Seungman Han
Seungman Han
Add max performance results max float16 op_name | dtype | size | dim | model | dir | ROCm pytorch | MIOpen HIP | Improvement -- | -- | --...
> @seungmanhan A couple of questions before we continue with review. The ReduceExtreme operations look similar to the existing TensorReduce. What prevents or hinders us from extending the existing TensorReduce...
The driver is being modified while resolving merge conflicts.
@junliume Even if I do a clean build and make check, i cannot determine the cause of the failure in Jenkins - HIP Package. Please check the reason of failure.
> > @amberhassaan Have your review comments been addressed ? > > No, not quite. I haven't had a chance to leave a detailed review, but I'd like @seungmanhan to...
@junliume Can you please confirm why the github action is failing?
@junliume When I add miopen::IsUnset(MIOPEN_ENV(MIOPEN_TEST_ALL) in gtest, GitHub action fails. Can you please check the log [363dbe2](https://github.com/ROCm/MIOpen/pull/2883/commits/363dbe22159b8448f2e1384211e6329a293b59a5)?
> @seungmanhan > > > @junliume When I add miopen::IsUnset(MIOPEN_ENV(MIOPEN_TEST_ALL) in gtest, GitHub action fails. Can you please check the log [363dbe2](https://github.com/ROCm/MIOpen/pull/2883/commits/363dbe22159b8448f2e1384211e6329a293b59a5)? > > You can check locally > >...
@junliume Can you find out why PRs keep failing at the build stage? (https://github.com/ROCm/MIOpen/pull/2883/commits/6255e87b863630c107bb830b1c9f9d5f69e58ef4)
@junliume Can you confirm what is being tested in 'Jenkins - Fp32 Hip All gfx90a' or send me the script?