Add HELION_AUTOTUNE_PRECOMPILE_REPS setting to run kernel multiple times during spawn precompile#1914
Open
yf225 wants to merge 1 commit intopytorch:mainfrom
Open
Add HELION_AUTOTUNE_PRECOMPILE_REPS setting to run kernel multiple times during spawn precompile#1914yf225 wants to merge 1 commit intopytorch:mainfrom
yf225 wants to merge 1 commit intopytorch:mainfrom
Conversation
…mes during spawn precompile Summary: When using spawn precompile mode, a kernel may compile and run once successfully but cause an illegal memory access (IMA) after repeated execution during benchmarking. Since benchmarking runs in the main process, this corrupts the GPU state and crashes the entire autotuning session. This adds a `autotune_precompile_reps` setting (env var `HELION_AUTOTUNE_PRECOMPILE_REPS`, default 1) that controls how many times the kernel is executed in the spawn subprocess during precompile. Setting it to e.g. 10 helps catch IMAs that only manifest after repeated runs, while keeping the failure isolated in the subprocess. Differential Revision: D99175421
meta-cla
bot
added
the
CLA Signed
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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:
When using spawn precompile mode, a kernel may compile and run once
successfully but cause an illegal memory access (IMA) after repeated
execution during benchmarking. Since benchmarking runs in the main
process, this corrupts the GPU state and crashes the entire autotuning
session.
This adds a
autotune_precompile_repssetting (env varHELION_AUTOTUNE_PRECOMPILE_REPS, default 1) that controls how manytimes the kernel is executed in the spawn subprocess during precompile.
Setting it to e.g. 10 helps catch IMAs that only manifest after
repeated runs, while keeping the failure isolated in the subprocess.
Differential Revision: D99175421