| commit | 67555c0e7786cd4205fa5fabb6f22bd4149faee1 | [log] [tgz] |
|---|---|---|
| author | bjacob <benoitjacob@google.com> | Tue May 30 15:40:13 2023 -0400 |
| committer | GitHub <noreply@github.com> | Tue May 30 15:40:13 2023 -0400 |
| tree | b9eae846cb98892987dc018d934b7df64de5c0a5 | |
| parent | bd5174b66f7f17933ef6dcdf8d3bc89204dda410 [diff] |
Drop conditionals and configured headers from the ukernels build (#13834) This drop CMake `if()` and configured headers from the ukernels build. Instead: * The uarch-specific code paths are unconditionally added and then `#if`'d out inside C code based on compiler version macros. This simplifies the build ahead of multi-arch bitcode (#13804) and paves the way for Bazel support on par with CMake. * The `iree_uk_ssize_t` type is unconditionally `int64_t`, being more clearly branded a size/index type more than a "pointer-size" type. This wasn't completely clear until the recent bitcode push, but now that I'm looking at LLVM IR where `ptr` is an opaque type and there isn't even a concept of pointer size, while size/index have to use some explicitly sized integer type, this change reflects the reality of what this type is used for. It would be possible to make this `int32_t` instead of `int64_t` for best performance on 32-bit architectures but for now making it unconditionally `int64_t` is an easy first step and keeps what we were currently doing in standalone builds.
IREE (Intermediate Representation Execution Environment, pronounced as “eerie”) is an MLIR-based end-to-end compiler and runtime that lowers Machine Learning (ML) models to a unified IR that scales up to meet the needs of the datacenter and down to satisfy the constraints and special considerations of mobile and edge deployments.
See our website for project details, user guides, and instructions on building from source.
IREE is still in its early phase. We have settled down on the overarching infrastructure and are actively improving various software components as well as project logistics. It is still quite far from ready for everyday use and is made available without any support at the moment. With that said, we welcome any kind of feedback on any communication channels!
See our website for more information.
IREE is licensed under the terms of the Apache 2.0 License with LLVM Exceptions. See LICENSE for more information.