Fix adv benchmarks.yaml: test args do not match kernel compile-time constants#247
Merged
zjin-lcf merged 1 commit intoJun 6, 2026
Merged
Conversation
The advCubatureHex3D kernel hardcodes p_Nq=8 and p_cubNq=16. These require N=7 (Nq=8) and cubN=15 (cubNq=16) respectively. Previous args [16,16,16] gave Nq=17 != p_Nq=8, causing shared memory out-of-bounds reads and incorrect results on all platforms. Verified on AMD MI300X (gfx942): adv 7 15 16 1 -> PASS (Nelements=1,4,16,64 all pass) adv 16 16 16 1 -> FAIL (index 8192: 0.691294 != 0.583546) Fixes: ORNL#246
Collaborator
|
Thank you for the pull request. |
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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
Fixes incorrect test arguments for the
advbenchmark that cause correctness failures on all GPU platforms (CUDA, HIP, SYCL, OpenMP).Fixes #246
Root Cause
The
advCubatureHex3Dkernel insrc/adv-{cuda,hip}/adv.hhardcodes compile-time constants:p_Nq = 8→ requires N = 7 (Nq = N+1 = 8)p_cubNq = 16→ requires cubN = 15 (cubNq = 16)The previous args
[�\, �\, �\]gave Nq=17 ≠ p_Nq=8, causing shared memory out-of-bounds reads and wrong results on all platforms.Change
Verification (AMD MI300X gfx942, ROCm 7.2)
adv 16 16 16 1adv 7 15 1 1adv 7 15 4 1adv 7 15 16 1adv 7 15 64 1Failure occurs on all platforms (CUDA/HIP/SYCL/OpenMP) — not AMD-specific.