Two outputs per iteration in ConvPerChannelD1OD24_5x5 Change-Id: I8929a0273e935bcf538095c554dc4be1900ba5e3
diff --git a/tflm/opt/conv_s8.cc b/tflm/opt/conv_s8.cc index a92433f..0dc432b 100644 --- a/tflm/opt/conv_s8.cc +++ b/tflm/opt/conv_s8.cc
@@ -220,6 +220,7 @@ } if (input_depth == 1 && filter_width == 5 && filter_height == 5 && + stride_width == 2 && output_depth == 24) { RUN_KERNEL(kelvin::opt::ConvPerChannelD1OD24_5x5); }
diff --git a/tflm/opt/conv_s8_d1.cc b/tflm/opt/conv_s8_d1.cc index 09f4110..c03a7f5 100644 --- a/tflm/opt/conv_s8_d1.cc +++ b/tflm/opt/conv_s8_d1.cc
@@ -67,6 +67,67 @@ } // namespace +#define FLT_0_0 v0 +#define FLT_0_1 v3 +#define FLT_0_2 v6 +#define FLT_0_3 v9 +#define FLT_0_4 v12 + +#define FLT_1_0 v1 +#define FLT_1_1 v4 +#define FLT_1_2 v7 +#define FLT_1_3 v10 +#define FLT_1_4 v13 + +#define FLT_2_0 v2 +#define FLT_2_1 v5 +#define FLT_2_2 v8 +#define FLT_2_3 v11 +#define FLT_2_4 v14 + +#define FLT_3_0 v15 +#define FLT_3_1 v16 +#define FLT_3_2 v17 +#define FLT_3_3 v18 +#define FLT_3_4 v19 + +#define FLT_HOLE v20 +#define FLT_4_0 v21 +#define FLT_4_1 v22 +#define FLT_4_2 v23 +#define FLT_4_3 v24 +#define FLT_4_4 v25 + +#define INPUT_0_0 v26 +#define INPUT_0_1 v29 +#define INPUT_0_2 v32 +#define INPUT_0_3 v35 +#define INPUT_0_4 v38 + +#define INPUT_1_0 v27 +#define INPUT_1_1 v30 +#define INPUT_1_2 v33 +#define INPUT_1_3 v36 +#define INPUT_1_4 v39 + +#define INPUT_2_0 v28 +#define INPUT_2_1 v31 +#define INPUT_2_2 v34 +#define INPUT_2_3 v37 +#define INPUT_2_4 v40 + +#define INPUT_3_0 v41 +#define INPUT_3_1 v42 +#define INPUT_3_2 v43 +#define INPUT_3_3 v44 +#define INPUT_3_4 v45 + +#define INPUT_4_0 v46 +#define INPUT_4_1 v47 +#define INPUT_4_2 v48 +#define INPUT_4_3 v49 +#define INPUT_4_4 v50 + #define CALCULATE_IN_X(in_x_origin) \ { \ _Pragma("GCC unroll 5") for (int i = 0; i < 5; ++i) { \ @@ -83,93 +144,93 @@ #define PAD_ROW_0(input_offset) \ { \ - vdup_b_x(v27, -input_offset); \ - vdup_b_x(v28, -input_offset); \ - vdup_b_x(v29, -input_offset); \ - vdup_b_x(v30, -input_offset); \ - vdup_b_x(v31, -input_offset); \ + vdup_b_x(INPUT_0_0, -input_offset); \ + vdup_b_x(INPUT_0_1, -input_offset); \ + vdup_b_x(INPUT_0_2, -input_offset); \ + vdup_b_x(INPUT_0_3, -input_offset); \ + vdup_b_x(INPUT_0_4, -input_offset); \ } #define PAD_ROW_1(input_offset) \ { \ - vdup_b_x(v32, -input_offset); \ - vdup_b_x(v33, -input_offset); \ - vdup_b_x(v34, -input_offset); \ - vdup_b_x(v35, -input_offset); \ - vdup_b_x(v36, -input_offset); \ + vdup_b_x(INPUT_1_0, -input_offset); \ + vdup_b_x(INPUT_1_1, -input_offset); \ + vdup_b_x(INPUT_1_2, -input_offset); \ + vdup_b_x(INPUT_1_3, -input_offset); \ + vdup_b_x(INPUT_1_4, -input_offset); \ } #define PAD_ROW_2(input_offset) \ - { \ - vdup_b_x(v37, -input_offset); \ - vdup_b_x(v38, -input_offset); \ - vdup_b_x(v39, -input_offset); \ - vdup_b_x(v40, -input_offset); \ - vdup_b_x(v41, -input_offset); \ + { \ + vdup_b_x(INPUT_2_0, -input_offset); \ + vdup_b_x(INPUT_2_1, -input_offset); \ + vdup_b_x(INPUT_2_2, -input_offset); \ + vdup_b_x(INPUT_2_3, -input_offset); \ + vdup_b_x(INPUT_2_4, -input_offset); \ } #define PAD_ROW_3(input_offset) \ { \ - vdup_b_x(v42, -input_offset); \ - vdup_b_x(v43, -input_offset); \ - vdup_b_x(v44, -input_offset); \ - vdup_b_x(v45, -input_offset); \ - vdup_b_x(v46, -input_offset); \ + vdup_b_x(INPUT_3_0, -input_offset); \ + vdup_b_x(INPUT_3_1, -input_offset); \ + vdup_b_x(INPUT_3_2, -input_offset); \ + vdup_b_x(INPUT_3_3, -input_offset); \ + vdup_b_x(INPUT_3_4, -input_offset); \ } #define PAD_ROW_4(input_offset) \ { \ - vdup_b_x(v47, -input_offset); \ - vdup_b_x(v48, -input_offset); \ - vdup_b_x(v49, -input_offset); \ - vdup_b_x(v50, -input_offset); \ - vdup_b_x(v51, -input_offset); \ + vdup_b_x(INPUT_4_0, -input_offset); \ + vdup_b_x(INPUT_4_1, -input_offset); \ + vdup_b_x(INPUT_4_2, -input_offset); \ + vdup_b_x(INPUT_4_3, -input_offset); \ + vdup_b_x(INPUT_4_4, -input_offset); \ } #define LOAD_ROW_0(p_input, input_width, in_y, in_x) \ { \ const int8_t* p_row = p_input + (in_y[0] * input_width); \ - vdup_b_x(v27, *(p_row + in_x[0])); \ - vdup_b_x(v28, *(p_row + in_x[1])); \ - vdup_b_x(v29, *(p_row + in_x[2])); \ - vdup_b_x(v30, *(p_row + in_x[3])); \ - vdup_b_x(v31, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_0_0, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_0_1, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_0_2, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_0_3, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_0_4, *(p_row + in_x[4])); \ } #define LOAD_ROW_1(p_input, input_width, in_y, in_x) \ { \ const int8_t* p_row = p_input + (in_y[1] * input_width); \ - vdup_b_x(v32, *(p_row + in_x[0])); \ - vdup_b_x(v33, *(p_row + in_x[1])); \ - vdup_b_x(v34, *(p_row + in_x[2])); \ - vdup_b_x(v35, *(p_row + in_x[3])); \ - vdup_b_x(v36, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_1_0, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_1_1, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_1_2, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_1_3, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_1_4, *(p_row + in_x[4])); \ } #define LOAD_ROW_2(p_input, input_width, in_y, in_x) \ { \ const int8_t* p_row = p_input + (in_y[2] * input_width); \ - vdup_b_x(v37, *(p_row + in_x[0])); \ - vdup_b_x(v38, *(p_row + in_x[1])); \ - vdup_b_x(v39, *(p_row + in_x[2])); \ - vdup_b_x(v40, *(p_row + in_x[3])); \ - vdup_b_x(v41, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_2_0, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_2_1, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_2_2, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_2_3, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_2_4, *(p_row + in_x[4])); \ } #define LOAD_ROW_3(p_input, input_width, in_y, in_x) \ { \ const int8_t* p_row = p_input + (in_y[3] * input_width); \ - vdup_b_x(v42, *(p_row + in_x[0])); \ - vdup_b_x(v43, *(p_row + in_x[1])); \ - vdup_b_x(v44, *(p_row + in_x[2])); \ - vdup_b_x(v45, *(p_row + in_x[3])); \ - vdup_b_x(v46, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_3_0, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_3_1, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_3_2, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_3_3, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_3_4, *(p_row + in_x[4])); \ } #define LOAD_ROW_4(p_input, input_width, in_y, in_x) \ { \ const int8_t* p_row = p_input + (in_y[4] * input_width); \ - vdup_b_x(v47, *(p_row + in_x[0])); \ - vdup_b_x(v48, *(p_row + in_x[1])); \ - vdup_b_x(v49, *(p_row + in_x[2])); \ - vdup_b_x(v50, *(p_row + in_x[3])); \ - vdup_b_x(v51, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_4_0, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_4_1, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_4_2, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_4_3, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_4_4, *(p_row + in_x[4])); \ } #define H_PAD_OR_LOAD_ROW_0(p_input, input_width, input_offset, in_y, in_x) \ @@ -178,29 +239,29 @@ } else { \ const int8_t* p_row = p_input + (in_y[0] * input_width); \ if (in_x[0] < 0 || in_x[0] >= input_width) { \ - vdup_b_x(v27, -input_offset); \ + vdup_b_x(INPUT_0_0, -input_offset); \ } else { \ - vdup_b_x(v27, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_0_0, *(p_row + in_x[0])); \ } \ if (in_x[1] < 0 || in_x[1] >= input_width) { \ - vdup_b_x(v28, -input_offset); \ + vdup_b_x(INPUT_0_1, -input_offset); \ } else { \ - vdup_b_x(v28, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_0_1, *(p_row + in_x[1])); \ } \ if (in_x[2] < 0 || in_x[2] >= input_width) { \ - vdup_b_x(v29, -input_offset); \ + vdup_b_x(INPUT_0_2, -input_offset); \ } else { \ - vdup_b_x(v29, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_0_2, *(p_row + in_x[2])); \ } \ if (in_x[3] < 0 || in_x[3] >= input_width) { \ - vdup_b_x(v30, -input_offset); \ + vdup_b_x(INPUT_0_3, -input_offset); \ } else { \ - vdup_b_x(v30, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_0_3, *(p_row + in_x[3])); \ } \ if (in_x[4] < 0 || in_x[4] >= input_width) { \ - vdup_b_x(v31, -input_offset); \ + vdup_b_x(INPUT_0_4, -input_offset); \ } else { \ - vdup_b_x(v31, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_0_4, *(p_row + in_x[4])); \ } \ } @@ -210,29 +271,29 @@ } else { \ const int8_t* p_row = p_input + (in_y[1] * input_width); \ if (in_x[0] < 0 || in_x[0] >= input_width) { \ - vdup_b_x(v32, -input_offset); \ + vdup_b_x(INPUT_1_0, -input_offset); \ } else { \ - vdup_b_x(v32, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_1_0, *(p_row + in_x[0])); \ } \ if (in_x[1] < 0 || in_x[1] >= input_width) { \ - vdup_b_x(v33, -input_offset); \ + vdup_b_x(INPUT_1_1, -input_offset); \ } else { \ - vdup_b_x(v33, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_1_1, *(p_row + in_x[1])); \ } \ if (in_x[2] < 0 || in_x[2] >= input_width) { \ - vdup_b_x(v34, -input_offset); \ + vdup_b_x(INPUT_1_2, -input_offset); \ } else { \ - vdup_b_x(v34, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_1_2, *(p_row + in_x[2])); \ } \ if (in_x[3] < 0 || in_x[3] >= input_width) { \ - vdup_b_x(v35, -input_offset); \ + vdup_b_x(INPUT_1_3, -input_offset); \ } else { \ - vdup_b_x(v35, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_1_3, *(p_row + in_x[3])); \ } \ if (in_x[4] < 0 || in_x[4] >= input_width) { \ - vdup_b_x(v36, -input_offset); \ + vdup_b_x(INPUT_1_4, -input_offset); \ } else { \ - vdup_b_x(v36, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_1_4, *(p_row + in_x[4])); \ } \ } @@ -242,29 +303,29 @@ } else { \ const int8_t* p_row = p_input + (in_y[2] * input_width); \ if (in_x[0] < 0 || in_x[0] >= input_width) { \ - vdup_b_x(v37, -input_offset); \ + vdup_b_x(INPUT_2_0, -input_offset); \ } else { \ - vdup_b_x(v37, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_2_0, *(p_row + in_x[0])); \ } \ if (in_x[1] < 0 || in_x[1] >= input_width) { \ - vdup_b_x(v38, -input_offset); \ + vdup_b_x(INPUT_2_1, -input_offset); \ } else { \ - vdup_b_x(v38, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_2_1, *(p_row + in_x[1])); \ } \ if (in_x[2] < 0 || in_x[2] >= input_width) { \ - vdup_b_x(v39, -input_offset); \ + vdup_b_x(INPUT_2_2, -input_offset); \ } else { \ - vdup_b_x(v39, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_2_2, *(p_row + in_x[2])); \ } \ if (in_x[3] < 0 || in_x[3] >= input_width) { \ - vdup_b_x(v40, -input_offset); \ + vdup_b_x(INPUT_2_3, -input_offset); \ } else { \ - vdup_b_x(v40, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_2_3, *(p_row + in_x[3])); \ } \ if (in_x[4] < 0 || in_x[4] >= input_width) { \ - vdup_b_x(v41, -input_offset); \ + vdup_b_x(INPUT_2_4, -input_offset); \ } else { \ - vdup_b_x(v41, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_2_4, *(p_row + in_x[4])); \ } \ } @@ -274,29 +335,29 @@ } else { \ const int8_t* p_row = p_input + (in_y[3] * input_width); \ if (in_x[0] < 0 || in_x[0] >= input_width) { \ - vdup_b_x(v42, -input_offset); \ + vdup_b_x(INPUT_3_0, -input_offset); \ } else { \ - vdup_b_x(v42, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_3_0, *(p_row + in_x[0])); \ } \ if (in_x[1] < 0 || in_x[1] >= input_width) { \ - vdup_b_x(v43, -input_offset); \ + vdup_b_x(INPUT_3_1, -input_offset); \ } else { \ - vdup_b_x(v43, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_3_1, *(p_row + in_x[1])); \ } \ if (in_x[2] < 0 || in_x[2] >= input_width) { \ - vdup_b_x(v44, -input_offset); \ + vdup_b_x(INPUT_3_2, -input_offset); \ } else { \ - vdup_b_x(v44, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_3_2, *(p_row + in_x[2])); \ } \ if (in_x[3] < 0 || in_x[3] >= input_width) { \ - vdup_b_x(v45, -input_offset); \ + vdup_b_x(INPUT_3_3, -input_offset); \ } else { \ - vdup_b_x(v45, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_3_3, *(p_row + in_x[3])); \ } \ if (in_x[4] < 0 || in_x[4] >= input_width) { \ - vdup_b_x(v46, -input_offset); \ + vdup_b_x(INPUT_3_4, -input_offset); \ } else { \ - vdup_b_x(v46, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_3_4, *(p_row + in_x[4])); \ } \ } @@ -306,29 +367,29 @@ } else { \ const int8_t* p_row = p_input + (in_y[4] * input_width); \ if (in_x[0] < 0 || in_x[0] >= input_width) { \ - vdup_b_x(v47, -input_offset); \ + vdup_b_x(INPUT_4_0, -input_offset); \ } else { \ - vdup_b_x(v47, *(p_row + in_x[0])); \ + vdup_b_x(INPUT_4_0, *(p_row + in_x[0])); \ } \ if (in_x[1] < 0 || in_x[1] >= input_width) { \ - vdup_b_x(v48, -input_offset); \ + vdup_b_x(INPUT_4_1, -input_offset); \ } else { \ - vdup_b_x(v48, *(p_row + in_x[1])); \ + vdup_b_x(INPUT_4_1, *(p_row + in_x[1])); \ } \ if (in_x[2] < 0 || in_x[2] >= input_width) { \ - vdup_b_x(v49, -input_offset); \ + vdup_b_x(INPUT_4_2, -input_offset); \ } else { \ - vdup_b_x(v49, *(p_row + in_x[2])); \ + vdup_b_x(INPUT_4_2, *(p_row + in_x[2])); \ } \ if (in_x[3] < 0 || in_x[3] >= input_width) { \ - vdup_b_x(v50, -input_offset); \ + vdup_b_x(INPUT_4_3, -input_offset); \ } else { \ - vdup_b_x(v50, *(p_row + in_x[3])); \ + vdup_b_x(INPUT_4_3, *(p_row + in_x[3])); \ } \ if (in_x[4] < 0 || in_x[4] >= input_width) { \ - vdup_b_x(v51, -input_offset); \ + vdup_b_x(INPUT_4_4, -input_offset); \ } else { \ - vdup_b_x(v51, *(p_row + in_x[4])); \ + vdup_b_x(INPUT_4_4, *(p_row + in_x[4])); \ } \ } @@ -371,15 +432,15 @@ { \ vld_w_x_m(v60, swizzled_bias_data); \ adwinit_v(v60, v60); \ - adwconv_vxv(v60, v27, cmds, v0); \ - adwconv_vxv(v60, v30, cmds, v3); \ - adwconv_vxv(v60, v33, cmds, v6); \ - adwconv_vxv(v60, v36, cmds, v9); \ - adwconv_vxv(v60, v39, cmds, v12); \ - adwconv_vxv(v60, v42, cmds, v15); \ - adwconv_vxv(v60, v45, cmds, v18); \ - adwconv_vxv(v60, v48, cmds, v21); \ - vdwconv_vxv(v60, v51, cmds, v24); \ + adwconv_vxv(v60, INPUT_0_0, cmds, FLT_0_0); \ + adwconv_vxv(v60, INPUT_0_1, cmds, FLT_0_1); \ + adwconv_vxv(v60, INPUT_0_2, cmds, FLT_0_2); \ + adwconv_vxv(v60, INPUT_0_3, cmds, FLT_0_3); \ + adwconv_vxv(v60, INPUT_0_4, cmds, FLT_0_4); \ + adwconv_vxv(v60, INPUT_3_0, cmds, FLT_3_0); \ + adwconv_vxv(v60, INPUT_3_3, cmds, FLT_3_3); \ + adwconv_vxv(v60, INPUT_3_4, cmds, FLT_HOLE); \ + vdwconv_vxv(v60, INPUT_4_2, cmds, FLT_4_2); \ } #define OUTPUT(output_activation_min, output_activation_max, output_offset, \ @@ -459,42 +520,42 @@ } } } + const int8_t* p_flt_0 = swizzled_filter_data.get() + (0 * filter_width * 24); const int8_t* p_flt_1 = swizzled_filter_data.get() + (1 * filter_width * 24); const int8_t* p_flt_2 = swizzled_filter_data.get() + (2 * filter_width * 24); const int8_t* p_flt_3 = swizzled_filter_data.get() + (3 * filter_width * 24); const int8_t* p_flt_4 = swizzled_filter_data.get() + (4 * filter_width * 24); - vld_b_l_xx(v0, p_flt_0 + (0 * 24), 24); - vld_b_l_xx(v1, p_flt_0 + (1 * 24), 24); - vld_b_l_xx(v2, p_flt_0 + (2 * 24), 24); - vld_b_l_xx(v3, p_flt_0 + (3 * 24), 24); - vld_b_l_xx(v4, p_flt_0 + (4 * 24), 24); + vld_b_lp_xx(FLT_0_0, p_flt_0, 24); + vld_b_lp_xx(FLT_0_1, p_flt_0, 24); + vld_b_lp_xx(FLT_0_2, p_flt_0, 24); + vld_b_lp_xx(FLT_0_3, p_flt_0, 24); + vld_b_lp_xx(FLT_0_4, p_flt_0, 24); - vld_b_l_xx(v5, p_flt_1 + (0 * 24), 24); - vld_b_l_xx(v6, p_flt_1 + (1 * 24), 24); - vld_b_l_xx(v7, p_flt_1 + (2 * 24), 24); - vld_b_l_xx(v8, p_flt_1 + (3 * 24), 24); - vld_b_l_xx(v9, p_flt_1 + (4 * 24), 24); + vld_b_lp_xx(FLT_1_0, p_flt_1, 24); + vld_b_lp_xx(FLT_1_1, p_flt_1, 24); + vld_b_lp_xx(FLT_1_2, p_flt_1, 24); + vld_b_lp_xx(FLT_1_3, p_flt_1, 24); + vld_b_lp_xx(FLT_1_4, p_flt_1, 24); - vld_b_l_xx(v10, p_flt_2 + (0 * 24), 24); - vld_b_l_xx(v11, p_flt_2 + (1 * 24), 24); - vld_b_l_xx(v12, p_flt_2 + (2 * 24), 24); - vld_b_l_xx(v13, p_flt_2 + (3 * 24), 24); - vld_b_l_xx(v14, p_flt_2 + (4 * 24), 24); + vld_b_lp_xx(FLT_2_0, p_flt_2, 24); + vld_b_lp_xx(FLT_2_1, p_flt_2, 24); + vld_b_lp_xx(FLT_2_2, p_flt_2, 24); + vld_b_lp_xx(FLT_2_3, p_flt_2, 24); + vld_b_lp_xx(FLT_2_4, p_flt_2, 24); - vld_b_l_xx(v15, p_flt_3 + (0 * 24), 24); - vld_b_l_xx(v16, p_flt_3 + (1 * 24), 24); - vld_b_l_xx(v17, p_flt_3 + (2 * 24), 24); - vld_b_l_xx(v18, p_flt_3 + (3 * 24), 24); - vld_b_l_xx(v19, p_flt_3 + (4 * 24), 24); + vld_b_lp_xx(FLT_3_0, p_flt_3, 24); + vld_b_lp_xx(FLT_3_1, p_flt_3, 24); + vld_b_lp_xx(FLT_3_2, p_flt_3, 24); + vld_b_lp_xx(FLT_3_3, p_flt_3, 24); + vld_b_lp_xx(FLT_3_4, p_flt_3, 24); - vld_b_l_xx(v20, p_flt_4 + (0 * 24), 24); - vld_b_l_xx(v21, p_flt_4 + (1 * 24), 24); - vld_b_l_xx(v22, p_flt_4 + (2 * 24), 24); - vld_b_l_xx(v23, p_flt_4 + (3 * 24), 24); - vld_b_l_xx(v24, p_flt_4 + (4 * 24), 24); - vdup_b_x(v25, 0); - vdup_b_x(v26, 