openxla / xla

A machine learning compiler for GPUs, CPUs, and ML accelerators
Apache License 2.0
2.64k stars 418 forks source link

[XLA:GPU] Disable `xla_gpu_enable_pipelined_all_gather` by default. #17924

Closed copybara-service[bot] closed 6 days ago

copybara-service[bot] commented 1 week ago

[XLA:GPU] Disable xla_gpu_enable_pipelined_all_gather by default.

This breaks the natural optimization barrier of a while loop.