xla
xla copied to clipboard
openxla
A machine learning compiler for GPUs, CPUs, and ML accelerators
#sdy export NamedComputationOps with backend_configs to CallOps. Needed to make sure host callbacks in XLA aren't inlined but also propagated through.
![copybara-service[bot] avatar](https://avatars.githubusercontent.com/in/44061?v=4 "copybara-service[bot] avatar"){kind=avatar}
[XLA:GPU] Improve error message for unsupported dot algorithms. The current error message can be a bit confusing because it will report that an algorithm is "unsupported on the current device"...
The leak comes from [MakeDebugOptionsFlag](https://github.com/openxla/xla/blob/0d9511953b74eb614b9db1dcd45f100d89067c6b/xla/debug_options_flags.cc#L342). In real usage, the function is run only once in a program's whole lifetime. This test calls it more than once and got memory leaks....
This is the issue I reported originally at https://github.com/tensorflow/tensorflow/issues/49944 The problem is that the values returned by the float to uint8 cast are always in the wrong order. I.e. the...
Integrate LLVM at llvm/llvm-project@ac2a2816e3fe Updates LLVM usage to match [ac2a2816e3fe](https://github.com/llvm/llvm-project/commit/ac2a2816e3fe)