
-    if (resultShapeRank != expectedRank) {
-      return nullptr;
-    }
-
-    int resultShapeIndex = 0;
-    for (int64_t i = 0; i < axisValue; ++i) {
-      if (!resultShape.isDimDynamic(resultShapeIndex++)) continue;
-      GENERATE_RANKED_DIM_OP(paramsShapeValue, i);
-    }
-
-    for (int64_t i = batchDimValue; i < indicesRank; ++i) {
-      if (!resultShape.isDimDynamic(resultShapeIndex++)) continue;
-      GENERATE_RANKED_DIM_OP(indicesShapeValue, i);
-    }
-
-    for (int64_t i = axisValue + 1;
-         i < paramsRank && resultShapeIndex < resultShapeRank; ++i) {
-      if (!resultShape.isDimDynamic(resultShapeIndex++)) continue;
-      GENERATE_RANKED_DIM_OP(paramsShapeValue, i);
-    }
-  }
-#undef GENERATE_RANKED_DIM_OP
-
-  return builder.create<MakeRankedShapeOp>(loc, resultShape, dynamicDims);
-}
-
-}  // namespace
-
-// Creates a custom op shape builder for XLA-HLO ops that are not otherwise
-// supported through traits or other declarative means.
-void populateXlaHloCustomOpShapeBuilder(CustomOpShapeBuilderList &builders) {
-  auto &b = builders.make<CallbackCustomOpShapeBuilder>();
-  // NOTE: Most of these *should not* be "custom ops". They should be coming
-  // from declarative shape information, but that doesn't exist yet.
-#define INSERT_EW_OP(OpTy) \
-  b.insertOpRankedShapeBuilder<OpTy>(rewriteXlaNaryElementwiseOpShape<OpTy>);
-  INSERT_EW_OP(AddOp);
-  INSERT_EW_OP(Atan2Op);
-  INSERT_EW_OP(DivOp);
-  INSERT_EW_OP(MaxOp);
-  INSERT_EW_OP(MinOp);
-  INSERT_EW_OP(MulOp);
-  INSERT_EW_OP(PowOp);
-  INSERT_EW_OP(RemOp);
-  INSERT_EW_OP(ShiftLeftOp);
-  INSERT_EW_OP(ShiftRightArithmeticOp);
-  INSERT_EW_OP(ShiftRightLogicalOp);
-  INSERT_EW_OP(SubOp);
-  INSERT_EW_OP(CompareOp);
-  INSERT_EW_OP(ClampOp);
-
-  b.insertOpRankedShapeBuilder<SelectOp>(rewriteSelectOp);
-  b.insertOpRankedShapeBuilder<DotOp>(rewriteXlaDotOpShape);
-  b.insertOpRankedShapeBuilder<RankedBroadcastInDimOp>(
-      rewriteShapexRankedBroadcastInDim);
-  b.insertOpRankedShapeBuilder<iree_compiler::Shape::IotaOp>(rewriteShapexIota);
-  b.insertOpRankedShapeBuilder<ReduceOp>(rewriteReduce);
-  b.insertOpRankedShapeBuilder<TransposeOp>(rewriteTranspose);
-  b.insertOpRankedShapeBuilder<mhlo::DotGeneralOp>(rewriteDotGeneral);
-  b.insertOpRankedShapeBuilder<mhlo::DynamicReshapeOp>(rewriteDynamicReshape);
-  b.insertOpRankedShapeBuilder<mhlo::TorchIndexSelectOp>(
-      rewriteTorchIndexSelect);
-}
-
-}  // namespace mhlo
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.h b/iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.h
deleted file mode 100644
index e1783f4..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.h
+++ /dev/null
@@ -1,23 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#ifndef IREE_COMPILER_DIALECT_SHAPE_IR_XLAHLOSHAPEBUILDER_H_
-#define IREE_COMPILER_DIALECT_SHAPE_IR_XLAHLOSHAPEBUILDER_H_
-
-#include "iree/compiler/Dialect/Shape/IR/ShapeInterface.h"
-
-namespace mlir {
-namespace mhlo {
-
-// Creates a custom op shape builder for XLA-HLO ops that are not otherwise
-// supported through traits or other declarative means.
-void populateXlaHloCustomOpShapeBuilder(
-    iree_compiler::Shape::CustomOpShapeBuilderList &builders);
-
-}  // namespace mhlo
-}  // namespace mlir
-
-#endif  // IREE_COMPILER_DIALECT_SHAPE_IR_XLAHLOSHAPEBUILDER_H_
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/test/BUILD b/iree/compiler/Dialect/Shape/Plugins/XLA/test/BUILD
deleted file mode 100644
index e7eb829..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/test/BUILD
+++ /dev/null
@@ -1,26 +0,0 @@
-# Copyright 2020 The IREE Authors
-#
-# Licensed under the Apache License v2.0 with LLVM Exceptions.
-# See https://llvm.org/LICENSE.txt for license information.
-# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-load("//iree:lit_test.bzl", "iree_lit_test_suite")
-load("//build_tools/bazel:enforce_glob.bzl", "enforce_glob")
-
-package(
-    default_visibility = ["//visibility:public"],
-    features = ["layering_check"],
-    licenses = ["notice"],  # Apache 2.0
-)
-
-iree_lit_test_suite(
-    name = "lit",
-    srcs = enforce_glob(
-        ["custom_ops.mlir"],
-        include = ["*.mlir"],
-    ),
-    data = [
-        "//iree/tools:IreeFileCheck",
-        "//iree/tools:iree-opt",
-    ],
-)
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/test/CMakeLists.txt b/iree/compiler/Dialect/Shape/Plugins/XLA/test/CMakeLists.txt
deleted file mode 100644
index 2fd4f81..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/test/CMakeLists.txt
+++ /dev/null
@@ -1,23 +0,0 @@
-################################################################################
-# Autogenerated by build_tools/bazel_to_cmake/bazel_to_cmake.py from           #
-# iree/compiler/Dialect/Shape/Plugins/XLA/test/BUILD                           #
-#                                                                              #
-# Use iree_cmake_extra_content from iree/build_defs.oss.bzl to add arbitrary   #
-# CMake-only content.                                                          #
-#                                                                              #
-# To disable autogeneration for this file entirely, delete this header.        #
-################################################################################
-
-iree_add_all_subdirs()
-
-iree_lit_test_suite(
-  NAME
-    lit
-  SRCS
-    "custom_ops.mlir"
-  DATA
-    iree::tools::IreeFileCheck
-    iree::tools::iree-opt
-)
-
-### BAZEL_TO_CMAKE_PRESERVES_ALL_CONTENT_BELOW_THIS_LINE ###
diff --git a/iree/compiler/Dialect/Shape/Plugins/XLA/test/custom_ops.mlir b/iree/compiler/Dialect/Shape/Plugins/XLA/test/custom_ops.mlir
deleted file mode 100644
index 4a03bf9..0000000
--- a/iree/compiler/Dialect/Shape/Plugins/XLA/test/custom_ops.mlir
+++ /dev/null
@@ -1,48 +0,0 @@
-// RUN: iree-opt -split-input-file -verify-diagnostics -iree-shape-materialize-calculations %s | IreeFileCheck %s
-
-// CHECK-LABEL: @transpose
-func @transpose(%arg0: tensor<?x7x10xf32>, %arg1: !shapex.ranked_shape<[?,7,10]>) -> (tensor<7x?x10xf32>, !shapex.ranked_shape<[7,?,10]>) {
-  %tied = shapex.tie_shape %arg0, %arg1 : tensor<?x7x10xf32>, !shapex.ranked_shape<[?,7,10]>
-  %0 = "mhlo.transpose"(%tied) {permutation = dense<[1, 0, 2]> : tensor<3xi64>} :
-      (tensor<?x7x10xf32>) -> tensor<7x?x10xf32>
-  // CHECK-DAG: %[[RESULT:.+]] = "mhlo.transpose"
-  // CHECK-DAG: %[[DIM:.+]] = shapex.ranked_dim %arg1[0]
-  // CHECK-DAG: %[[SHAPE:.+]] = shapex.make_ranked_shape %[[DIM]]
-  %1 = shapex.get_ranked_shape %0 : tensor<7x?x10xf32> -> !shapex.ranked_shape<[7,?,10]>
-  // CHECK: return %[[RESULT]], %[[SHAPE]]
-  return %0, %1 : tensor<7x?x10xf32>, !shapex.ranked_shape<[7,?,10]>
-}
-
-// -----
-// CHECK-LABEL: func @dot_general
-func @dot_general(%arg0: tensor<?x?x?xf32>, %arg1: tensor<?x?x?xf32>,
-                  %arg2: !shapex.ranked_shape<[?,?,?]>, %arg3: !shapex.ranked_shape<[?,?,?]>) -> !shapex.ranked_shape<[?,?,?]> {
-  %tie0 = shapex.tie_shape %arg0, %arg2 : tensor<?x?x?xf32>, !shapex.ranked_shape<[?,?,?]>
-  %tie1 = shapex.tie_shape %arg1, %arg3 : tensor<?x?x?xf32>, !shapex.ranked_shape<[?,?,?]>
-
-  // Extents are taken directly from args.
-  // CHECK-DAG: %[[EXTENT0:.+]] = shapex.ranked_dim %arg2[0]
-  // CHECK-DAG: %[[EXTENT1:.+]] = shapex.ranked_dim %arg2[1]
-  // CHECK-DAG: %[[EXTENT2:.+]] = shapex.ranked_dim %arg3[2]
-  // CHECK-DAG: %[[SHAPE:.+]] = shapex.make_ranked_shape %[[EXTENT0]], %[[EXTENT1]], %[[EXTENT2]]
-  // CHECK-DAG: return %[[SHAPE]]
-  %0 = "mhlo.dot_general"(%tie0, %tie1) { dot_dimension_numbers = {
-    lhs_batching_dimensions = dense<0> : tensor<1xi64>,
-    lhs_contracting_dimensions = dense<2> : tensor<1xi64>,
-    rhs_batching_dimensions = dense<0> : tensor<1xi64>,
-    rhs_contracting_dimensions = dense<1> : tensor<1xi64>
-  }} : (tensor<?x?x?xf32>, tensor<?x?x?xf32>) -> tensor<?x?x?xf32>
-  %1 = shapex.get_ranked_shape %0 : tensor<?x?x?xf32> -> !shapex.ranked_shape<[?,?,?]>
-  return %1 : !shapex.ranked_shape<[?,?,?]>
-}
-
-// -----
-
-// CHECK-LABEL: func @dynamic_reshape
-func @dynamic_reshape(%arg0: tensor<?xf32>, %arg1: tensor<2xindex>) -> !shapex.ranked_shape<[?,?]> {
-  // CHECK-DAG: %[[SHAPE:.+]] = "shapex.from_extent_tensor"(%arg1)
-  // CHECK-DAG: return %[[SHAPE]]
-  %0 = "mhlo.dynamic_reshape"(%arg0, %arg1) : (tensor<?xf32>, tensor<2xindex>) -> tensor<?x?xf32>
-  %1 = shapex.get_ranked_shape %0 : tensor<?x?xf32> -> !shapex.ranked_shape<[?,?]>
-  return %1 : !shapex.ranked_shape<[?,?]>
-}
diff --git a/iree/compiler/Dialect/Shape/Transforms/BUILD b/iree/compiler/Dialect/Shape/Transforms/BUILD
index ff64c9d..5efac09 100644
--- a/iree/compiler/Dialect/Shape/Transforms/BUILD
+++ b/iree/compiler/Dialect/Shape/Transforms/BUILD
@@ -14,21 +14,13 @@
     name = "Transforms",
     srcs = [
         "CleanupPlaceholdersPass.cpp",
-        "ConvertHLOToShapeDialectPass.cpp",
-        "ConvertShapeToStandard.cpp",
         "FunctionSignatureExpansionPass.cpp",
-        "HoistShapeCalculationsPass.cpp",
-        "MaterializeShapeCalculations.cpp",
-        "MaterializeShapeCalculationsPass.cpp",
-        "TieDynamicShapesPass.cpp",
     ],
     hdrs = [
         "Passes.h",
-        "Patterns.h",
     ],
     deps = [
         "//iree/compiler/Dialect/Shape/IR",
-        "//iree/compiler/Dialect/Shape/Plugins/XLA:XlaHloShapeBuilder",
         "//iree/compiler/Dialect/Shape/Utils:TypeConversion",
         "//iree/compiler/Utils",
         "@llvm-project//llvm:Support",
@@ -40,6 +32,5 @@
         "@llvm-project//mlir:Support",
         "@llvm-project//mlir:TensorDialect",
         "@llvm-project//mlir:Transforms",
-        "@mlir-hlo//:hlo",
     ],
 )
diff --git a/iree/compiler/Dialect/Shape/Transforms/CMakeLists.txt b/iree/compiler/Dialect/Shape/Transforms/CMakeLists.txt
index 1ad3803..5f50ced 100644
--- a/iree/compiler/Dialect/Shape/Transforms/CMakeLists.txt
+++ b/iree/compiler/Dialect/Shape/Transforms/CMakeLists.txt
@@ -15,16 +15,9 @@
     Transforms
   HDRS
     "Passes.h"
-    "Patterns.h"
   SRCS
     "CleanupPlaceholdersPass.cpp"
-    "ConvertHLOToShapeDialectPass.cpp"
-    "ConvertShapeToStandard.cpp"
     "FunctionSignatureExpansionPass.cpp"
-    "HoistShapeCalculationsPass.cpp"
-    "MaterializeShapeCalculations.cpp"
-    "MaterializeShapeCalculationsPass.cpp"
-    "TieDynamicShapesPass.cpp"
   DEPS
     LLVMSupport
     MLIRAnalysis
@@ -36,10 +29,8 @@
     MLIRTensor
     MLIRTransforms
     iree::compiler::Dialect::Shape::IR
-    iree::compiler::Dialect::Shape::Plugins::XLA::XlaHloShapeBuilder
     iree::compiler::Dialect::Shape::Utils::TypeConversion
     iree::compiler::Utils
-    tensorflow::mlir_hlo
   PUBLIC
 )
 
diff --git a/iree/compiler/Dialect/Shape/Transforms/ConvertHLOToShapeDialectPass.cpp b/iree/compiler/Dialect/Shape/Transforms/ConvertHLOToShapeDialectPass.cpp
deleted file mode 100644
index 7e68a46..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/ConvertHLOToShapeDialectPass.cpp
+++ /dev/null
@@ -1,100 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include <numeric>
-
-#include "iree/compiler/Dialect/Shape/IR/ShapeDialect.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeTypes.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
-#include "mlir-hlo/Dialect/mhlo/IR/hlo_ops.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/IR/BuiltinTypes.h"
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Pass/PassRegistry.h"
-#include "mlir/Transforms/DialectConversion.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-namespace {
-
-class ConvertDynamicBroadcastInDim
-    : public OpConversionPattern<mhlo::DynamicBroadcastInDimOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      mhlo::DynamicBroadcastInDimOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    mhlo::DynamicBroadcastInDimOp::Adaptor adapter(operands);
-    Value rankedShape = rewriter.create<Shape::FromExtentTensorOp>(
-        op.getLoc(), adapter.output_dimensions());
-    rewriter.replaceOpWithNewOp<Shape::RankedBroadcastInDimOp>(
-        op, op.getType(), adapter.operand(), rankedShape,
-        op.broadcast_dimensions());
-    return success();
-  }
-};
-
-class ConvertDynamicIota : public OpConversionPattern<mhlo::DynamicIotaOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      mhlo::DynamicIotaOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    auto resultTy = op.getType().cast<ShapedType>();
-    if (resultTy.getRank() != 1) {
-      return failure();
-    }
-
-    auto rankedShape = rewriter.create<Shape::FromExtentTensorOp>(
-        op.getLoc(), op.getOperand());
-    rewriter.replaceOpWithNewOp<Shape::IotaOp>(op, op.getType(), rankedShape);
-    return success();
-  }
-};
-
-class ConvertHLOToShapePass
-    : public PassWrapper<ConvertHLOToShapePass, FunctionPass> {
-  void getDependentDialects(DialectRegistry &registry) const override {
-    registry.insert<ShapeDialect>();
-  }
-
-  void runOnFunction() override {
-    ConversionTarget conversionTarget(getContext());
-    OwningRewritePatternList conversionPatterns(&getContext());
-
-    conversionTarget.addLegalDialect<ShapeDialect>();
-    conversionTarget.addLegalDialect<StandardOpsDialect>();
-    conversionTarget.addLegalDialect<mhlo::MhloDialect>();
-
-    conversionTarget.addIllegalOp<mhlo::DynamicBroadcastInDimOp>();
-    conversionPatterns.insert<ConvertDynamicBroadcastInDim>(&getContext());
-
-    conversionTarget.addIllegalOp<mhlo::DynamicIotaOp>();
-    conversionPatterns.insert<ConvertDynamicIota>(&getContext());
-
-    if (failed(applyPartialConversion(getFunction(), conversionTarget,
-                                      std::move(conversionPatterns)))) {
-      return signalPassFailure();
-    }
-  }
-};
-
-}  // namespace
-
-// Converts shape-sensitive HLOs to be based on facilities in the shape
-// dialect.
