feat: redirect Triton cache into magi managed cache tree - #53
Merged
Conversation
Previously only TORCHINDUCTOR_CACHE_DIR was set, leaving Triton to use
its default ~/.triton/cache which only stores cubin files. By also setting
TRITON_CACHE_DIR to {cache_root_dir}/triton_cache, all Triton compilation
artifacts (.source, .ptx, .json, .cubin) are preserved alongside the
Inductor cache, enabling post-hoc analysis of autotuning decisions.
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
TRITON_CACHE_DIRalongsideTORCHINDUCTOR_CACHE_DIRin_compilation_context(), pointing to{cache_root_dir}/triton_cache.~/.triton/cachewhich only stores.cubinfiles. With this change, all compilation artifacts (.source,.ptx,.json,.cubin) are preserved in the managed cache tree.Motivation
When investigating non-determinism in compiled VAE models, we needed to inspect Triton kernel source code and autotuning metadata. However,
magi_compileronly stored.cubinfiles becauseTRITON_CACHE_DIRwas not redirected into the managed tree. This PR fixes the gap.Changes
config.py: Addtriton_cache_dump_path()returning{cache_root_dir}/triton_cache_api.py: Import and usetriton_cache_dump_path, addTRITON_CACHE_DIRto the env patch dictTest plan
triton_cache/directory is created underMAGI_COMPILE_CACHE_ROOT_DIRafter compilation.source,.ptx,.jsonartifacts are present (not just.cubin)