[FRONTEND] Add max_trials_per_task parameter to static_shape_tuning_pipeline #18159
+35
−6
Conversation
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
…ipeline This commit introduces the max_trials_per_task parameter to the static_shape_tuning_pipeline, giving users explicit control over the maximum number of trials allocated to each task during MetaSchedule tuning.
Hzfengsy
reviewed
Jul 24, 2025
| gpu_out = vm["main"](gpu_data, *gpu_params).numpy() | ||
| gpu_out = vm["main"](gpu_data, *gpu_params) | ||
|
|
||
| gpu_out = np.array([gpu_out[0].numpy()[0][j] for j in range(1000)]) |
There was a problem hiding this comment.
What's the purpose of this change?
There was a problem hiding this comment.
I have simplified the numpy conversion.
There was a problem hiding this comment.
After the new FFI was introduced, vm["main"](...) returns an object of <class 'tvm.ffi.container.Array'>, so a [0] needs to be added.
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Outdated suggestions cannot be applied.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.
This PR introduces the
max_trials_per_taskparameter to thestatic_shape_tuning_pipeline, giving users explicit control over the maximum number of trials allocated to each task during MetaSchedule tuning.Currently, the number of tuning trials per task per iteration is determined by
min(max_trials_per_iter=64, total_trials). This leads to potential issues when tuning on GPUs:When the user-specified
total_trialsis:(number_of_tasks * 64)but greater thanmax_trials_per_iter=64, ormax_trials_per_iter=64itself,Some TIR functions will not be tuned at all, and fail to be bound to threads properly.