-std::unique_ptr<OperationPass<FuncOp>> createConvertHLOToShapePass() {
-  return std::make_unique<Shape::ConvertHLOToShapePass>();
-}
-
-static PassRegistration<Shape::ConvertHLOToShapePass> pass(
-    "iree-shape-convert-hlo",
-    "Converts dynamic shape dependent HLO ops to shaped variants.");
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Transforms/ConvertShapeToStandard.cpp b/iree/compiler/Dialect/Shape/Transforms/ConvertShapeToStandard.cpp
deleted file mode 100644
index fc803bc..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/ConvertShapeToStandard.cpp
+++ /dev/null
@@ -1,72 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/Dialect/Tensor/IR/Tensor.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Support/LogicalResult.h"
-#include "mlir/Transforms/DialectConversion.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-
-namespace {
-
-class ConvertFromExtent : public OpConversionPattern<FromExtentTensorOp> {
-  using OpConversionPattern::OpConversionPattern;
-  LogicalResult matchAndRewrite(
-      FromExtentTensorOp op, ArrayRef<Value> operands,
-      ConversionPatternRewriter &rewriter) const override {
-    auto input = op.extent_tensor();
-    ShapedType inputTy = input.getType().cast<ShapedType>();
-    if (!inputTy.hasRank() || inputTy.getRank() != 1) {
-      return failure();
-    }
-
-    llvm::SmallVector<Value, 4> extracted_elements;
-    auto valueCount = inputTy.getDimSize(0);
-    extracted_elements.reserve(valueCount);
-    for (int i = 0; i < valueCount; i++) {
-      auto index = rewriter.create<ConstantIndexOp>(op.getLoc(), i);
-      Value dim = rewriter.create<tensor::ExtractOp>(
-          op.getLoc(), inputTy.getElementType(), input, index.getResult());
-      if (!dim.getType().isIndex()) {
-        dim = rewriter.create<IndexCastOp>(op.getLoc(), rewriter.getIndexType(),
-                                           dim);
-      }
-      extracted_elements.push_back(dim);
-    }
-
-    SmallVector<int64_t, 4> dims;
-    dims.resize(valueCount, -1);
-    rewriter.replaceOpWithNewOp<Shape::MakeRankedShapeOp>(
-        op, Shape::RankedShapeType::get(dims, op.getContext()),
-        extracted_elements);
-
-    return success();
-  }
-};
-
-}  // namespace
-
-// Populates patterns that will convert shape calculations into standard ops.
-void populateShapeToStandardConversionPatterns(
-    OwningRewritePatternList &patterns, MLIRContext *context) {
-  patterns.insert<ConvertFromExtent>(context);
-}
-
-// Sets up legality for shape calculation materialization conversions.
-void setupShapeToStandardLegality(ConversionTarget &target) {
-  target.addIllegalOp<FromExtentTensorOp>();
-  target.addLegalOp<Shape::MakeRankedShapeOp>();
-}
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Transforms/HoistShapeCalculationsPass.cpp b/iree/compiler/Dialect/Shape/Transforms/HoistShapeCalculationsPass.cpp
deleted file mode 100644
index 1019f5c..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/HoistShapeCalculationsPass.cpp
+++ /dev/null
@@ -1,151 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include <algorithm>
-#include <iterator>
-
-#include "iree/compiler/Dialect/Shape/IR/ShapeDialect.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeTypes.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/IR/Dominance.h"
-#include "mlir/Pass/Pass.h"
-#include "mlir/Pass/PassRegistry.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-namespace {
-
-bool isSimpleShapeCalculationOp(Operation *op) {
-  // The op must have no side effects.
-  if (!MemoryEffectOpInterface::hasNoEffect(op) || op->getNumRegions() != 0) {
-    return false;
-  }
-  // The op should operate on types that are likely shape calculations.
-  // The exact predicate used here isn't too important. The main important thing
-  // is that we don't want to include ops on tensors.
-  for (Type type : op->getOperandTypes()) {
-    if (type.isa<TensorType>()) {
-      return false;
-    }
-  }
-  return true;
-}
-
-// Return an operation in `block` that defines `v`, if one exists.
-Operation *getDefiningOpInBlock(Value v, Block &block) {
-  if (OpResult opResult = v.dyn_cast<OpResult>()) {
-    if (opResult.getOwner()->getBlock() == &block) {
-      return opResult.getOwner();
-    }
-  }
-  return nullptr;
-}
-
-DenseSet<Operation *> calculateOpsToHoist(Block &block) {
-  // Strategy:
-  // Backward DFS from shapex.tie_shape shape operands (second
-  // operands), staying within the block and not incorporating ops that don't
-  // satisfy `isSimpleShapeCalculationOp`.
-
-  SmallVector<Operation *, 16> worklist;
-  // The return value, and also used as a "visited" set for our DFS below.
-  DenseSet<Operation *> opsToHoistSet;
-
-  // The worklist is initially populated with shape-producing ops defined in
-  // this block.
-  //
-  // We are only hoisting ops within the block, so block arguments and any
-  // values defined outside the block (which already dominate the entire block)
-  // don't matter.
-  for (Operation &op : block) {
-    if (auto tieShape = dyn_cast<Shape::TieShapeOp>(op)) {
-      if (Operation *op = getDefiningOpInBlock(tieShape.shape(), block)) {
-        worklist.push_back(op);
-      }
-    }
-  }
-  while (!worklist.empty()) {
-    Operation *op = worklist.pop_back_val();
-    if (!isSimpleShapeCalculationOp(op)) {
-      continue;
-    }
-    if (opsToHoistSet.insert(op).second) {
-      for (Value v : op->getOperands()) {
-        if (Operation *op = getDefiningOpInBlock(v, block)) {
-          worklist.push_back(op);
-        }
-      }
-    }
-  }
-  return opsToHoistSet;
-}
-
-void hoistOps(DenseSet<Operation *> opsToHoistSet, Block &block) {
-  auto opsToHoist = llvm::to_vector<16>(opsToHoistSet);
-  llvm::stable_sort(opsToHoist, [&](Operation *lhs, Operation *rhs) {
-    return lhs->isBeforeInBlock(rhs);
-  });
-
-  for (Operation *op : opsToHoist) {
-    Operation *moveAfter = nullptr;
-    for (Value operand : op->getOperands()) {
-      if (Operation *definingOp = getDefiningOpInBlock(operand, block)) {
-        if (moveAfter == nullptr || moveAfter->isBeforeInBlock(definingOp)) {
-          moveAfter = definingOp;
-        }
-      }
-    }
-    if (moveAfter != nullptr) {
-      op->moveAfter(moveAfter);
-    } else {
-      op->moveBefore(&block, block.begin());
-    }
-  }
-}
-
-// Best-effort pass for hoisting shape calculations earlier in the program.
-// We currently don't provide any hard guarantees about exactly what invariants
-// are established by this pass.
-//
-// The goal of this pass is to unblock further progress on dynamic shape
-// support. One pragmatic thing we observe is that for IREE, dispatch region
-// formation requires that when there is a `shapex.tie_shape %tensor, %shape`
-// op, to even properly form the dispatch region, IREE needs `%shape` to
-// dominate `%tensor` since the dispatch region's "workload" is derived from the
-// shape.
-//
-// This pass doesn't have a cost model, so it shouldn't be considered a generic
-// "hoist stuff to make things faster" type of pass. It's strictly a
-// best-effort pass to make certain lowerings work, albeit on somewhat shaky
-// ground. Longer-term, IREE's dispatch region formation will use a more
-// sophisticated algorithm and the analysis/hoisting done here will be a
-// byproduct of the dispatch region formation legality analysis/preparation.
-class HoistShapeCalculations
-    : public PassWrapper<HoistShapeCalculations, FunctionPass> {
- public:
-  void runOnFunction() override {
-    auto func = getFunction();
-    for (Block &block : func) {
-      DenseSet<Operation *> opsToHoist = calculateOpsToHoist(block);
-      hoistOps(opsToHoist, block);
-    }
-  }
-};
-}  // namespace
-
-std::unique_ptr<OperationPass<FuncOp>> createHoistShapeCalculationsPass() {
-  return std::make_unique<HoistShapeCalculations>();
-}
-
-static PassRegistration<HoistShapeCalculations> pass(
-    "iree-shape-hoist-shape-calculations",
-    "Best-effort shape calculation hoisting.");
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Transforms/MaterializeShapeCalculations.cpp b/iree/compiler/Dialect/Shape/Transforms/MaterializeShapeCalculations.cpp
deleted file mode 100644
index 7ca07aa..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/MaterializeShapeCalculations.cpp
+++ /dev/null
@@ -1,287 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include "iree/compiler/Dialect/Shape/IR/Builders.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeDialect.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeInterface.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeTypes.h"
-#include "iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Patterns.h"
-#include "iree/compiler/Utils/PatternUtils.h"
-#include "llvm/Support/Debug.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/IR/BuiltinTypes.h"
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/OpDefinition.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Pass/PassRegistry.h"
-#include "mlir/Support/LogicalResult.h"
-
-#define DEBUG_TYPE "iree-shape"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-
-namespace {
-
-// Gets a CustomOpShapeBuilderList for expanding shapes of custom ops.
-// By default, returns nullptr, which will not handle custom op shapes.
-// TODO(laurenzo): Since it isn't clear yet whether we need this facility
-// long term (i.e. this should come from the ops themselves), we are just
-// hard-linking it here at the expense of a dependency problem. Decouple this
-// if the facility persists.
-const CustomOpShapeBuilderList *getCustomOpShapeBuilder() {
-  static CustomOpShapeBuilderList globalBuilders = ([]() {
-    CustomOpShapeBuilderList builders;
-    mhlo::populateXlaHloCustomOpShapeBuilder(builders);
-    return builders;
-  })();
-  return &globalBuilders;
-}
-
-Value rewriteShapexRankedBroadcastShape(
-    RankedBroadcastShapeOp op, RankedBroadcastShapeOp::Adaptor operands,
-    OpBuilder &builder) {
-  auto lhs = operands.lhs();
-  auto rhs = operands.rhs();
-  auto loc = op.getLoc();
-  auto resultRs = op.getResult().getType().cast<RankedShapeType>();
-
-  auto c1 = builder.create<ConstantIndexOp>(loc, 1);
-  // Entries are the extent of the output along that dimension corresponding to
-  // the given side, or 1 (which is neutral w.r.t. broadcasting).
-  SmallVector<Value, 4> lhsResultExtents(resultRs.getRank(), c1);
-  SmallVector<Value, 4> rhsResultExtents(resultRs.getRank(), c1);
-
-  for (auto dim : llvm::enumerate(op.lhs_broadcast_dimensions())) {
-    auto inputDim = dim.index();
-    auto outputDim = dim.value().getZExtValue();
-    lhsResultExtents[outputDim] =
-        builder.create<RankedDimOp>(loc, lhs, inputDim);
-  }
-  for (auto dim : llvm::enumerate(op.rhs_broadcast_dimensions())) {
-    auto inputDim = dim.index();
-    auto outputDim = dim.value().getZExtValue();
-    rhsResultExtents[outputDim] =
-        builder.create<RankedDimOp>(loc, rhs, inputDim);
-  }
-
-  SmallVector<Value, 4> resultExtents;
-  for (auto t : llvm::zip(lhsResultExtents, rhsResultExtents)) {
-    auto lhsExtent = std::get<0>(t);
-    auto rhsExtent = std::get<1>(t);
-    auto ugt =
-        builder.create<CmpIOp>(loc, CmpIPredicate::ugt, lhsExtent, rhsExtent);
-    auto max = builder.create<SelectOp>(loc, ugt, lhsExtent, rhsExtent);
-    resultExtents.push_back(max);
-    // TODO(silvasean): Create error handling code for invalid broadcasts.