0); + vdup_b_x(FLT_HOLE, 0); + vld_b_lp_xx(FLT_4_0, p_flt_4, 24); + vld_b_lp_xx(FLT_4_1, p_flt_4, 24); + vld_b_lp_xx(FLT_4_2, p_flt_4, 24); + vld_b_lp_xx(FLT_4_3, p_flt_4, 24); + vld_b_lp_xx(FLT_4_4, p_flt_4, 24); union { vdwconv_u8_t dwconv; @@ -524,7 +585,7 @@ int8_t* local_output_data = output_data + out_channel; int in_y[5]; - int in_x[5]; + int in_x[7]; int out_y = 0; const int8_t* p_input = input_data; // Handle top row padding @@ -615,6 +676,109 @@ local_output_data, n_channels); local_output_data += output_depth; } + for (; out_x + 2 <= (output_width - pad_width); out_x += 2) { + const int in_x_origin = (out_x * stride_width) - pad_width; + + #pragma GCC unroll 7 + for (int i = 0; i < 7; ++i) { + in_x[i] = in_x_origin + (dilation_width_factor * i); + } + const int8_t* p_rows[5]; + #pragma GCC unroll 5 + for (int i = 0; i < 5; ++i) { + p_rows[i] = p_input + (in_y[i] * input_width); + } + + vdup_b_x(INPUT_0_0, *(p_rows[0] + in_x[0])); + vdup_b_x(INPUT_0_1, *(p_rows[0] + in_x[1])); + vdup_b_x(INPUT_0_2, *(p_rows[0] + in_x[2])); + vdup_b_x(INPUT_0_3, *(p_rows[0] + in_x[3])); + vdup_b_x(INPUT_0_4, *(p_rows[0] + in_x[4])); + + vdup_b_x(INPUT_1_0, *(p_rows[1] + in_x[0])); + vdup_b_x(INPUT_1_1, *(p_rows[1] + in_x[1])); + vdup_b_x(INPUT_1_2, *(p_rows[1] + in_x[2])); + vdup_b_x(INPUT_1_3, *(p_rows[1] + in_x[3])); + vdup_b_x(INPUT_1_4, *(p_rows[1] + in_x[4])); + + vdup_b_x(INPUT_2_0, *(p_rows[2] + in_x[0])); + vdup_b_x(INPUT_2_1, *(p_rows[2] + in_x[1])); + vdup_b_x(INPUT_2_2, *(p_rows[2] + in_x[2])); + vdup_b_x(INPUT_2_3, *(p_rows[2] + in_x[3])); + vdup_b_x(INPUT_2_4, *(p_rows[2] + in_x[4])); + + vdup_b_x(INPUT_3_0, *(p_rows[3] + in_x[0])); + vdup_b_x(INPUT_3_1, *(p_rows[3] + in_x[1])); + vdup_b_x(INPUT_3_2, *(p_rows[3] + in_x[2])); + vdup_b_x(INPUT_3_3, *(p_rows[3] + in_x[3])); + vdup_b_x(INPUT_3_4, *(p_rows[3] + in_x[4])); + + vdup_b_x(INPUT_4_0, *(p_rows[4] + in_x[0])); + vdup_b_x(INPUT_4_1, *(p_rows[4] + in_x[1])); + vdup_b_x(INPUT_4_2, *(p_rows[4] + in_x[2])); + vdup_b_x(INPUT_4_3, *(p_rows[4] + in_x[3])); + vdup_b_x(INPUT_4_4, *(p_rows[4] + in_x[4])); + + vld_w_x_m(v60, swizzled_bias_data); + adwinit_v(v60, v60); + adwconv_vxv(v60, INPUT_0_0, cmds, FLT_0_0); + adwconv_vxv(v60, INPUT_0_1, cmds, FLT_0_1); + adwconv_vxv(v60, INPUT_0_2, cmds, FLT_0_2); + adwconv_vxv(v60, INPUT_0_3, cmds, FLT_0_3); + adwconv_vxv(v60, INPUT_0_4, cmds, FLT_0_4); + adwconv_vxv(v60, INPUT_3_0, cmds, FLT_3_0); + adwconv_vxv(v60, INPUT_3_3, cmds, FLT_3_3); + adwconv_vxv(v60, INPUT_3_4, cmds, FLT_HOLE); + vdwconv_vxv(v60, INPUT_4_2, cmds, FLT_4_2); + vmv_v(INPUT_0_0, v60); + vmv_v(INPUT_1_0, v61); + vmv_v(INPUT_2_0, v62); + vmv_v(INPUT_0_1, v63); + + vdup_b_x(INPUT_3_0, *(p_rows[3] + in_x[5])); + vdup_b_x(INPUT_