OpenXLA Discuss

1–30 of 155

Welcome to the OpenXLA Project community!

Here are other ways to get involved in the project:

Attend an OpenXLA Community Meeting
Subscribe to the OpenXLA Youtube Channel
Follow the OpenXLA Twitter
Contribute to OpenXLA on GitHub

0 selected

Steeve Morin, Florian Reichl2

Mar 11

Triton/XLA on ROCm has very bad performance

Hi Steve, I see you've also opened https://github.com/openxla/xla/issues/23574 and are getting

unread,

Triton/XLA on ROCm has very bad performance

Hi Steve, I see you've also opened https://github.com/openxla/xla/issues/23574 and are getting

Mar 11

Steeve Morin, … George Karpenkov7

Mar 3

Very bad performance when xla_gpu_enable_triton_gemm=true on MI210/MI300X

Hey folks, Jumping back on this. We have tried running the Triton implementation of Paged Attention

unread,

Very bad performance when xla_gpu_enable_triton_gemm=true on MI210/MI300X

Hey folks, Jumping back on this. We have tried running the Triton implementation of Paged Attention

Mar 3

Feb 20

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Mar 18, 2025 9am - 10am (PDT) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Mar 18, 2025 9am - 10am (PDT) (openxla-discuss@openxla.org)

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

Feb 20

Aditi Joshi, Chris Jones2

Feb 18

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Feb 18, 2025 9am - 9:45am (PST) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Feb 18, 2025 9am - 9:45am (PST) (openxla-discuss@openxla.org)

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

Feb 18

Feb 17

July 16, Tues, 9-10 am PT OpenXLA Monthly Community meeting

Friendly reminder that tomorrow morning at 9 am PT is our February OpenXLA Community meeting. Talk

unread,

July 16, Tues, 9-10 am PT OpenXLA Monthly Community meeting

Friendly reminder that tomorrow morning at 9 am PT is our February OpenXLA Community meeting. Talk

Feb 17

Feb 13

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Feb 18, 2025 9am - 10am (PST) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Feb 18, 2025 9am - 10am (PST) (openxla-discuss@openxla.org)

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

Feb 13

Jeff Huynh, Rahul Joshi5

Feb 13

Include token as part of the input/output tuple in all-gather and reduce-scatter

Hi Rahul, I would like to revisit this issue as we see all-reduce has support for tuple but reduce-

unread,

Include token as part of the input/output tuple in all-gather and reduce-scatter

Hi Rahul, I would like to revisit this issue as we see all-reduce has support for tuple but reduce-

Feb 13

Md Faijul Amin, … Alexander Belyaev13

Feb 11

XLA GPU MLIR Codegen

We can also chat over VC, if it is more convenient for you.

unread,

XLA GPU MLIR Codegen

We can also chat over VC, if it is more convenient for you.

Feb 11

Jan 14

tf.aliasing vs jax.buffer_donor

Hi folks. Is there a reason to use tf.aliasing instead of the higher level jax.buffer_donor ? It

unread,

tf.aliasing vs jax.buffer_donor

Hi folks. Is there a reason to use tf.aliasing instead of the higher level jax.buffer_donor ? It

Jan 14

Sandeep Dasgupta, Jan Pfeifer6

Jan 14

HLO Componentization: Progress update

Hello Everybody, This is a follow-up to the previous announcement about the deprecation of certain

unread,

HLO Componentization: Progress update

Hello Everybody, This is a follow-up to the previous announcement about the deprecation of certain

Jan 14

Jan 14

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Jan 14, 2025 9am - 9:30am (PST) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Jan 14, 2025 9am - 9:30am (PST) (openxla-discuss@openxla.org)

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

Jan 14

12/23/24

Timeline for upgrading Bazel 6.5.0 to a newer LTS

What is the timeline for OpenXLA projects to upgrade from Bazel 6.5.0 to a newer LTS like 8.0.0?

unread,

Timeline for upgrading Bazel 6.5.0 to a newer LTS

What is the timeline for OpenXLA projects to upgrade from Bazel 6.5.0 to a newer LTS like 8.0.0?

12/23/24

12/20/24

Updated invitation: OpenXLA Community Meeting - Monthly @ Monthly from 9am to 10am on the third Tuesday from Tue Sep 17 to Mon Jan 20, 2025 (PDT) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Monthly from 9am to 10am on the third Tuesday from Tue Sep 17 to Mon Jan 20, 2025 (PDT) (openxla-discuss@openxla.org)

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

12/20/24

Cyril Bortolato

12/12/24

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Jan 14, 2025 9am - 9:45am (PST) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Jan 14, 2025 9am - 9:45am (PST) (openxla-discuss@openxla.org)

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

12/12/24

12/5/24

Inserting custom-calls inside HLOComputation for customized runtime instrumentation

Hi, I'm trying to rewrite HLOModule in order to insert custom-calls inside each HLOComputation.

unread,

Inserting custom-calls inside HLOComputation for customized runtime instrumentation

Hi, I'm trying to rewrite HLOModule in order to insert custom-calls inside each HLOComputation.