-    // Use vm.cond_fail (or something that lowers to that).
-  }
-
-  // MakeRankedShapeOp only accepts the dynamic dims, so filter appropriately.
-  SmallVector<Value, 4> filteredResultExtents;
-  for (int i = 0, e = resultRs.getRank(); i < e; i++) {
-    if (resultRs.isDimDynamic(i)) {
-      filteredResultExtents.push_back(resultExtents[i]);
-    }
-  }
-
-  return builder.create<MakeRankedShapeOp>(loc, resultRs,
-                                           filteredResultExtents);
-}
-
-LogicalResult expandGatherExtentsOp(GatherExtentsOp op,
-                                    GatherExtentsOp::Adaptor operands,
-                                    PatternRewriter &rewriter) {
-  // Calculate cumulative sums of the ranks of each operand, which allows
-  // us to map each index to its corresponding operand easily.
-  SmallVector<int64_t, 6> cumsum;
-  cumsum.push_back(0);
-  for (auto operand : operands.shapes()) {
-    auto rank = operand.getType().cast<Shape::RankedShapeType>().getRank();
-    cumsum.push_back(cumsum.back() + rank);
-  }
-
-  // For each index, extract the relevant extent from the operands.
-  SmallVector<Value, 6> extents;
-  for (auto index : op.indices().getValues<int64_t>()) {
-    auto it = llvm::upper_bound(cumsum, index) - 1;
-    auto operandNum = std::distance(cumsum.begin(), it);
-    auto dimNum = index - *it;
-    auto extent = rewriter.create<Shape::RankedDimOp>(
-        op.getLoc(), operands.shapes()[operandNum], dimNum);
-    extents.push_back(extent);
-  }
-
-  // Due to a quirk of MakeRankedShapeOp, we only want the dynamic
-  // dimensions.
-  SmallVector<Value, 6> onlyDynamicExtents;
-  auto resultType = op.result().getType().cast<Shape::RankedShapeType>();
-  for (int i = 0, e = resultType.getRank(); i < e; i++) {
-    if (resultType.isDimDynamic(i)) {
-      onlyDynamicExtents.push_back(extents[i]);
-    }
-  }
-
-  rewriter.replaceOpWithNewOp<Shape::MakeRankedShapeOp>(op, resultType,
-                                                        onlyDynamicExtents);
-  return success();
-}
-
-LogicalResult expandRankedBroadcastShapePattern(
-    RankedBroadcastShapeOp bcastOp, RankedBroadcastShapeOp::Adaptor operands,
-    PatternRewriter &rewriter) {
-  auto newValue =
-      rewriteShapexRankedBroadcastShape(bcastOp, operands, rewriter);
-  if (!newValue) return failure();
-
-  rewriter.replaceOp(bcastOp, newValue);
-  return success();
-}
-
-LogicalResult rewriteSameOperandsAndResultShape(GetRankedShapeOp getShapeOp,
-                                                Operation *inputOperation,
-                                                PatternRewriter &rewriter) {
-  llvm::SmallVector<Value, 4> inputOperands(inputOperation->getOperands());
-  auto combinedShapeOp = buildCastInputsToResultShape(
-      inputOperation->getLoc(), getShapeOp.getRankedShape(), inputOperands,
-      rewriter);
-  if (!combinedShapeOp) return failure();
-  rewriter.replaceOp(getShapeOp, {combinedShapeOp});
-  return success();
-}
-
-// Matches the case where the input to a GetRankedShapeOp is another
-// operation. This is the primary supported case as other rewrites should
-// have isolated function/block boundaries with TieShape ops.
-LogicalResult rewriteInputOp(GetRankedShapeOp getShapeOp,
-                             GetRankedShapeOp::Adaptor operands,
-                             Operation *inputOperation,
-                             PatternRewriter &rewriter) {
-  // SameOperandsAndResultShape trait.
-  if (inputOperation->hasTrait<OpTrait::SameOperandsAndResultShape>() ||
-      inputOperation->hasTrait<OpTrait::SameOperandsAndResultType>()) {
-    return rewriteSameOperandsAndResultShape(getShapeOp, inputOperation,
-                                             rewriter);
-  }
-
-  // Custom shapes.
-  auto customOpShapeBuilder = getCustomOpShapeBuilder();
-  if (customOpShapeBuilder) {
-    auto resultShape = getShapeOp.getRankedShape();
-    for (auto &shapeBuilder : *customOpShapeBuilder) {
-      Value customShape =
-          shapeBuilder->buildRankedShape(resultShape, inputOperation, rewriter);
-      if (customShape) {
-        rewriter.replaceOp(getShapeOp, customShape);
-        return success();
-      }
-    }
-  }
-
-  return failure();
-}
-
-void rewriteRuntimeShape(GetRankedShapeOp getShapeOp,
-                         GetRankedShapeOp::Adaptor operands,
-                         PatternRewriter &rewriter) {
-  auto shapeType = getShapeOp.shape().getType().dyn_cast<RankedShapeType>();
-  SmallVector<Value, 4> dynamicDims;
-  for (int64_t i = 0, e = shapeType.getRank(); i < e; ++i) {
-    if (!shapeType.isDimDynamic(i)) continue;
-    dynamicDims.push_back(rewriter.create<memref::DimOp>(
-        getShapeOp.getLoc(), operands.operand(), i));
-  }
-
-  // TODO(laurenzo): Remove once further along (it is fine to be unsupported
-  // as it will fall back to generic), but in these early phases, it is
-  // extremely useful to be chatty about this fallback.
-  auto inputOperation = operands.operand().getDefiningOp();
-  if (inputOperation) {
-    inputOperation->emitRemark()
-        << "unable to materialize shape calculation (unsupported op '"
-        << inputOperation->getName() << "'?): falling back to runtime "
-        << "resolution";
-  }
-
-  rewriter.replaceOpWithNewOp<MakeRankedShapeOp>(getShapeOp, shapeType,
-                                                 dynamicDims);
-}
-
-// Low benefit fallback pattern to materialize a ranked shape.
-LogicalResult materializeRankedShapePattern(GetRankedShapeOp getShapeOp,
-                                            GetRankedShapeOp::Adaptor operands,
-                                            PatternRewriter &rewriter) {
-  // Check for static shape and elide.
-  auto operandType = operands.operand().getType().dyn_cast<RankedTensorType>();
-  auto shapeType = getShapeOp.shape().getType().dyn_cast<RankedShapeType>();
-  if (operandType && shapeType && operandType.hasStaticShape()) {
-    rewriter.replaceOpWithNewOp<ConstRankedShapeOp>(getShapeOp, shapeType);
-    return success();
-  }
-
-  // Check for input operation (unless if a small set of shape ops).
-  if (auto inputOperation = operands.operand().getDefiningOp()) {
-    // Materialize a shape function if possible.
-    LLVM_DEBUG(llvm::dbgs() << "** SHAPE: MATERIALIZE FOR INPUT OP: "
-                            << *inputOperation << "\n");
-    if (!failed(
-            rewriteInputOp(getShapeOp, operands, inputOperation, rewriter))) {
-      return success();
-    }
-  }
-
-  // Runtime fallback.
-  LLVM_DEBUG(llvm::dbgs() << "** SHAPE: RUNTIME RESOLUTION\n");
-  rewriteRuntimeShape(getShapeOp, operands, rewriter);
-  return success();
-}
-
-// Matches a tie_shape -> get_ranked_shape pattern and resolves it statically.
-// This must be a higher benefit than materializeRankedShapePattern.
-LogicalResult passThroughTiedGetRankedShapePattern(
-    GetRankedShapeOp getShapeOp, GetRankedShapeOp::Adaptor operands,
-    PatternRewriter &rewriter) {
-  // Check for input operation (unless if a small set of shape ops).
-  Operation *inputOperation = operands.operand().getDefiningOp();
-  if (auto tieOp = llvm::dyn_cast_or_null<TieShapeOp>(inputOperation)) {
-    LLVM_DEBUG(llvm::dbgs() << "** SHAPE: PASS-THROUGH tie_shape\n");
-    rewriter.replaceOp(getShapeOp, tieOp.shape());
-    return success();
-  }
-  return failure();
-}
-
-}  // namespace
-
-void setupMaterializeShapeCalculationsLegality(ConversionTarget &target) {
-  // We explicitly want to convert these ops, eliminating them.
-  target.addIllegalOp<GetRankedShapeOp>();
-  target.addIllegalOp<RankedBroadcastShapeOp>();
-  target.addIllegalOp<GatherExtentsOp>();
-}
-
-void populateMaterializeShapeCalculationsConversionPatterns(
-    OwningRewritePatternList &patterns, MLIRContext *context) {
-  // Fallback patterns.
-  insertConversionPattern(patterns, context, expandRankedBroadcastShapePattern,
-                          /*benefit=*/1);
-  insertConversionPattern(patterns, context, expandGatherExtentsOp,
-                          /*benefit=*/1);
-  insertConversionPattern(patterns, context, materializeRankedShapePattern,
-                          /*benefit=*/1);
-
-  // High benefit patterns.
-  insertConversionPattern(patterns, context,
-                          passThroughTiedGetRankedShapePattern,
-                          /*benefit=*/10);
-}
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Transforms/MaterializeShapeCalculationsPass.cpp b/iree/compiler/Dialect/Shape/Transforms/MaterializeShapeCalculationsPass.cpp
deleted file mode 100644
index db4d414..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/MaterializeShapeCalculationsPass.cpp
+++ /dev/null
@@ -1,90 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include "iree/compiler/Dialect/Shape/IR/Builders.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeDialect.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeInterface.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeTypes.h"
-#include "iree/compiler/Dialect/Shape/Plugins/XLA/XlaHloShapeBuilder.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Patterns.h"
-#include "iree/compiler/Utils/PatternUtils.h"
-#include "mlir/Dialect/MemRef/IR/MemRef.h"
-#include "mlir/Dialect/StandardOps/IR/Ops.h"
-#include "mlir/IR/BuiltinTypes.h"
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/OpDefinition.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Pass/PassRegistry.h"
-#include "mlir/Support/LogicalResult.h"
-#include "mlir/Transforms/GreedyPatternRewriteDriver.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-namespace {
-
-class MaterializeShapeCalculationsPass
-    : public PassWrapper<MaterializeShapeCalculationsPass, FunctionPass> {
- public:
-  void getDependentDialects(DialectRegistry &registry) const override {
-    registry.insert<memref::MemRefDialect>();
-  }
-
-  void runOnFunction() override {
-    auto *context = &getContext();
-    // First run conversion.
-    ConversionTarget target(*context);
-
-    // Allow any Shape dialect ops to persist.
-    target.addLegalDialect<ShapeDialect>();
-
-    // Shape functions and runtime resolution generate ops in the memref and
-    // standard dialect.
-    target.addLegalDialect<memref::MemRefDialect>();
-    target.addLegalDialect<StandardOpsDialect>();
-
-    setupMaterializeShapeCalculationsLegality(target);
-    OwningRewritePatternList conversionPatterns(&getContext());
-    populateMaterializeShapeCalculationsConversionPatterns(conversionPatterns,
-                                                           context);
-    if (failed(applyPartialConversion(getOperation(), target,
-                                      std::move(conversionPatterns)))) {
-      signalPassFailure();
-      return;
-    }
-
-    // And then canonicalize shape ops.
-    // TODO(laurenzo): I would prefer to get the list of ops in the dialect
-    // versus doing this, but I don't know that is possible.
-    OwningRewritePatternList patterns(&getContext());
-    CastCompatibleShapeOp::getCanonicalizationPatterns(patterns, context);
-    GetRankedShapeOp::getCanonicalizationPatterns(patterns, context);
-    MakeRankedShapeOp::getCanonicalizationPatterns(patterns, context);
-    RankedDimOp::getCanonicalizationPatterns(patterns, context);
-    RankedDimsOp::getCanonicalizationPatterns(patterns, context);
-    TieShapeOp::getCanonicalizationPatterns(patterns, context);
-    FromExtentTensorOp::getCanonicalizationPatterns(patterns, context);
-    (void)applyPatternsAndFoldGreedily(getOperation(), std::move(patterns));
-  }
-};
-
-}  // namespace
-
-// For any function which contains dynamic dims in its inputs or results,
-// rewrites it so that the dynamic dims are passed in/out.
-std::unique_ptr<OperationPass<FuncOp>>
-createMaterializeShapeCalculationsPass() {
-  return std::make_unique<Shape::MaterializeShapeCalculationsPass>();
-}
-
-static PassRegistration<Shape::MaterializeShapeCalculationsPass> pass(
-    "iree-shape-materialize-calculations", "Materializes shape calculations.");
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Transforms/Passes.h b/iree/compiler/Dialect/Shape/Transforms/Passes.h
index 06a6fa6..71bbe97 100644
--- a/iree/compiler/Dialect/Shape/Transforms/Passes.h
+++ b/iree/compiler/Dialect/Shape/Transforms/Passes.h
@@ -27,38 +27,15 @@
 std::unique_ptr<OperationPass<FuncOp>>
 createExpandFunctionRankedShapeDimsPass();
 
-// For any dynamically shaped edges in a function, introduces an appropriate
-// get_ranked_shape and corresponding tie_shape op to make the association.
-//
-// Any op contained in a region transitively owned by an op with a name in
-// `doNotRecurseOpNames` is not tied.
-std::unique_ptr<OperationPass<FuncOp>> createTieDynamicShapesPass(
-    ArrayRef<std::string> doNotRecurseOpNames = {});
-
-// Materializes shape calculations for any get_ranked_shape ops.
-std::unique_ptr<OperationPass<FuncOp>> createMaterializeShapeCalculationsPass();
-
 // Cleans up any unnecessary shape placeholder ops. Can be run after all
 // shape calculation code has been lowered.
 std::unique_ptr<OperationPass<FuncOp>> createCleanupShapePlaceholdersPass();
 
-// Converts shape-sensitive HLOs to be based on facilities in the shape
-// dialect.