3_1, *(p_rows[3] + in_x[6])); + + vmv_v(INPUT_4_0, INPUT_4_2); + vmv_v(INPUT_4_1, INPUT_4_3); + vmv_v(INPUT_4_2, INPUT_4_4); + vdup_b_x(INPUT_4_3, *(p_rows[4] + in_x[5])); + vdup_b_x(INPUT_4_4, *(p_rows[4] + in_x[6])); + + vld_w_x_m(v60, swizzled_bias_data); + adwinit_v(v60, v60); + adwconv_vxv(v60, INPUT_0_2, cmds, FLT_0_0); + adwconv_vxv(v60, INPUT_0_3, cmds, FLT_0_1); + + vmv_v(INPUT_0_2, INPUT_0_0); + vmv_v(INPUT_1_2, INPUT_1_0); + vmv_v(INPUT_2_2, INPUT_2_0); + vmv_v(INPUT_0_3, INPUT_0_1); + + vdup_b_x(INPUT_0_0, *(p_rows[0] + in_x[5])); + vdup_b_x(INPUT_0_1, *(p_rows[0] + in_x[6])); + vdup_b_x(INPUT_1_0, *(p_rows[1] + in_x[5])); + vdup_b_x(INPUT_1_1, *(p_rows[1] + in_x[6])); + vdup_b_x(INPUT_2_0, *(p_rows[2] + in_x[5])); + vdup_b_x(INPUT_2_1, *(p_rows[2] + in_x[6])); + + adwconv_vxv(v60, INPUT_0_4, cmds, FLT_0_2); + adwconv_vxv(v60, INPUT_0_0, cmds, FLT_0_3); + adwconv_vxv(v60, INPUT_0_1, cmds, FLT_0_4); + adwconv_vxv(v60, INPUT_3_2, cmds, FLT_3_0); + adwconv_vxv(v60, INPUT_3_0, cmds, FLT_3_3); + adwconv_vxv(v60, INPUT_3_4, cmds, FLT_HOLE); + vdwconv_vxv(v60, INPUT_4_2, cmds, FLT_4_2); + INT32_TO_INT8_OUTPUT_PIPELINE_INPLACE2( + v60, INPUT_0_2, v52, v56, output_activation_min, + output_activation_max, output_offset + ); + vsraqs_b_vx(INPUT_0_2, INPUT_0_2, 0); + vst_b_l_xx(INPUT_0_2, local_output_data, n_channels); + local_output_data += output_depth; + vsraqs_b_vx(v60, v60, 0); + vst_b_l_xx(v60, local_output_data, n_channels); + local_output_data += output_depth; + } for (; out_x < (output_width - pad_width); ++out_x) { const int in_x_origin = (out_x * stride_width) - pad_width; @@ -736,6 +900,62 @@ #undef PAD_ROW_4 #undef CALCULATE_IN_X #undef CALCULATE_IN_Y +#undef INPUT_0_0 +#undef INPUT_0_1 +#undef INPUT_0_2 +#undef INPUT_0_3 +#undef INPUT_0_4 +#undef INPUT_1_0 +#undef INPUT_1_1 +#undef INPUT_1_2 +#undef INPUT_1_3 +#undef INPUT_1_4 +#undef INPUT_2_0 +#undef INPUT_2_1 +#undef INPUT_2_2 +#undef INPUT_2_3 +#undef INPUT_2_4 +#undef INPUT_3_0 +#undef INPUT_3_1 +#undef INPUT_3_2 +#undef INPUT_3_3 +#undef INPUT_3_4 +#undef INPUT_4_0 +#undef INPUT_4_1 +#undef INPUT_4_2 +#undef INPUT_4_3 +#undef INPUT_4_4 +#undef INPUT_0_5 +#undef INPUT_1_5 +#undef INPUT_2_5 +#undef INPUT_3_5 +#undef INPUT_4_5 +#undef FLT_0_0 +#undef FLT_0_1 +#undef FLT_0_2 +#undef FLT_0_3 +#undef FLT_0_4 +#undef FLT_1_0 +#undef FLT_1_1 +#undef FLT_1_2 +#undef FLT_1_3 +#undef FLT_1_4 +#undef FLT_2_0 +#undef FLT_2_1 +#undef FLT_2_2 +#undef FLT_2_3 +#undef FLT_2_4 +#undef FLT_3_0 +#undef FLT_3_1 +#undef FLT_3_2 +#undef FLT_3_3 +#undef FLT_3_4 +#undef FLT_HOLE +#undef FLT_4_0 +#undef FLT_4_1 +#undef FLT_4_2 +#undef FLT_4_3 +#undef FLT_4_4 void ConvPerChannelD1( const tflite::ConvParams& params, const int32_t* output_multiplier,