Conversation
Bump SimpleModel dim 128→1024 and seq_len 256→512 so GPU compute dominates over kernel-launch overhead, eliminating flaky ratio assertions on fast GPUs (B300/H100).
This branch has not been deployed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Bump SimpleModel dim 128→1024 and seq_len 256→512 in
test_simple_model_timing_class_function_instance_methodso GPU compute dominates over kernel-launch overhead.Problem: With dim=128, timings are ~0.2ms where kernel launch jitter causes 4-6x ratio swings, making the
max/min < 1.2assertion flaky on fast GPUs (B300/H100).Fix: Larger model → ~0.24ms median with stdev ~0.005ms → ratio stable at ~1.02.
Tested on B300 (SM103): all 4 entry points converge to 0.238-0.244ms median.