-std::unique_ptr<OperationPass<FuncOp>> createConvertHLOToShapePass();
-
-// Best-effort hoisting of shape calculations to attempt to establish the
-// invariant that shapex.tie_shape second operand dominates the first operand.
-std::unique_ptr<OperationPass<FuncOp>> createHoistShapeCalculationsPass();
-
 // Register all Passes
 inline void registerShapePasses() {
   createExpandFunctionDynamicDimsPass();
   createExpandFunctionRankedShapeDimsPass();
-  createTieDynamicShapesPass();
-  createMaterializeShapeCalculationsPass();
   createCleanupShapePlaceholdersPass();
-  createConvertHLOToShapePass();
-  createHoistShapeCalculationsPass();
 }
 
 }  // namespace Shape
diff --git a/iree/compiler/Dialect/Shape/Transforms/Patterns.h b/iree/compiler/Dialect/Shape/Transforms/Patterns.h
deleted file mode 100644
index aa3039d..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/Patterns.h
+++ /dev/null
@@ -1,37 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#ifndef IREE_COMPILER_DIALECT_SHAPE_TRANSFORMS_PATTERNS_H_
-#define IREE_COMPILER_DIALECT_SHAPE_TRANSFORMS_PATTERNS_H_
-
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Transforms/DialectConversion.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-
-// Sets up legality for shape calculation materialization conversions.
-void setupMaterializeShapeCalculationsLegality(ConversionTarget &target);
-
-// Populates patterns that will materialize shape calculations for any
-// GetRankedShape and related ops.
-void populateMaterializeShapeCalculationsConversionPatterns(
-    OwningRewritePatternList &patterns, MLIRContext *context);
-
-// Sets up legality for shape calculation materialization conversions.
-void setupShapeToStandardLegality(ConversionTarget &target);
-
-// Populates patterns that will convert shape calculations into standard ops.
-void populateShapeToStandardConversionPatterns(
-    OwningRewritePatternList &patterns, MLIRContext *context);
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
-
-#endif  // IREE_COMPILER_DIALECT_SHAPE_TRANSFORMS_PATTERNS_H_
diff --git a/iree/compiler/Dialect/Shape/Transforms/TieDynamicShapesPass.cpp b/iree/compiler/Dialect/Shape/Transforms/TieDynamicShapesPass.cpp
deleted file mode 100644
index 4a8c217..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/TieDynamicShapesPass.cpp
+++ /dev/null
@@ -1,109 +0,0 @@
-// Copyright 2020 The IREE Authors
-//
-// Licensed under the Apache License v2.0 with LLVM Exceptions.
-// See https://llvm.org/LICENSE.txt for license information.
-// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
-
-#include "iree/compiler/Dialect/Shape/IR/ShapeDialect.h"
-#include "iree/compiler/Dialect/Shape/IR/ShapeOps.h"
-#include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
-#include "mlir/IR/Builders.h"
-#include "mlir/IR/BuiltinTypes.h"
-#include "mlir/IR/MLIRContext.h"
-#include "mlir/IR/PatternMatch.h"
-#include "mlir/Pass/PassRegistry.h"
-
-namespace mlir {
-namespace iree_compiler {
-namespace Shape {
-
-// Return true if `op` has a proper ancestor whose op name is in `opNames`.
-static bool hasProperAncestorWithOpName(Operation *op,
-                                        const DenseSet<StringRef> &opNames) {
-  while ((op = op->getParentOp())) {
-    if (opNames.contains(op->getName().getStringRef())) {
-      return true;
-    }
-  }
-  return false;
-}
-
-namespace {
-
-class TieDynamicShapesPass
-    : public PassWrapper<TieDynamicShapesPass, FunctionPass> {
- public:
-  TieDynamicShapesPass() = default;
-  TieDynamicShapesPass(const TieDynamicShapesPass &) {}
-  explicit TieDynamicShapesPass(ArrayRef<std::string> doNotRecurseOpNames_) {
-    doNotRecurseOpNames = doNotRecurseOpNames_;
-  }
-  void getDependentDialects(DialectRegistry &registry) const override {
-    registry.insert<ShapeDialect>();
-  }
-
-  void runOnFunction() override {
-    DenseSet<StringRef> doNotRecurseOpNameSet;
-    for (auto &s : doNotRecurseOpNames) {
-      doNotRecurseOpNameSet.insert(s);
-    }
-    getFunction().walk([&](Operation *nestedOp) {
-      if (hasProperAncestorWithOpName(nestedOp, doNotRecurseOpNameSet)) {
-        return;
-      }
-      for (auto result : nestedOp->getResults()) {
-        rewriteOperationResult(nestedOp, result);
-      }
-    });
-  }
-
-  void rewriteOperationResult(Operation *op, Value result) {
-    if (llvm::isa<TieShapeOp>(op)) return;
-    auto shapedType = result.getType().dyn_cast<ShapedType>();
-    if (!shapedType || shapedType.hasStaticShape()) return;
-
-    // Only ranked is supported currently.
-    if (!shapedType.hasRank()) return;
-
-    // Skip un-ambiguous ties that already exist.
-    if (result.hasOneUse() &&
-        llvm::dyn_cast_or_null<TieShapeOp>(result.use_begin()->getOwner())) {
-      return;
-    }
-
-    OpBuilder builder(&getContext());
-    builder.setInsertionPointAfter(op);
-    auto getShapeOp = builder.create<GetRankedShapeOp>(op->getLoc(), result);
-    auto tieOp = builder.create<TieShapeOp>(op->getLoc(), result.getType(),
-                                            result, getShapeOp);
-
-    // Replace: {result} -> {tieOp, getShapeOp, ...origUses}
-    // With: {result} -> {tieOp -> ...origUses, getShapeOp}
-    result.replaceAllUsesWith(tieOp);
-    tieOp.getOperation()->replaceUsesOfWith(tieOp, result);
-    getShapeOp.getOperation()->replaceUsesOfWith(tieOp, result);
-  }
-
-  ListOption<std::string> doNotRecurseOpNames{
-      *this, "do-not-recurse-op-names",
-      llvm::cl::desc("comma separated list of op names (e.g. "
-                     "`my_dialect.my_op,my_dialect.my_other_op`) whose regions "
-                     "should not have their ops tied"),
-      llvm::cl::ZeroOrMore, llvm::cl::MiscFlags::CommaSeparated};
-};
-
-}  // namespace
-
-// For any function which contains dynamic dims in its inputs or results,
-// rewrites it so that the dynamic dims are passed in/out.
-std::unique_ptr<OperationPass<FuncOp>> createTieDynamicShapesPass(
-    ArrayRef<std::string> doNotRecurseOpNames) {
-  return std::make_unique<Shape::TieDynamicShapesPass>(doNotRecurseOpNames);
-}
-
-static PassRegistration<Shape::TieDynamicShapesPass> pass(
-    "iree-shape-tie-dynamic", "Ties any dynamic shapes in a function.");
-
-}  // namespace Shape
-}  // namespace iree_compiler
-}  // namespace mlir
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/BUILD b/iree/compiler/Dialect/Shape/Transforms/test/BUILD
index bff70b7..84fca24 100644
--- a/iree/compiler/Dialect/Shape/Transforms/test/BUILD
+++ b/iree/compiler/Dialect/Shape/Transforms/test/BUILD
@@ -17,13 +17,8 @@
     name = "lit",
     srcs = enforce_glob(
         [
-            "convert_hlo_to_shape_dialect.mlir",
             "expand_function_dynamic_dims.mlir",
             "expand_function_ranked_shape_dims.mlir",
-            "hoist_shape_calculations.mlir",
-            "materialize_shape_calculations.mlir",
-            "tie_dynamic_shapes.mlir",
-            "tie_dynamic_shapes_no_recurse.mlir",
         ],
         include = ["*.mlir"],
     ),
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/CMakeLists.txt b/iree/compiler/Dialect/Shape/Transforms/test/CMakeLists.txt
index 6c0229e..8e40e7d 100644
--- a/iree/compiler/Dialect/Shape/Transforms/test/CMakeLists.txt
+++ b/iree/compiler/Dialect/Shape/Transforms/test/CMakeLists.txt
@@ -14,13 +14,8 @@
   NAME
     lit
   SRCS
-    "convert_hlo_to_shape_dialect.mlir"
     "expand_function_dynamic_dims.mlir"
     "expand_function_ranked_shape_dims.mlir"
-    "hoist_shape_calculations.mlir"
-    "materialize_shape_calculations.mlir"
-    "tie_dynamic_shapes.mlir"
-    "tie_dynamic_shapes_no_recurse.mlir"
   DATA
     iree::tools::IreeFileCheck
     iree::tools::iree-opt
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/convert_hlo_to_shape_dialect.mlir b/iree/compiler/Dialect/Shape/Transforms/test/convert_hlo_to_shape_dialect.mlir
deleted file mode 100644
index 1ddddcb..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/test/convert_hlo_to_shape_dialect.mlir
+++ /dev/null
@@ -1,10 +0,0 @@
-// RUN: iree-opt -split-input-file -verify-diagnostics -iree-shape-convert-hlo %s | IreeFileCheck %s
-
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: tensor<2xindex>) -> tensor<?x?xf32> {
-  // CHECK-DAG: %[[SHAPE:.+]] = "shapex.from_extent_tensor"(%arg1) : (tensor<2xindex>) -> !shapex.ranked_shape<[?,?]>
-  // CHECK-DAG: %[[BROADCASTED:.+]] = "shapex.ranked_broadcast_in_dim"(%arg0, %0) {broadcast_dimensions = dense<1> : tensor<1xi64>}
-  // CHECK-DAG: return %[[BROADCASTED]]
-  %0 = "mhlo.dynamic_broadcast_in_dim"(%arg0, %arg1) {broadcast_dimensions = dense<[1]> : tensor<1xi64>}: (tensor<?xf32>, tensor<2xindex>) -> tensor<?x?xf32>
-  return %0 : tensor<?x?xf32>
-}
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/hoist_shape_calculations.mlir b/iree/compiler/Dialect/Shape/Transforms/test/hoist_shape_calculations.mlir
deleted file mode 100644
index 1571fd5..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/test/hoist_shape_calculations.mlir
+++ /dev/null
@@ -1,70 +0,0 @@
-// RUN: iree-opt -allow-unregistered-dialect -split-input-file -verify-diagnostics -iree-shape-hoist-shape-calculations %s | IreeFileCheck %s
-
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: index) {
-  // CHECK: shapex.make_ranked_shape
-  // CHECK: addf
-  %t = addf %arg0, %arg0 : tensor<?xf32>
-  %shape = shapex.make_ranked_shape %arg1 : (index) -> !shapex.ranked_shape<[?]>
-  shapex.tie_shape %t, %shape : tensor<?xf32>, !shapex.ranked_shape<[?]>
-  return
-}
-
-// -----
-
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>) {
-  // CHECK: addf
-  // CHECK: memref.dim
-  // CHECK: shapex.make_ranked_shape
-  %t = addf %arg0, %arg0 : tensor<?xf32>
-  %c0 = constant 0 : index
-  %dim = memref.dim %t, %c0 : tensor<?xf32>
-  %shape = shapex.make_ranked_shape %dim : (index) -> !shapex.ranked_shape<[?]>
-  shapex.tie_shape %t, %shape : tensor<?xf32>, !shapex.ranked_shape<[?]>
-  return
-}
-
-// -----
-
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: index) {
-  // CHECK: addi
-  // CHECK: muli
-  // CHECK: shapex.make_ranked_shape
-  // CHECK: some_dialect.some_op
-  "some_dialect.some_op"() : () -> ()
-  %addi = addi %arg1, %arg1 : index
-  %dim = muli %addi, %addi : index
-  %shape = shapex.make_ranked_shape %dim : (index) -> !shapex.ranked_shape<[?]>
-  shapex.tie_shape %arg0, %shape : tensor<?xf32>, !shapex.ranked_shape<[?]>
-  return
-}
-
-// -----
-
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?xf32>, %arg1: index) {
-  // CHECK: some_dialect.some_op
-  // CHECK: some_dialect.side_effecting_muli
-  // CHECK: shapex.make_ranked_shape
-  "some_dialect.some_op"() : () -> ()
-  %dim = "some_dialect.side_effecting_muli"(%arg1, %arg1) : (index, index) -> index
-  %shape = shapex.make_ranked_shape %dim : (index) -> !shapex.ranked_shape<[?]>
-  shapex.tie_shape %arg0, %shape : tensor<?xf32>, !shapex.ranked_shape<[?]>
-  return
-}
-
-// -----
-
-// CHECK-LABEL: func @f
-func @f(%arg0: tensor<?x?xf32>) {
-  // CHECK: constant
-  // CHECK: constant
-  // CHECK: shapex.make_ranked_shape
-  %c0 = constant 0 : index
-  %c1 = constant 1 : index
-  %0 = shapex.make_ranked_shape %c0, %c1 : (index, index) -> !shapex.ranked_shape<[?,?]>
-  %1 = shapex.tie_shape %arg0, %0 : tensor<?x?xf32>, !shapex.ranked_shape<[?,?]>
-  return
-}
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/materialize_shape_calculations.mlir b/iree/compiler/Dialect/Shape/Transforms/test/materialize_shape_calculations.mlir
deleted file mode 100644
index 58966ec..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/test/materialize_shape_calculations.mlir
+++ /dev/null
@@ -1,88 +0,0 @@
-// RUN: iree-opt -allow-unregistered-dialect -split-input-file -verify-diagnostics -iree-shape-materialize-calculations -iree-shape-cleanup-placeholders %s | IreeFileCheck %s
-
-// CHECK-LABEL: @compileTime
-// CHECK-SAME: %[[T:[^:[:space:]]+]]: tensor<?x2xf32>
-// CHECK-SAME: %[[SHAPE:[^:[:space:]]+]]: !shapex.ranked_shape<[?,2]>
-func @compileTime(%arg0 : tensor<?x2xf32>, %arg1 : !shapex.ranked_shape<[?,2]>) -> (tensor<?x2xf32> , !shapex.ranked_shape<[?,2]>) {
-  // CHECK-NOT: shapex.tie_shape
-  // CHECK-NOT: shapex.get_ranked_shape
-  %0 = shapex.tie_shape %arg0, %arg1 : tensor<?x2xf32>, !shapex.ranked_shape<[?,2]>
-  // CHECK: %[[ABS:.+]] = "mhlo.abs"(%[[T]])
-  // The only thing special about abs is that we have a compile time shape
-  // calculation for it.
