xla
xla copied to clipboard
openxla
A machine learning compiler for GPUs, CPUs, and ML accelerators
[xla:gpu] Only instantiate cuda graph if the captured function gets called frequently
![copybara-service[bot] avatar](https://avatars.githubusercontent.com/in/44061?v=4 "copybara-service[bot] avatar"){kind=avatar}
WeakRefCounted sends notification before destructors. This makes the behavior inside the weak notification more defined. e.g. if the weak notification's receiver is owned by the WeakRefCounted that sends the notification,...
Convert local rendevzous table to use WeakPtr. A new EagerContext stat is added for inspecting number of 'leaked' local Rendezvous objects. I checked this metric manually and with this CL...
[LAX:RBG] Allow any type to RngBitGenerator [XLA:PRNG] Support narrow rbg
Integrate LLVM at llvm/llvm-project@bf8f684efff3 Updates LLVM usage to match [bf8f684efff3](https://github.com/llvm/llvm-project/commit/bf8f684efff3)
Reverts f4d70c211dea602d10263e3c3cd1abc9805e2f2c
Add xla_extension APIs for getting stablehlo serialization versions