-
Notifications
You must be signed in to change notification settings - Fork 75
[UT] Fix test_precise_math
#2776
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Conversation
victor-eds
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I may be missing some context (can you add PR description, please?), but I see our test is indeed failing and we are playing tricks to get it to pass. Would llvm/llvm-project#88222 (comment) fix the test without needing to change it?
Added PR description.
That is a different issue, we currently need to use cpu result as reference, or else we are comparing with the imprecise result. |
third_party/intel/lib/TritonIntelGPUToLLVM/ElementwiseOpToLLVM.cpp
Outdated
Show resolved
Hide resolved
victor-eds
left a comment
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM keeping as is after offline conversation
Signed-off-by: Whitney Tsang <[email protected]>
Signed-off-by: Whitney Tsang <[email protected]>
f50c16a to
fc59ae5
Compare
Before this PR, precise sqrt is lowered to
__imf_sqrtf, which is not precise.__imf_sqrt_rnis added in the new device library updated in #2774.This PR uses
__imf_sqrt_rnto get the precise sqrt implemented.