-  %1 = "mhlo.abs"(%0) : (tensor<?x2xf32>) -> (tensor<?x2xf32>)
-  %2 = shapex.get_ranked_shape %1 : tensor<?x2xf32> -> !shapex.ranked_shape<[?,2]>
-  // CHECK: return %[[ABS]], %[[SHAPE]]
-  return %1, %2 : tensor<?x2xf32>, !shapex.ranked_shape<[?,2]>
-}
-
-// -----
-
-// CHECK-LABEL: @f
-func @f(%arg0 : !shapex.ranked_shape<[?]>, %arg1 : !shapex.ranked_shape<[?]>) -> (!shapex.ranked_shape<[?]>) {
-  // CHECK-DAG: %[[LHSEXTENT:.+]] = shapex.ranked_dim %arg0[0]
-  // CHECK-DAG: %[[RHSEXTENT:.+]] = shapex.ranked_dim %arg1[0]
-  // CHECK-DAG: %[[GT:.+]] = cmpi ugt, %[[LHSEXTENT]], %[[RHSEXTENT]] : index
-  // CHECK-DAG: %[[MAX:.+]] = select %[[GT]], %[[LHSEXTENT]], %[[RHSEXTENT]] : index
-  // CHECK-DAG: %[[RS:.+]] = shapex.make_ranked_shape %[[MAX]]
-  // CHECK-DAG: return %[[RS]]
-  %0 = "shapex.ranked_broadcast_shape"(%arg0, %arg1) {
-    lhs_broadcast_dimensions = dense<[0]> : tensor<1xi64>,
-    rhs_broadcast_dimensions = dense<[0]> : tensor<1xi64>
-  } : (!shapex.ranked_shape<[?]>, !shapex.ranked_shape<[?]>) -> !shapex.ranked_shape<[?]>
-  return %0 : !shapex.ranked_shape<[?]>
-}
-
-// -----
-// CHECK-LABEL: @runTimeFallback
-// CHECK-SAME: %[[T:[^:[:space:]]+]]: tensor<?x2xf32>
-// CHECK-SAME: %[[SHAPE:[^:[:space:]]+]]: !shapex.ranked_shape<[?,2]>
-func @runTimeFallback(%arg0 : tensor<?x2xf32>, %arg1 : !shapex.ranked_shape<[?,2]>) -> (tensor<?x2xf32> , !shapex.ranked_shape<[?,2]>) {
-  // CHECK-NOT: shapex.tie_shape
-  // CHECK-NOT: shapex.get_ranked_shape
-  %0 = shapex.tie_shape %arg0, %arg1 : tensor<?x2xf32>, !shapex.ranked_shape<[?,2]>
-  // CHECK: %[[C0:.*]] = constant 0 : index
-  // CHECK: %[[RESULT:.+]] = "unknown_op"
-  // @expected-remark @+1 {{unable to materialize shape calculation}}
-  %1 = "unknown_op"(%0) : (tensor<?x2xf32>) -> (tensor<?x2xf32>)
-  // CHECK: %[[DIM:.+]] = memref.dim %[[RESULT]], %[[C0]]
-  // CHECK: %[[RESULT_SHAPE:.+]] = shapex.make_ranked_shape %[[DIM]]
-  // CHECK: return %[[RESULT]], %[[RESULT_SHAPE]]
-  %2 = shapex.get_ranked_shape %1 : tensor<?x2xf32> -> !shapex.ranked_shape<[?,2]>
-  return %1, %2 : tensor<?x2xf32>, !shapex.ranked_shape<[?,2]>
-}
-
-// -----
-// CHECK-LABEL: func @f
-func @f(%arg0: index, %arg1: index) -> (index, index, index) {
-  %0 = shapex.make_ranked_shape %arg0, %arg1 : (index, index) -> !shapex.ranked_shape<[?,?]>
-  %1 = "shapex.gather_extents"(%0) {indices = dense<[1, 1, 0]> : tensor<3xi64>} : (!shapex.ranked_shape<[?,?]>) -> !shapex.ranked_shape<[?,?,?]>
-  %2:3 = shapex.ranked_dims %1 : !shapex.ranked_shape<[?,?,?]> -> index, index, index
-  // CHECK: return %arg1, %arg1, %arg0
-  return %2#0, %2#1, %2#2 : index, index, index
-}
-
-// -----
-// CHECK-LABEL: func @f
-func @f(%arg0: index, %arg1: index, %arg2: index) -> (index, index, index, index) {
-  %0 = shapex.make_ranked_shape %arg0 : (index) -> !shapex.ranked_shape<[?]>
-  %1 = shapex.make_ranked_shape %arg1 : (index) -> !shapex.ranked_shape<[?]>
-  %2 = shapex.make_ranked_shape %arg2 : (index) -> !shapex.ranked_shape<[?]>
-  %gathered = "shapex.gather_extents"(%0, %1, %2) {indices = dense<[2, 2, 1, 0]> : tensor<4xi64>} : (!shapex.ranked_shape<[?]>, !shapex.ranked_shape<[?]>, !shapex.ranked_shape<[?]>) -> !shapex.ranked_shape<[?,?,?,?]>
-  %extents:4 = shapex.ranked_dims %gathered : !shapex.ranked_shape<[?,?,?,?]> -> index, index, index, index
-  // CHECK: return %arg2, %arg2, %arg1, %arg0
-  return %extents#0, %extents#1, %extents#2, %extents#3 : index, index, index, index
-}
-
-// -----
-// CHECK-LABEL: func @f
-func @f(%arg0: index, %arg1: index, %arg2: index) -> (index, index, index, index, index, index) {
-  %0 = shapex.make_ranked_shape %arg0 : (index) -> !shapex.ranked_shape<[?,2]>
-  %1 = shapex.make_ranked_shape %arg1 : (index) -> !shapex.ranked_shape<[3,?]>
-  %2 = shapex.make_ranked_shape %arg2 : (index) -> !shapex.ranked_shape<[7,?]>
-  %gathered = "shapex.gather_extents"(%0, %1, %2) {indices = dense<[0, 1, 2, 3, 4, 5]> : tensor<6xi64>} : (!shapex.ranked_shape<[?,2]>, !shapex.ranked_shape<[3,?]>, !shapex.ranked_shape<[7,?]>) -> !shapex.ranked_shape<[?,2,3,?,7,?]>
-  %extents:6 = shapex.ranked_dims %gathered : !shapex.ranked_shape<[?,2,3,?,7,?]> -> index, index, index, index, index, index
-  // CHECK: return %arg0, %c2{{.*}}, %c3{{.*}}, %arg1, %c7{{.*}}, %arg2
-  return %extents#0, %extents#1, %extents#2, %extents#3, %extents#4, %extents#5 : index, index, index, index, index, index
-}
-
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/tie_dynamic_shapes.mlir b/iree/compiler/Dialect/Shape/Transforms/test/tie_dynamic_shapes.mlir
deleted file mode 100644
index 1e557ab..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/test/tie_dynamic_shapes.mlir
+++ /dev/null
@@ -1,50 +0,0 @@
-// RUN: iree-opt -split-input-file -verify-diagnostics -iree-shape-tie-dynamic %s | IreeFileCheck %s
-
-// CHECK-LABEL: @tieIntermediates
-// CHECK-SAME: %[[T1:[^:[:space:]]+]]: tensor
-// CHECK-SAME: %[[T2:[^:[:space:]]+]]: tensor
-func @tieIntermediates(%arg1 : tensor<?xf32>, %arg2 : tensor<?xf32>) -> tensor<?xf32> {
-  // CHECK-DAG: %[[SUM:.+]] = addf %[[T1]], %[[T2]]
-  // CHECK-DAG: %[[RS:.+]] = shapex.get_ranked_shape %[[SUM]]
-  // CHECK-DAG: %[[TIED_SUM:.+]] = shapex.tie_shape %[[SUM]], %[[RS]]
-  // CHECK-DAG: return %[[TIED_SUM]]
-  %0 = addf %arg1, %arg2 : tensor<?xf32>
-  return %0 : tensor<?xf32>
-}
-
-// -----
-// CHECK-LABEL: @noTieStatic
-// CHECK-SAME: %[[T1:[^:[:space:]]+]]: tensor
-// CHECK-SAME: %[[T2:[^:[:space:]]+]]: tensor
-func @noTieStatic(%arg1 : tensor<1xf32>, %arg2 : tensor<1xf32>) -> tensor<1xf32> {
-  // CHECK-NOT: shapex.tie_shape
-  // CHECK: %[[SUM:.+]] = addf %[[T1]], %[[T2]]
-  // CHECK: return %[[SUM]]
-  %0 = addf %arg1, %arg2 : tensor<1xf32>
-  return %0 : tensor<1xf32>
-}
-
-// -----
-// CHECK-LABEL: @noTieUnranked
-// CHECK-SAME: %[[T1:[^:[:space:]]+]]: tensor
-// CHECK-SAME: %[[T2:[^:[:space:]]+]]: tensor
-func @noTieUnranked(%arg1 : tensor<*xf32>, %arg2 : tensor<*xf32>) -> tensor<*xf32> {
-  // CHECK-NOT: shapex.tie_shape
-  // CHECK: %[[SUM:.+]] = addf %[[T1]], %[[T2]]
-  // CHECK: return %[[SUM]]
-  %0 = addf %arg1, %arg2 : tensor<*xf32>
-  return %0 : tensor<*xf32>
-}
-
-// CHECK-LABEL: @noTieRedundant
-// CHECK-SAME: %[[T1:[^:[:space:]]+]]: tensor
-// CHECK-SAME: %[[T2:[^:[:space:]]+]]: tensor
-// CHECK-SAME: %[[RS1:[^:[:space:]]+]]: !shapex.ranked_shape
-func @noTieRedundant(%arg1 : tensor<?xf32>, %arg2 : tensor<?xf32>, %arg3 : !shapex.ranked_shape<[?]>) -> tensor<?xf32> {
-  // CHECK-DAG: %[[SUM:.+]] = addf %[[T1]], %[[T2]]
-  // CHECK-DAG: %[[TIED_SUM:.+]] = shapex.tie_shape %[[SUM]], %[[RS1]]
-  // CHECK-DAG: return %[[TIED_SUM]]
-  %0 = addf %arg1, %arg2 : tensor<?xf32>
-  %1 = shapex.tie_shape %0, %arg3 : tensor<?xf32>, !shapex.ranked_shape<[?]>
-  return %1 : tensor<?xf32>
-}
diff --git a/iree/compiler/Dialect/Shape/Transforms/test/tie_dynamic_shapes_no_recurse.mlir b/iree/compiler/Dialect/Shape/Transforms/test/tie_dynamic_shapes_no_recurse.mlir
deleted file mode 100644
index a866bf8..0000000
--- a/iree/compiler/Dialect/Shape/Transforms/test/tie_dynamic_shapes_no_recurse.mlir
+++ /dev/null
@@ -1,37 +0,0 @@
-// RUN: iree-opt -allow-unregistered-dialect -split-input-file -verify-diagnostics -iree-shape-tie-dynamic='do-not-recurse-op-names=my_dialect.region1,my_dialect.region2' %s | IreeFileCheck %s
-
-// CHECK-LABEL: @doTie
-func @doTie(%arg1 : tensor<?xf32>) {
-  // CHECK: shapex.tie_shape
-  %0 = "my_dialect.munge"(%arg1) : (tensor<?xf32>) -> tensor<?xf32>
-  return
-}
-
-// CHECK-LABEL: @noTieDoNotRecurse
-func @noTieDoNotRecurse(%arg1 : tensor<?xf32>) {
-  // CHECK: my_dialect.region1
-  "my_dialect.region1"() ( {
-    // CHECK: my_dialect.munge
-    %0 = "my_dialect.munge"(%arg1) : (tensor<?xf32>) -> tensor<?xf32>
-    // CHECK-NOT: shapex.tie_shape
-    // CHECK: my_dialect.terminator
-    "my_dialect.terminator"() : () -> ()
-  }) : () -> ()
-  // CHECK: my_dialect.region2
-  "my_dialect.region2"() ( {
-    // CHECK: my_dialect.munge
-    %1 = "my_dialect.munge"(%arg1) : (tensor<?xf32>) -> tensor<?xf32>
-    // CHECK-NOT: shapex.tie_shape
-    // CHECK: my_dialect.terminator
-    "my_dialect.terminator"() : () -> ()
-  }) : () -> ()
-  // CHECK: my_dialect.region3
-  "my_dialect.region3"() ( {
-    // CHECK: my_dialect.munge
-    %1 = "my_dialect.munge"(%arg1) : (tensor<?xf32>) -> tensor<?xf32>
-    // CHECK: shapex.tie_shape
-    // CHECK: my_dialect.terminator
-    "my_dialect.terminator"() : () -> ()
-  }) : () -> ()
-  return
-}
diff --git a/iree/hal/cuda/graph_command_buffer.c b/iree/hal/cuda/graph_command_buffer.c
index 6c49ef6..9b7fdc9 100644
--- a/iree/hal/cuda/graph_command_buffer.c
+++ b/iree/hal/cuda/graph_command_buffer.c
@@ -35,7 +35,7 @@
   void* current_descriptor[];
 } iree_hal_cuda_graph_command_buffer_t;
 
-static const size_t max_binding_count = 64;
+#define IREE_HAL_CUDA_MAX_BINDING_COUNT 64
 
 extern const iree_hal_command_buffer_vtable_t
     iree_hal_cuda_graph_command_buffer_vtable;
@@ -61,8 +61,8 @@
                        "cuGraphCreate");
   iree_hal_cuda_graph_command_buffer_t* command_buffer = NULL;
   size_t total_size = sizeof(*command_buffer) +
-                      max_binding_count * sizeof(void*) +
-                      max_binding_count * sizeof(CUdeviceptr);
+                      IREE_HAL_CUDA_MAX_BINDING_COUNT * sizeof(void*) +
+                      IREE_HAL_CUDA_MAX_BINDING_COUNT * sizeof(CUdeviceptr);
   iree_status_t status = iree_allocator_malloc(
       context->host_allocator, total_size, (void**)&command_buffer);
   if (iree_status_is_ok(status)) {
@@ -77,8 +77,9 @@
     command_buffer->last_node = NULL;
 
     CUdeviceptr* device_ptrs =
-        (CUdeviceptr*)(command_buffer->current_descriptor + max_binding_count);
-    for (size_t i = 0; i < max_binding_count; i++) {
+        (CUdeviceptr*)(command_buffer->current_descriptor +
+                       IREE_HAL_CUDA_MAX_BINDING_COUNT);
+    for (size_t i = 0; i < IREE_HAL_CUDA_MAX_BINDING_COUNT; i++) {
       command_buffer->current_descriptor[i] = &device_ptrs[i];
     }
 
@@ -320,6 +321,21 @@
                           "need cuda implementation");
 }
 
+// Tie together the binding index and its index in |bindings| array.
