[rocm6.4_internal_testing] Replaced ROCm specific skips to generalized conditions #2100
Conversation
|
Jenkins build for d98fc67a7dfdce9665b6c5a05acdaeea8ea82a67 commit finished as FAILURE |
| @unittest.skipIf(IS_FBCODE, "Not yet runnable in fbcode") | ||
| @unittest.skipIf(not SM80OrLater, "bfloat16 only supported in sm80+") | ||
| @unittest.skipIf(not PLATFORM_SUPPORTS_FLASH_ATTENTION, "Some archs don't support SDPA") |
There was a problem hiding this comment.
Does this test fail when using math backend on CUDA as well?
There was a problem hiding this comment.
I observed that if we force math backend in MI300 and even at A100 for exactly this test, it fails.
see: https://github.com/ROCm/frameworks-internal/issues/11952
there you can also find an upstream pytorch issue I filed..
But the logic here is clear:
- If FA (goes into AOTriton) is not supported, then code goes into math backend and fails and there's nothing we can do about it (until upstream fixes math backend) -> must skip the test
- If FA is enabled (this depends on the architecture and/or if TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL is set, see:
| @@ -1361,16 +1361,6 @@ def printErrors(self) -> None: | |||
| IS_ARM64 = platform.machine() in ('arm64', 'aarch64') | |||
| IS_S390X = platform.machine() == "s390x" | |||
|
|
|||
| NAVI32_ARCH = "gfx1101" | |||
|
|
|||
| def is_navi_arch(): | |||
There was a problem hiding this comment.
I believe you have checked nobody is using this function.
There was a problem hiding this comment.
Yeah, it was introduced by me in previous cherry-pick
pruthvistony
deleted the
iupaikov_sdpa_skips_generalization_rocm6.4_internal_testing
branch
|
!cherry-pick --onto release/2.7 |
|
!cherry-pick --onto rocm6.5_internal_testing |
|
Created branch autogenerated/release/2.7_cherry-pick_pr-2100 and #2126. It contains a merge conflict. Please resolve it |
|
Can't perform the cherry-pick keyword: unexpected error |
|
Created branch autogenerated/rocm6.5_internal_testing_cherry-pick_pr-2100 and #2127. It contains a merge conflict. Please resolve it |
|
Can't perform the cherry-pick keyword: unexpected error |
…d conditions (#2100) It's an internal change, removes unnecessary arch mentions and generalizes conditions. Needs to be cherry-picked into rocm6.5_internal_testing and release/2.7 .and now also into release/2.6
… Replaced ROCm specific skips to generalized conditions (#2127) Cherry-pick of #2100 Need to resolve conflicts --------- Co-authored-by: iupaikov-amd <[email protected]>
…m specific skips to generalized conditions (#2126) Cherry-pick of #2100 Need to resolve conflicts --------- Co-authored-by: iupaikov-amd <[email protected]>
|
!cherry-pick --onto release/2.6 |
|
Created branch autogenerated/release/2.6_cherry-pick_pr-2100 and #2260. It contains a merge conflict. Please resolve it |
|
!cherry-pick --onto release/2.5 |
1 similar comment
|
!cherry-pick --onto release/2.5 |
|
Created branch autogenerated/release/2.5_cherry-pick_pr-2100 and #2261. It contains a merge conflict. Please resolve it |
|
Created branch autogenerated/release/2.5_cherry-pick_pr-2100 and #2262. It contains a merge conflict. Please resolve it |
It's an internal change, removes unnecessary arch mentions and generalizes conditions.
Needs to be cherry-picked into rocm6.5_internal_testing and release/2.7
Cherry-picked to release/2.6 branch via #2260
Cherry-picked to release/2.5 branch via #2261
Cherry-picked to release/2.5 branch via #2262