Kali Uday Balleda
Kali Uday Balleda
> Debugging this is very tricky, changing even the smallest things (e.g. setting specific breakpoints in GDB) cause the reproducer to fail, even though the generated code looks identical. >...
> > it is better to synchronize and evict > > That will kill performance. > > What specifically is the effect of calling `zeContextEvictMemory` right before `zeMemFree`? Just calling...
This issue impacts hungarian-hip benchmark as well.
@maleadt https://github.com/JuliaGPU/oneAPI.jl/pull/467 please confirm if this works post tool chain update.
@luraess when we tried to import IGG following errors are encountered, based on the log IGG may need to be updated to use CUDA 5.0.0. Can you make a release...
> > @luraess, Can you give us a general idea what it takes to support oneAPI.jl in IGG? > > I can see that this message has not been answered...
> @kballeda I added more LAPACK routines and updated `support.toml` for the wrappers of your PR #348. @amontoison Thanks for adding ormqr, unmqr, orgqr, ungqr primitives to the list. It's...
> I will add some tests asap. What can we do to fix CI? @amontoison https://github.com/JuliaGPU/oneAPI.jl/issues/357 - Discussion on this topic
@maleadt I have filed a Git issue https://github.com/JuliaGPU/oneAPI.jl/issues/357 . This issue is observed as part of our oneAPI.jl CI flow. Can you please take a look ?
> @kballeda Can I help on something here? I would like to interface the additional LAPACK routines `orgqr` and `ormqr`. Sure Thank you !