+typedef struct {
+  uint32_t index;
+  uint32_t binding;
+} iree_hal_cuda_binding_mapping_t;
+
+// Helper to sort the binding based on their binding index.
+static int compare_binding_index(const void* a, const void* b) {
+  const iree_hal_cuda_binding_mapping_t buffer_a =
+      *(const iree_hal_cuda_binding_mapping_t*)a;
+  const iree_hal_cuda_binding_mapping_t buffer_b =
+      *(const iree_hal_cuda_binding_mapping_t*)b;
+  return buffer_a.binding < buffer_b.binding ? -1 : 1;
+}
+
 static iree_status_t iree_hal_cuda_graph_command_buffer_push_descriptor_set(
     iree_hal_command_buffer_t* base_command_buffer,
     iree_hal_executable_layout_t* executable_layout, uint32_t set,
@@ -327,15 +343,27 @@
     const iree_hal_descriptor_set_binding_t* bindings) {
   iree_hal_cuda_graph_command_buffer_t* command_buffer =
       iree_hal_cuda_graph_command_buffer_cast(base_command_buffer);
+  // Convention with the compiler side. We map bindings to kernel argument.
+  // We compact the bindings to get a dense set of arguments and keep them order
+  // based on the binding index.
+  // Sort the binding based on the binding index and map the array index to the
+  // argument index.
+  iree_hal_cuda_binding_mapping_t binding_used[IREE_HAL_CUDA_MAX_BINDING_COUNT];
   for (iree_host_size_t i = 0; i < binding_count; i++) {
-    uint32_t arg_index = bindings[i].binding;
-    assert(arg_index < max_binding_count &&
-           "binding index larger than the max expected.");
+    iree_hal_cuda_binding_mapping_t buffer = {i, bindings[i].binding};
+    binding_used[i] = buffer;
+  }
+  qsort(binding_used, binding_count, sizeof(iree_hal_cuda_binding_mapping_t),
+        compare_binding_index);
+  assert(binding_count < IREE_HAL_CUDA_MAX_BINDING_COUNT &&
+         "binding count larger than the max expected.");
+  for (iree_host_size_t i = 0; i < binding_count; i++) {
+    iree_hal_descriptor_set_binding_t binding = bindings[binding_used[i].index];
     CUdeviceptr device_ptr =
         iree_hal_cuda_buffer_device_pointer(
-            iree_hal_buffer_allocated_buffer(bindings[i].buffer)) +
-        iree_hal_buffer_byte_offset(bindings[i].buffer) + bindings[i].offset;
-    *((CUdeviceptr*)command_buffer->current_descriptor[arg_index]) = device_ptr;
+            iree_hal_buffer_allocated_buffer(binding.buffer)) +
+        iree_hal_buffer_byte_offset(binding.buffer) + binding.offset;
+    *((CUdeviceptr*)command_buffer->current_descriptor[i]) = device_ptr;
   }
   return iree_ok_status();
 }
diff --git a/iree/samples/variables_and_state/CMakeLists.txt b/iree/samples/variables_and_state/CMakeLists.txt
new file mode 100644
index 0000000..e6244af
--- /dev/null
+++ b/iree/samples/variables_and_state/CMakeLists.txt
@@ -0,0 +1,18 @@
+# Copyright 2021 The IREE Authors
+#
+# Licensed under the Apache License v2.0 with LLVM Exceptions.
+# See https://llvm.org/LICENSE.txt for license information.
+# SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+set(_NAME "iree_samples_variables_and_state")
+add_executable(${_NAME} "")
+target_sources(${_NAME}
+  PRIVATE
+    main.c
+)
+
+set_target_properties(${_NAME} PROPERTIES OUTPUT_NAME "variables-and-state")
+
+target_link_libraries(${_NAME}
+  iree_runtime_runtime
+)
diff --git a/iree/samples/variables_and_state/README.md b/iree/samples/variables_and_state/README.md
new file mode 100644
index 0000000..b219ea2
--- /dev/null
+++ b/iree/samples/variables_and_state/README.md
@@ -0,0 +1,103 @@
+# "Variables and State" sample
+
+This sample shows how to
+
+1. Create a TensorFlow program that tracks state with an internal variable
+2. Import that program into IREE's compiler
+3. Compile that program to an IREE VM bytecode module
+4. Load the compiled program using IREE's high level runtime C API
+5. Call exported functions on the loaded program to interact with the internal
+   variable
+
+Steps 1-3 are performed in Python via the
+[`variables_and_state.ipynb`](./variables_and_state.ipynb)
+[Colab](https://research.google.com/colaboratory/) notebook:
+
+[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/google/iree/blob/main/iree/samples/variables_and_state/variables_and_state.ipynb)
+
+Steps 4-5 are in [`main.c`](./main.c)
+
+The program used to demonstrate tracks a single integer counter and provides
+a few functions for interacting with it:
+
+```python
+class CounterModule(tf.Module):
+  def __init__(self):
+    super().__init__()
+    self.counter = tf.Variable(0)
+
+  @tf.function(input_signature=[])
+  def get_value(self):
+    return self.counter
+
+  @tf.function(input_signature=[tf.TensorSpec([], tf.int32)])
+  def set_value(self, new_value):
+    self.counter.assign(new_value)
+
+  @tf.function(input_signature=[tf.TensorSpec([], tf.int32)])
+  def add_to_value(self, x):
+    self.counter.assign(self.counter + x)
+
+  @tf.function(input_signature=[])
+  def reset_value(self):
+    self.set_value(0)
+```
+
+## Background
+
+Just like in other programming models, _variables_ in this context are used to
+represent some persistent state that programs may manipulate. See TensorFlow's
+[Introduction to Variables](https://www.tensorflow.org/guide/variable) for
+more information on using variables in TensorFlow.
+
+Mutable variables and internal program state are modeled natively in IREE along
+with the rest of a program's executable code and data. Variables are not given
+special treatment - program authors must define explicit exported functions for
+interacting with them and then invoke these functions as they would any other
+functions in the compiled programs.
+
+## Instructions
+
+1. Run the Colab notebook and download the `counter.mlir` and
+   `counter_vmvx.vmfb` files it generates
+2. Compile the `iree_samples_variables_and_state` CMake target (see
+   [here](https://google.github.io/iree/building-from-source/getting-started/)
+   for general instructions on building using CMake)
+
+    ```
+    cmake -B ../iree-build/ -DCMAKE_BUILD_TYPE=RelWithDebInfo .
+    cmake --build ../iree-build/ --target iree_samples_variables_and_state
+    ```
+3. Run the sample binary:
+
+   ```
+   ../iree-build/iree/samples/variables_and_state/variables-and-state \
+       /path/to/counter_vmvx.vmfb vmvx
+   ```
+
+### Changing compilation options
+
+The provided Colab notebook imports the TensorFlow program into MLIR and then
+compiles it further to an IREE VM bytecode module for IREE's reference CPU
+backend named "VMVX". To use a different backend, set compilation options, or
+include source code changes to the compiler, you can compile the imported MLIR
+file using IREE's tools on your own machine.
+
+For example, to use IREE's `dylib-llvm-aot` target, which is optimized for CPU
+execution using LLVM, refer to the
+[documentation](https://google.github.io/iree/deployment-configurations/cpu-dylib/)
+and compile the imported `counter.mlir` file using `iree-translate`:
+
+```
+../iree-build/iree/tools/iree-translate \
+    -iree-mlir-to-vm-bytecode-module \
+    -iree-hal-target-backends=dylib-llvm-aot \
+    counter.mlir -o counter_dylib.vmfb
+```
+
+then run the program with that new VM bytecode module:
+
+```
+../iree-build/iree/samples/variables_and_state/variables-and-state \
+    /path/to/counter_dylib.vmfb dylib
+```
diff --git a/iree/samples/variables_and_state/main.c b/iree/samples/variables_and_state/main.c
new file mode 100644
index 0000000..9a7a288
--- /dev/null
+++ b/iree/samples/variables_and_state/main.c
@@ -0,0 +1,241 @@
+// Copyright 2021 The IREE Authors
+//
+// Licensed under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#include <stdio.h>
+
+#include "iree/runtime/api.h"
+
+iree_status_t counter_get_value(iree_runtime_session_t* session,
+                                int* out_value) {
+  iree_runtime_call_t call;
+  IREE_RETURN_IF_ERROR(iree_runtime_call_initialize_by_name(
+      session, iree_make_cstring_view("module.get_value"), &call));
+
+  iree_status_t status = iree_ok_status();
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_call_invoke(&call, /*flags=*/0);
+  }
+  iree_hal_buffer_view_t* buffer_view = NULL;
+  if (iree_status_is_ok(status)) {
+    status =
+        iree_runtime_call_outputs_pop_front_buffer_view(&call, &buffer_view);
+  }
+  iree_hal_buffer_mapping_t buffer_mapping;
+  if (iree_status_is_ok(status)) {
+    status = iree_hal_buffer_map_range(iree_hal_buffer_view_buffer(buffer_view),
+                                       IREE_HAL_MEMORY_ACCESS_READ, 0,
+                                       IREE_WHOLE_BUFFER, &buffer_mapping);
+  }
+  if (iree_status_is_ok(status)) {
+    *out_value = *buffer_mapping.contents.data;
+  }
+  iree_hal_buffer_unmap_range(&buffer_mapping);
+  iree_hal_buffer_view_release(buffer_view);
+
+  iree_runtime_call_deinitialize(&call);
+  return status;
+}
+
+iree_status_t counter_set_value(iree_runtime_session_t* session,
+                                int new_value) {
+  iree_runtime_call_t call;
+  IREE_RETURN_IF_ERROR(iree_runtime_call_initialize_by_name(
+      session, iree_make_cstring_view("module.set_value"), &call));
+
+  iree_hal_buffer_view_t* arg0 = NULL;
+  static const iree_hal_dim_t arg0_shape[1] = {1};
+  int arg0_data[1] = {new_value};
+
+  // TODO(scotttodd): use iree_hal_buffer_view_wrap_or_clone_heap_buffer
+  //   * debugging some apparent memory corruption with the stack-local value
+  iree_status_t status = iree_ok_status();
+  if (iree_status_is_ok(status)) {
+    status = iree_hal_buffer_view_clone_heap_buffer(
+        iree_runtime_session_device_allocator(session), arg0_shape,
+        IREE_ARRAYSIZE(arg0_shape), IREE_HAL_ELEMENT_TYPE_SINT_32,
+        IREE_HAL_MEMORY_TYPE_HOST_LOCAL | IREE_HAL_MEMORY_TYPE_DEVICE_VISIBLE,
+        IREE_HAL_BUFFER_USAGE_ALL,
+        iree_make_const_byte_span((void*)arg0_data, sizeof(arg0_data)), &arg0);
+  }
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_call_inputs_push_back_buffer_view(&call, arg0);
+  }
+  iree_hal_buffer_view_release(arg0);
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_call_invoke(&call, /*flags=*/0);
+  }
+
+  iree_runtime_call_deinitialize(&call);
+  return status;
+}
+
+iree_status_t counter_add_to_value(iree_runtime_session_t* session, int x) {
+  iree_runtime_call_t call;
+  IREE_RETURN_IF_ERROR(iree_runtime_call_initialize_by_name(
+      session, iree_make_cstring_view("module.add_to_value"), &call));
+
+  iree_hal_buffer_view_t* arg0 = NULL;
+  static const iree_hal_dim_t arg0_shape[1] = {1};
+  int arg0_data[1] = {x};
+
+  // TODO(scotttodd): use iree_hal_buffer_view_wrap_or_clone_heap_buffer
+  //   * debugging some apparent memory corruption with the stack-local value
+  iree_status_t status = iree_ok_status();
+  if (iree_status_is_ok(status)) {
+    status = iree_hal_buffer_view_clone_heap_buffer(
+        iree_runtime_session_device_allocator(session), arg0_shape,
+        IREE_ARRAYSIZE(arg0_shape), IREE_HAL_ELEMENT_TYPE_SINT_32,
+        IREE_HAL_MEMORY_TYPE_HOST_LOCAL | IREE_HAL_MEMORY_TYPE_DEVICE_VISIBLE,
+        IREE_HAL_BUFFER_USAGE_ALL,
+        iree_make_const_byte_span((void*)arg0_data, sizeof(arg0_data)), &arg0);
+  }
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_call_inputs_push_back_buffer_view(&call, arg0);
+  }
+  iree_hal_buffer_view_release(arg0);
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_call_invoke(&call, /*flags=*/0);
+  }
+
+  iree_runtime_call_deinitialize(&call);
+  return status;
+}
+
+iree_status_t counter_reset_value(iree_runtime_session_t* session) {
+  iree_runtime_call_t call;
+  IREE_RETURN_IF_ERROR(iree_runtime_call_initialize_by_name(
+      session, iree_make_cstring_view("module.reset_value"), &call));
+
+  iree_status_t status = iree_runtime_call_invoke(&call, /*flags=*/0);
+
+  iree_runtime_call_deinitialize(&call);
+  return status;
+}
+
+iree_status_t run_sample(iree_string_view_t bytecode_module_path,
+                         iree_string_view_t driver_name) {
+  iree_status_t status = iree_ok_status();
+
+  //===-------------------------------------------------------------------===//
+  // Instance configuration (this should be shared across sessions).
+  fprintf(stdout, "Configuring IREE runtime instance and '%s' device\n",
+          driver_name.data);
+  iree_runtime_instance_options_t instance_options;
+  iree_runtime_instance_options_initialize(IREE_API_VERSION_LATEST,
+                                           &instance_options);
+  iree_runtime_instance_options_use_all_available_drivers(&instance_options);
+  iree_runtime_instance_t* instance = NULL;
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_instance_create(&instance_options,
+                                          iree_allocator_system(), &instance);
+  }
+  // TODO(#5724): move device selection into the compiled modules.
