Repository navigation
Conversation
Introduce backend dispatch, native CPU execution, and cached launch arguments. Add autotuning for kernels and multi-kernel factories.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR adds an experimental CPU backend for running cuTile kernels on Linux hosts, including NumPy-compatible arrays and CPU PyTorch tensors. Select it with ct.set_backend("cpu"); custom backends can be provided through the Backend protocol. CPU launches are synchronous, and their OpenMP thread count can be configured with ct.backend_options.
The backend interecepts TileIR and compiles them through tileir-to-mlir and upstream MLIR passes. It caches compiled kernels, and reuses cached launch argument packing for supported inputs. The PR also adds ct.autotune, which benchmarks configurations per input layout and reuses the best choice for kernels and multi-kernel factories.
CPU samples and backend/compiler tests are included. The CPU backend requires a separate CUTILE_BUILD_VARIANT=cpu build and is experimental; performance and correctness should be evaluated for each workload.