OpenXLA Discuss

Hello everyone, I am seeking help with the review of PR-16585 that introduces support for the

Hello, We've seen that enabling xla_gpu_enable_triton_gemm will severly (~50-100x) impact

OpenXLA Community Meeting - Monthly Join with Google Meet – OpenXLA Community Meeting Notes This

OpenXLA Community Meeting - Monthly Join with Google Meet – OpenXLA Community Meeting Notes This

On Fri, Sep 6, 2024 at 12:54 PM Peter Hawkins <phaw...@google.com> wrote: Hmm.. indeed we'

Hi everyone, Quick reminder that we have our monthly OpenXLA Community meeting happening tomorrow,

Dear StableHLO Community, The current spec for scatter op in StableHLO - https://github.com/openxla/

Hello everyone, I wanted to share the RFC for adding support for the f8E4M3 and f8E3M4 types in

OpenXLA Community Meeting - Monthly Join with Google Meet – OpenXLA Community Meeting Notes This

Hi Kevin, Thanks for your response! Yes let's continue the discussion on the GH issue. Thank you,

Hi all, Now that AMD has set up their CI, I was wondering if it would be possible to find

OpenXLA Community Meeting - Monthly Join with Google Meet – OpenXLA Community Meeting Notes This

OpenXLA Community Meeting - Monthly Join with Google Meet – OpenXLA Community Meeting Notes This

Hello everyone! We are glad to share a new RFC for adding quantization support for CPU in OpenXLA.

Thanks for the answer ! Do you have minimal reproducers for the bugs? Not really unfortunately. The

+Vadym Matsishevskyi, is this run-time increase an expected side-effect of using hermetic Python? Is

Hi OpenXLA Community, Just a friendly reminder to please complete the OpenXLA documentation feedback

Hi everyone, As mentioned during the community meeting, here is the RFC on adding support for int2

cc liuyuanqi...@bytedance.com who authored the MHLO canonicalization in https://github.com/

OpenXLA Community Meeting - Monthly Join with Google Meet – You have been invited by ajoshi@openxla.

Hi folks, Friendly reminder that our monthly OpenXLA meeting is happening tomorrow, June 18, at 9 am

OpenXLA Community Meeting - Monthly Join with Google Meet – You have been invited by Aditi Joshi to

Hello Axel I have just added an https://github.com/openxla/stablehlo/issues/2373 to address similar

Yes, that is the idea. It will disable them through the runtime at this point. This transform is

On a second thought we do not need modifications involving bitcast (step 3) and use the following

NOTE: External mailing list. +Frederik Gossen +Greg Olechwierowicz Makes sense to me, thanks! On Sun,

Thanks Sandeep for your reply. I guess dot_general is the only operator in the opset to have this

Thanks! Regards Mini Goel On Fri, May 31, 2024 at 12:11 PM Eugene Zhulenev <ezhu...@google.com

The generated MLIR python ones allow one to define handlers for attribute types so that you should be

Hello Everyone LowerComplex mhlo pass uses applyPatternsAndFoldGreedily. The doc says /// Also