+  iree_hal_device_t* device = NULL;
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_instance_try_create_default_device(
+        instance, driver_name, &device);
+  }
+  //===-------------------------------------------------------------------===//
+
+  //===-------------------------------------------------------------------===//
+  // Session configuration (one per loaded module to hold module state).
+  fprintf(stdout, "Creating IREE runtime session\n");
+  iree_runtime_session_options_t session_options;
+  iree_runtime_session_options_initialize(&session_options);
+  iree_runtime_session_t* session = NULL;
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_session_create_with_device(
+        instance, &session_options, device,
+        iree_runtime_instance_host_allocator(instance), &session);
+  }
+  iree_hal_device_release(device);
+
+  fprintf(stdout, "Loading bytecode module at '%s'\n",
+          bytecode_module_path.data);
+  if (iree_status_is_ok(status)) {
+    status = iree_runtime_session_append_bytecode_module_from_file(
+        session, bytecode_module_path.data);
+  }
+  //===-------------------------------------------------------------------===//
+
+  //===-------------------------------------------------------------------===//
+  // Call functions to manipulate the counter
+  fprintf(stdout, "Calling functions\n\n");
+
+  // 1. get_value() // initial value
+  int value;
+  if (iree_status_is_ok(status)) {
+    status = counter_get_value(session, &value);
+  }
+  fprintf(stdout, "Initial get_value()    : %d\n", value);
+
+  // 2. set_value(101)
+  if (iree_status_is_ok(status)) {
+    status = counter_set_value(session, 101);
+  }
+  if (iree_status_is_ok(status)) {
+    status = counter_get_value(session, &value);
+  }
+  fprintf(stdout, "After set_value(101)   : %d\n", value);
+
+  // 3. add_to_value(20)
+  if (iree_status_is_ok(status)) {
+    status = counter_add_to_value(session, 20);
+  }
+  if (iree_status_is_ok(status)) {
+    status = counter_get_value(session, &value);
+  }
+  fprintf(stdout, "After add_to_value(20) : %d\n", value);
+
+  // 4. add_to_value(-50)
+  if (iree_status_is_ok(status)) {
+    status = counter_add_to_value(session, -50);
+  }
+  if (iree_status_is_ok(status)) {
+    status = counter_get_value(session, &value);
+  }
+  fprintf(stdout, "After add_to_value(-50): %d\n", value);
+
+  // 5. reset_value()
+  if (iree_status_is_ok(status)) {
+    status = counter_reset_value(session);
+  }
+  if (iree_status_is_ok(status)) {
+    status = counter_get_value(session, &value);
+  }
+  fprintf(stdout, "After reset_value()    : %d\n", value);
+  //===-------------------------------------------------------------------===//
+
+  //===-------------------------------------------------------------------===//
+  // Cleanup.
+  iree_runtime_session_release(session);
+  iree_runtime_instance_release(instance);
+  //===-------------------------------------------------------------------===//
+
+  return status;
+}
+
+int main(int argc, char** argv) {
+  if (argc != 3) {
+    fprintf(
+        stderr,
+        "Usage: variables-and-state </path/to/counter.vmfb> <driver_name>\n");
+    fprintf(stderr, "  (See the README for this sample for details)\n ");
+    return -1;
+  }
+
+  iree_string_view_t bytecode_module_path = iree_make_cstring_view(argv[1]);
+  iree_string_view_t driver_name = iree_make_cstring_view(argv[2]);
+
+  iree_status_t result = run_sample(bytecode_module_path, driver_name);
+  if (!iree_status_is_ok(result)) {
+    iree_status_fprint(stderr, result);
+    iree_status_ignore(result);
+    return -1;
+  }
+  fprintf(stdout, "\nSuccess!\n");
+  return 0;
+}
diff --git a/iree/samples/variables_and_state/variables_and_state.ipynb b/iree/samples/variables_and_state/variables_and_state.ipynb
new file mode 100644
index 0000000..004a644
--- /dev/null
+++ b/iree/samples/variables_and_state/variables_and_state.ipynb
@@ -0,0 +1,482 @@
+{
+  "nbformat": 4,
+  "nbformat_minor": 0,
+  "metadata": {
+    "colab": {
+      "name": "variables_and_state.ipynb",
+      "provenance": [],
+      "collapsed_sections": []
+    },
+    "kernelspec": {
+      "display_name": "Python 3",
+      "name": "python3"
+    }
+  },
+  "cells": [
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "j_IgB6_3npkE"
+      },
+      "source": [
+        "```\n",
+        "Copyright 2021 The IREE Authors\n",
+        " \n",
+        "Licensed under the Apache License v2.0 with LLVM Exceptions.\n",
+        "See https://llvm.org/LICENSE.txt for license information.\n",
+        "SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception\n",
+        "```"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "h5s6ncerSpc5"
+      },
+      "source": [
+        "# Variables and State\n",
+        "\n",
+        "This notebook\n",
+        "\n",
+        "1. Creates a TensorFlow program with basic tf.Variable use\n",
+        "2. Imports that program into IREE's compiler\n",
+        "3. Compiles the imported program to an IREE VM bytecode module\n",
+        "4. Tests running the compiled VM module using IREE's runtime\n",
+        "5. Downloads compilation artifacts for use with the native (C API) sample application"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "s2bScbYkP6VZ",
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "outputId": "d0d207d5-ca38-4ef9-881b-f0464bccff8e"
+      },
+      "source": [
+        "#@title General setup\n",
+        "\n",
+        "import os\n",
+        "import tempfile\n",
+        "\n",
+        "ARTIFACTS_DIR = os.path.join(tempfile.gettempdir(), \"iree\", \"colab_artifacts\")\n",
+        "os.makedirs(ARTIFACTS_DIR, exist_ok=True)\n",
+        "print(f\"Using artifacts directory '{ARTIFACTS_DIR}'\")"
+      ],
+      "execution_count": 1,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "text": [
+            "Using artifacts directory '/tmp/iree/colab_artifacts'\n"
+          ],
+          "name": "stdout"
+        }
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "dBHgjTjGPOJ7"
+      },
+      "source": [
+        "## Create a program using TensorFlow and import it into IREE\n",
+        "\n",
+        "This program uses `tf.Variable` to track state internal to the program then exports functions which can be used to interact with that variable.\n",
+        "\n",
+        "Note that each function we want to be callable from our compiled program needs\n",
+        "to use `@tf.function` with an `input_signature` specified.\n",
+        "\n",
+        "References:\n",
+        "\n",
+        "* [\"Introduction to Variables\" Guide](https://www.tensorflow.org/guide/variable)\n",
+        "* [`tf.Variable` reference](https://www.tensorflow.org/api_docs/python/tf/Variable)\n",
+        "* [`tf.function` reference](https://www.tensorflow.org/api_docs/python/tf/function)"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "hwApbPstraWZ"
+      },
+      "source": [
+        "#@title Define a simple \"counter\" TensorFlow module\n",
+        "\n",
+        "import tensorflow as tf\n",
+        "\n",
+        "class CounterModule(tf.Module):\n",
+        "  def __init__(self):\n",
+        "    super().__init__()\n",
+        "    self.counter = tf.Variable(0)\n",
+        "\n",
+        "  @tf.function(input_signature=[])\n",
+        "  def get_value(self):\n",
+        "    return self.counter\n",
+        "    \n",
+        "  @tf.function(input_signature=[tf.TensorSpec([], tf.int32)])\n",
+        "  def set_value(self, new_value):\n",
+        "    self.counter.assign(new_value)\n",
+        "    \n",
+        "  @tf.function(input_signature=[tf.TensorSpec([], tf.int32)])\n",
+        "  def add_to_value(self, x):\n",
+        "    self.counter.assign(self.counter + x)\n",
+        "\n",
+        "  @tf.function(input_signature=[])\n",
+        "  def reset_value(self):\n",
+        "    self.set_value(0)"
+      ],
+      "execution_count": 2,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "k4aMPI2C7btB"
+      },
+      "source": [
+        "%%capture\n",
+        "!python -m pip install iree-compiler-snapshot iree-tools-tf-snapshot -f https://github.com/google/iree/releases"
+      ],
+      "execution_count": 3,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "3nSXZiZ_X8-P",
+        "outputId": "9dd6e263-4751-480e-d300-e424b4d7fd8d"
+      },
+      "source": [
+        "#@title Import the TensorFlow program into IREE as MLIR\n",
+        "\n",
+        "from IPython.display import clear_output\n",
+        "\n",
+        "from iree.compiler import tf as tfc\n",
+        "\n",
+        "compiler_module = tfc.compile_module(CounterModule(), import_only=True)\n",
+        "clear_output()  # Skip over TensorFlow's output.\n",
+        "\n",
+        "# Print the imported MLIR to see how the compiler views this TensorFlow program.\n",
+        "# Note IREE's `flow.variable` ops and the public (exported) functions.\n",
+        "print(\"Counter MLIR:\\n```\\n\", compiler_module.decode(\"utf-8\"), \"```\\n\")\n",
+        "\n",
+        "# Save the imported MLIR to disk.\n",
+        "imported_mlir_path = os.path.join(ARTIFACTS_DIR, \"counter.mlir\")\n",
+        "with open(imported_mlir_path, \"wt\") as output_file:\n",
+        "  output_file.write(compiler_module.decode(\"utf-8\"))\n",
+        "print(f\"Wrote MLIR to path '{imported_mlir_path}'\")"
+      ],
+      "execution_count": 4,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "text": [
+            "Counter MLIR:\n",
+            "```\n",
+            " #loc0 = loc(unknown)\n",
+            "module  {\n",
+            "  flow.variable @counter mutable dense<0> : tensor<i32> attributes {sym_visibility = \"private\"} loc(#loc0)\n",
+            "  func @add_to_value(%arg0: tensor<i32> {tf._user_specified_name = \"x\"} loc(unknown)) attributes {iree.module.export, iree.reflection = {abi = \"sip\", abiv = 1 : i32, sip = \"I8!S5!k0_0R1!\"}, tf._construction_context = \"kEagerRuntime\", tf._input_shapes = [#tf.shape<>, #tf.shape<>], tf.signature.is_stateful} {\n",
+            "    %0 = flow.variable.address @counter : !iree.ptr<tensor<i32>> loc(#loc0)\n",
+            "    %1 = flow.variable.load.indirect %0 : !iree.ptr<tensor<i32>> -> tensor<i32> loc(#loc1)\n",
+            "    %2 = mhlo.add %1, %arg0 : tensor<i32> loc(#loc2)\n",
+            "    flow.variable.store.indirect %2, %0 : tensor<i32> -> !iree.ptr<tensor<i32>> loc(#loc3)\n",
+            "    return loc(#loc0)\n",
+            "  } loc(#loc0)\n",
+            "  func @get_value() -> tensor<i32> attributes {iree.module.export, iree.reflection = {abi = \"sip\", abiv = 1 : i32, sip = \"I1!R3!_0\"}, tf._construction_context = \"kEagerRuntime\", tf._input_shapes = [#tf.shape<>], tf.signature.is_stateful} {\n",
+            "    %0 = flow.variable.address @counter : !iree.ptr<tensor<i32>> loc(#loc0)\n",
+            "    %1 = flow.variable.load.indirect %0 : !iree.ptr<tensor<i32>> -> tensor<i32> loc(#loc4)\n",
+            "    return %1 : tensor<i32> loc(#loc0)\n",
+            "  } loc(#loc0)\n",
+            "  func @reset_value() attributes {iree.module.export, iree.reflection = {abi = \"sip\", abiv = 1 : i32, sip = \"I1!R1!\"}, tf._construction_context = \"kEagerRuntime\", tf._input_shapes = [#tf.shape<>], tf.signature.is_stateful} {\n",
+            "    %0 = flow.variable.address @counter : !iree.ptr<tensor<i32>> loc(#loc0)\n",
+            "    %1 = mhlo.constant dense<0> : tensor<i32> loc(#loc5)\n",
+            "    flow.variable.store.indirect %1, %0 : tensor<i32> -> !iree.ptr<tensor<i32>> loc(#loc6)\n",
+            "    return loc(#loc0)\n",
+            "  } loc(#loc0)\n",
+            "  func @set_value(%arg0: tensor<i32> {tf._user_specified_name = \"new_value\"} loc(unknown)) attributes {iree.module.export, iree.reflection = {abi = \"sip\", abiv = 1 : i32, sip = \"I8!S5!k0_0R1!\"}, tf._construction_context = \"kEagerRuntime\", tf._input_shapes = [#tf.shape<>, #tf.shape<>], tf.signature.is_stateful} {\n",
+            "    %0 = flow.variable.address @counter : !iree.ptr<tensor<i32>> loc(#loc0)\n",
+            "    flow.variable.store.indirect %arg0, %0 : tensor<i32> -> !iree.ptr<tensor<i32>> loc(#loc7)\n",
+            "    return loc(#loc0)\n",
+            "  } loc(#loc0)\n",
+            "} loc(#loc0)\n",
+            "#loc1 = loc(callsite(\"ReadVariableOp@__inference_add_to_value_10\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/resource_variable_ops.py\":575:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/variables.py\":1074:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":20:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc2 = loc(callsite(\"add@__inference_add_to_value_10\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/math_ops.py\":1234:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/variables.py\":1074:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":20:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc3 = loc(callsite(\"AssignVariableOp@__inference_add_to_value_10\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/gen_resource_variable_ops.py\":155:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/resource_variable_ops.py\":903:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":20:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc4 = loc(callsite(\"ReadVariableOp@__inference_get_value_16\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/saved_model/save.py\":1104:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/saved_model/save.py\":1066:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/iree/compiler/tf.py\":204:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/iree/compiler/tf.py\":211:0 at callsite(\"<ipython-input-4-601a89dcf132>\":7:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\":2882:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\":2822:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/IPython/core/interactiveshell.py\":2718:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/ipykernel/zmqshell.py\":537:0 at \"/usr/local/lib/python3.7/dist-packages/ipykernel/ipkernel.py\":208:0))))))))))\n",