12/5/24

Ovidiu Marcu, … Bart Chrzaszcz6

12/4/24

Shardy with PJRT

Hello, Thank you for your time and feedback. I am looking at the idea to have a runtime/middleware

unread,

Shardy with PJRT

Hello, Thank you for your time and feedback. I am looking at the idea to have a runtime/middleware

12/4/24

Cyril Bortolato

12/3/24

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Dec 17, 2024 9am - 10am (PST) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Dec 17, 2024 9am - 10am (PST) (openxla-discuss@openxla.org)

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

12/3/24

11/14/24

Tune in virtually to the OpenXLA Dev Lab tomorrow!

Hey Everyone, I forgot to include the time slots for the sessions mentioned above, so here they are:

unread,

Tune in virtually to the OpenXLA Dev Lab tomorrow!

Hey Everyone, I forgot to include the time slots for the sessions mentioned above, so here they are:

11/14/24

Ben Huntsman, … Peter Hawkins5

11/12/24

Support for IBM Power CPU backend

One thing that seems plausible to me is that we could allow the user to provide an external or system

unread,

Support for IBM Power CPU backend

One thing that seems plausible to me is that we could allow the user to provide an external or system

11/12/24

11/5/24

Layout API propagation across PJRT C-APIs

Hello everyone, I'm trying to understand the propagation of Layout information across PJRT C-APIs

unread,

Layout API propagation across PJRT C-APIs

Hello everyone, I'm trying to understand the propagation of Layout information across PJRT C-APIs

11/5/24

10/28/24

Heads up: CHLO Experimental Ragged Ops

Hi everyone, We are working on a pull request to add experimental ragged ops to HLO, using optionally

unread,

Heads up: CHLO Experimental Ragged Ops

Hi everyone, We are working on a pull request to add experimental ragged ops to HLO, using optionally

10/28/24

10/15/24

[RFC] Add result accuracy to transcendental unary ops

Hi everyone, I posted the RFC on adding support for selecting implementation of transcendental unary

unread,

[RFC] Add result accuracy to transcendental unary ops

Hi everyone, I posted the RFC on adding support for selecting implementation of transcendental unary

10/15/24

10/14/24

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Oct 15, 2024 9am - 10am (PDT) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Oct 15, 2024 9am - 10am (PDT) (openxla-discuss@openxla.org)

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

10/14/24

10/11/24

Join Us for the 2nd OpenXLA Dev Lab

Hello OpenXLA Community! We're excited to invite you to our 2nd OpenXLA Dev Lab on November 14,

unread,

Join Us for the 2nd OpenXLA Dev Lab

Hello OpenXLA Community! We're excited to invite you to our 2nd OpenXLA Dev Lab on November 14,

10/11/24

Sergey Kozub CH

10/9/24

[RFC] Adding MX (microscaling) floating point types to StableHLO and XLA

Hello, everyone I posted RFCs for introducing MX primitive types to StableHLO and XLA: StableHLO RFC#

unread,

[RFC] Adding MX (microscaling) floating point types to StableHLO and XLA

Hello, everyone I posted RFCs for introducing MX primitive types to StableHLO and XLA: StableHLO RFC#

10/9/24

9/28/24

Long term support for infeed/outfeed

Hi folks, Thank you so much for the continuous work on XLA. Without it, we would not be able to bring

unread,

Long term support for infeed/outfeed

Hi folks, Thank you so much for the continuous work on XLA. Without it, we would not be able to bring

9/28/24

Joey Ye, Eugene Zhulenev2

9/27/24

Why was MLIR based CPU backend gone?

Maintenance. It was never fully implemented (launched by default) and was too much work to keep

unread,

Why was MLIR based CPU backend gone?

Maintenance. It was never fully implemented (launched by default) and was too much work to keep

9/27/24

Muneeb Anwar, … Eugene Zhulenev5

9/27/24

Enabling hlo op profiling for cpu

To make this work you'd need to make changes to xla/backends/cpu/runtime/thunk_executor.h. This

unread,

Enabling hlo op profiling for cpu

To make this work you'd need to make changes to xla/backends/cpu/runtime/thunk_executor.h. This

9/27/24

Perry Gibson, … Jan Pfeifer14

9/21/24

StableHLO and KV-cache

Thanks for the more specific reply Peter. You mentioned "you tell XLA you'd like to alias

unread,

StableHLO and KV-cache

Thanks for the more specific reply Peter. You mentioned "you tell XLA you'd like to alias

9/21/24

Cyril Bortolato

9/19/24

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Sep 24, 2024 9am - 9:45am (PDT) (openxla-discuss@openxla.org)

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

unread,

Updated invitation: OpenXLA Community Meeting - Monthly @ Tue Sep 24, 2024 9am - 9:45am (PDT) (openxla-discuss@openxla.org)

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

9/19/24

Search

Clear search

Close search

Google apps

Main menu