+            "#loc5 = loc(callsite(\"Const@__inference_reset_value_27\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":954:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":889:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":24:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "#loc6 = loc(callsite(callsite(\"AssignVariableOp@__inference_set_value_23\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/gen_resource_variable_ops.py\":155:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/resource_variable_ops.py\":903:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":16:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))) at callsite(\"StatefulPartitionedCall@__inference_reset_value_27\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":957:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":889:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":24:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0)))))))))))\n",
+            "#loc7 = loc(callsite(\"AssignVariableOp@__inference_set_value_23\"(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/gen_resource_variable_ops.py\":155:0) at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/ops/resource_variable_ops.py\":903:0 at callsite(\"<ipython-input-2-4934d5ff663c>\":16:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":982:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3971:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/def_function.py\":672:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/framework/func_graph.py\":999:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3289:0 at callsite(\"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3444:0 at \"/usr/local/lib/python3.7/dist-packages/tensorflow/python/eager/function.py\":3050:0))))))))))\n",
+            "\n",
+            " ```\n",
+            "\n",
+            "Wrote MLIR to path '/tmp/iree/colab_artifacts/counter.mlir'\n"
+          ],
+          "name": "stdout"
+        }
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "WCiRV6KRh3iA"
+      },
+      "source": [
+        "## Test the imported program\n",
+        "\n",
+        "_Note: you can stop after each step and use intermediate outputs with other tools outside of Colab._\n",
+        "\n",
+        "_See the [README](https://github.com/google/iree/tree/main/iree/samples/variables_and_state#changing-compilation-options) for more details and example command line instructions._\n",
+        "\n",
+        "* _The \"imported MLIR\" can be used by IREE's generic compiler tools_\n",
+        "* _The \"flatbuffer blob\" can be saved and used by runtime applications_\n",
+        "\n",
+        "_The specific point at which you switch from Python to native tools will depend on your project._"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "6TV6_Hdu6Xlf"
+      },
+      "source": [
+        "%%capture\n",
+        "!python -m pip install iree-compiler-snapshot -f https://github.com/google/iree/releases"
+      ],
+      "execution_count": 5,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "GF0dzDsbaP2w",
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "outputId": "5b475dfa-2108-4740-fa26-3891e0106817"
+      },
+      "source": [
+        "#@title Compile the imported MLIR further into an IREE VM bytecode module\n",
+        "\n",
+        "from iree.compiler import compile_str\n",
+        "\n",
+        "flatbuffer_blob = compile_str(compiler_module, target_backends=[\"vmvx\"])\n",
+        "\n",
+        "# Save the imported MLIR to disk.\n",
+        "flatbuffer_path = os.path.join(ARTIFACTS_DIR, \"counter_vmvx.vmfb\")\n",
+        "with open(flatbuffer_path, \"wb\") as output_file:\n",
+        "  output_file.write(flatbuffer_blob)\n",
+        "print(f\"Wrote .vmfb to path '{flatbuffer_path}'\")"
+      ],
+      "execution_count": 6,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "text": [
+            "Wrote .vmfb to path '/tmp/iree/colab_artifacts/counter_vmvx.vmfb'\n"
+          ],
+          "name": "stdout"
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "id": "G7g5eXYL6hWb"
+      },
+      "source": [
+        "%%capture\n",
+        "!python -m pip install iree-runtime-snapshot -f https://github.com/google/iree/releases"
+      ],
+      "execution_count": 7,
+      "outputs": []
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "h8cmF6nAfza0",
+        "outputId": "1fb57ac9-5d63-4ea9-e672-d4b60fabd126"
+      },
+      "source": [
+        "#@title Test running the compiled VM module using IREE's runtime\n",
+        "\n",
+        "from iree import runtime as ireert\n",
+        "\n",
+        "vm_module = ireert.VmModule.from_flatbuffer(flatbuffer_blob)\n",
+        "config = ireert.Config(\"vmvx\")\n",
+        "ctx = ireert.SystemContext(config=config)\n",
+        "ctx.add_vm_module(vm_module)"
+      ],
+      "execution_count": 8,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "text": [
+            "Created IREE driver vmvx: <iree.runtime.binding.HalDriver object at 0x7fe4a92227b0>\n",
+            "SystemContext driver=<iree.runtime.binding.HalDriver object at 0x7fe4a92227b0>\n"
+          ],
+          "name": "stderr"
+        }
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/"
+        },
+        "id": "CQffg1iQatkb",
+        "outputId": "1b8ba22f-8bd7-4a3d-c1f5-b732d2534cb2"
+      },
+      "source": [
+        "# Our @tf.functions are accessible by name on the module named 'module'\n",
+        "counter = ctx.modules.module\n",
+        "\n",
+        "print(counter.get_value())\n",
+        "counter.set_value(101)\n",
+        "print(counter.get_value())\n",
+        "\n",
+        "counter.add_to_value(20)\n",
+        "print(counter.get_value())\n",
+        "counter.add_to_value(-50)\n",
+        "print(counter.get_value())\n",
+        "\n",
+        "counter.reset_value()\n",
+        "print(counter.get_value())"
+      ],
+      "execution_count": 9,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "text": [
+            "0\n",
+            "101\n",
+            "121\n",
+            "71\n",
+            "0\n"
+          ],
+          "name": "stdout"
+        }
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {
+        "id": "wCvwX1IEokm6"
+      },
+      "source": [
+        "## Download compilation artifacts"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "metadata": {
+        "colab": {
+          "base_uri": "https://localhost:8080/",
+          "height": 85
+        },
+        "id": "bUaNUkS2ohRj",
+        "outputId": "e06ae6f1-1c6e-4f74-f044-9479670ffa41"
+      },
+      "source": [
+        "ARTIFACTS_ZIP = \"/tmp/variables_and_state_colab_artifacts.zip\"\n",
+        "\n",
+        "print(f\"Zipping '{ARTIFACTS_DIR}' to '{ARTIFACTS_ZIP}' for download...\")\n",
+        "!cd {ARTIFACTS_DIR} && zip -r {ARTIFACTS_ZIP} .\n",
+        "\n",
+        "# Note: you can also download files using Colab's file explorer\n",
+        "from google.colab import files\n",
+        "print(\"Downloading the artifacts zip file...\")\n",
+        "files.download(ARTIFACTS_ZIP)"
+      ],
+      "execution_count": 10,
+      "outputs": [
+        {
+          "output_type": "stream",
+          "text": [
+            "Zipping '/tmp/iree/colab_artifacts' to '/tmp/variables_and_state_colab_artifacts.zip' for download...\n",
+            "  adding: counter_vmvx.vmfb (deflated 67%)\n",
+            "  adding: counter.mlir (deflated 90%)\n",
+            "Downloading the artifacts zip file...\n"
+          ],
+          "name": "stdout"
+        },
+        {
+          "output_type": "display_data",
+          "data": {
+            "application/javascript": [
+              "\n",
+              "    async function download(id, filename, size) {\n",
+              "      if (!google.colab.kernel.accessAllowed) {\n",
+              "        return;\n",
+              "      }\n",
+              "      const div = document.createElement('div');\n",
+              "      const label = document.createElement('label');\n",
+              "      label.textContent = `Downloading \"${filename}\": `;\n",
+              "      div.appendChild(label);\n",
+              "      const progress = document.createElement('progress');\n",
+              "      progress.max = size;\n",
+              "      div.appendChild(progress);\n",
+              "      document.body.appendChild(div);\n",
+              "\n",
+              "      const buffers = [];\n",
+              "      let downloaded = 0;\n",
+              "\n",
+              "      const channel = await google.colab.kernel.comms.open(id);\n",
+              "      // Send a message to notify the kernel that we're ready.\n",
+              "      channel.send({})\n",
+              "\n",
+              "      for await (const message of channel.messages) {\n",
+              "        // Send a message to notify the kernel that we're ready.\n",
+              "        channel.send({})\n",
+              "        if (message.buffers) {\n",
+              "          for (const buffer of message.buffers) {\n",
+              "            buffers.push(buffer);\n",
+              "            downloaded += buffer.byteLength;\n",
+              "            progress.value = downloaded;\n",
+              "          }\n",
+              "        }\n",
+              "      }\n",
+              "      const blob = new Blob(buffers, {type: 'application/binary'});\n",
+              "      const a = document.createElement('a');\n",
+              "      a.href = window.URL.createObjectURL(blob);\n",
+              "      a.download = filename;\n",
+              "      div.appendChild(a);\n",
+              "      a.click();\n",
+              "      div.remove();\n",
+              "    }\n",
+              "  "
+            ],
+            "text/plain": [
+              "<IPython.core.display.Javascript object>"
+            ]
+          },
+          "metadata": {
+            "tags": []
+          }
+        },
+        {
+          "output_type": "display_data",
+          "data": {
+            "application/javascript": [
+              "download(\"download_c4a3a2ad-0f6b-4b60-941f-76bdbd562f9c\", \"variables_and_state_colab_artifacts.zip\", 3927)"
+            ],
+            "text/plain": [
+              "<IPython.core.display.Javascript object>"
+            ]
+          },
+          "metadata": {
+            "tags": []
+          }
+        }
+      ]
+    }
+  ]
+}
\ No newline at end of file
diff --git a/iree/test/e2e/vulkan_specific/dot_general.mlir b/iree/test/e2e/vulkan_specific/dot_general.mlir
index 11cd69d..250afb9 100644
--- a/iree/test/e2e/vulkan_specific/dot_general.mlir
+++ b/iree/test/e2e/vulkan_specific/dot_general.mlir
@@ -48,18 +48,19 @@
   return
 }
 
-func @large_dot_general2() attributes { iree.module.export } {
-  %lhs = iree.unfoldable_constant dense<1.0> : tensor<4x32x1024xf32>
-  %rhs = iree.unfoldable_constant dense<0.4> : tensor<4x1024x64xf32>
-  %res = "mhlo.dot_general"(%lhs, %rhs) {
-           dot_dimension_numbers = {
-             lhs_batching_dimensions = dense<0> : tensor<1xi64>,
-             lhs_contracting_dimensions = dense<2> : tensor<1xi64>,
-             rhs_batching_dimensions = dense<0> : tensor<1xi64>,
-             rhs_contracting_dimensions = dense<1> : tensor<1xi64>
-           },
-           precision_config = ["DEFAULT", "DEFAULT"]
-         } : (tensor<4x32x1024xf32>, tensor<4x1024x64xf32>) -> tensor<4x32x64xf32>
-  check.expect_almost_eq_const(%res, dense<409.596> : tensor<4x32x64xf32>) : tensor<4x32x64xf32>
-  return
-}
+// TODO(GH-6070): Re-enable this after fixing wrong answer on Pixel 4.
+// func @large_dot_general2() attributes { iree.module.export } {
+//   %lhs = iree.unfoldable_constant dense<1.0> : tensor<4x32x1024xf32>
+//   %rhs = iree.unfoldable_constant dense<0.4> : tensor<4x1024x64xf32>
+//   %res = "mhlo.dot_general"(%lhs, %rhs) {
+//            dot_dimension_numbers = {
+//              lhs_batching_dimensions = dense<0> : tensor<1xi64>,
+//              lhs_contracting_dimensions = dense<2> : tensor<1xi64>,
+//              rhs_batching_dimensions = dense<0> : tensor<1xi64>,
+//              rhs_contracting_dimensions = dense<1> : tensor<1xi64>
+//            },
+//            precision_config = ["DEFAULT", "DEFAULT"]
+//          } : (tensor<4x32x1024xf32>, tensor<4x1024x64xf32>) -> tensor<4x32x64xf32>
+//   check.expect_almost_eq_const(%res, dense<409.596> : tensor<4x32x64xf32>) : tensor<4x32x64xf32>
+//   return
+// }
diff --git a/iree/tools/BUILD b/iree/tools/BUILD
index 86e4fbd..0e729c9 100644
--- a/iree/tools/BUILD
+++ b/iree/tools/BUILD
@@ -89,7 +89,6 @@
         "//iree/compiler/Dialect/IREE/Transforms",
         "//iree/compiler/Dialect/Modules/VMVX/IR:VMVXDialect",
         "//iree/compiler/Dialect/Modules/VMVX/Transforms",
-        "//iree/compiler/Dialect/Shape/Conversion",
         "//iree/compiler/Dialect/Shape/IR",
         "//iree/compiler/Dialect/Shape/Transforms",
         "//iree/compiler/Dialect/VM/Analysis",
@@ -150,6 +149,7 @@
         ":init_xla_dialects",
         "//iree/compiler/Conversion",
         "//iree/compiler/Conversion:init_conversions",
+        "//iree/compiler/Conversion/HLOToLinalg",
         "//iree/compiler/Dialect/HAL/Conversion:Passes",
     ],
 )
diff --git a/iree/tools/CMakeLists.txt b/iree/tools/CMakeLists.txt
index 6f6a028..fb42f8f 100644
--- a/iree/tools/CMakeLists.txt
+++ b/iree/tools/CMakeLists.txt
@@ -163,7 +163,6 @@
       iree::compiler::Dialect::IREE::Transforms
       iree::compiler::Dialect::Modules::VMVX::IR::VMVXDialect
       iree::compiler::Dialect::Modules::VMVX::Transforms
-      iree::compiler::Dialect::Shape::Conversion
       iree::compiler::Dialect::Shape::IR
       iree::compiler::Dialect::Shape::Transforms
       iree::compiler::Dialect::VM::Analysis
diff --git a/iree/tools/init_iree_passes.h b/iree/tools/init_iree_passes.h
index 90ab127..41b2c83 100644
--- a/iree/tools/init_iree_passes.h
+++ b/iree/tools/init_iree_passes.h
@@ -20,7 +20,6 @@
 #include "iree/compiler/Dialect/HAL/Transforms/Passes.h"
 #include "iree/compiler/Dialect/IREE/Transforms/Passes.h"
 #include "iree/compiler/Dialect/Modules/VMVX/Transforms/Passes.h"
-#include "iree/compiler/Dialect/Shape/Conversion/Passes.h"
 #include "iree/compiler/Dialect/Shape/Transforms/Passes.h"
 #include "iree/compiler/Dialect/VM/Analysis/TestPasses.h"
 #include "iree/compiler/Dialect/VM/Transforms/Passes.h"
@@ -40,7 +39,6 @@
   IREE::Flow::registerFlowPasses();
   IREE::HAL::registerHALPasses();
   IREE::registerTransformPasses();
-  Shape::registerShapeConversionPasses();
   Shape::registerShapePasses();
   IREE::VM::registerVMPasses();
   IREE::VM::registerVMAnalysisTestPasses();