Intel & AMD Now Collaborating On Software Optimizations With PyTorch/oneDNN
Phoronix
Intel and AMD announce collaboration on software optimizations for PyTorch using the oneDNN library.Summary
Intel and AMD have jointly announced a collaboration to optimize software for PyTorch using the oneDNN library. Intel has long tuned oneDNN for its hardware, while AMD maintained ZenDNN as a compatible fork. Now, AMD's ZenDNN optimizations will be integrated into oneDNN, making it the default open-source library for x86 acceleration in PyTorch. Additionally, AMD will join the UXL Foundation as a steering member, contributing to the consortium that manages many former oneAPI projects. This partnership follows their work on the x86 Ecosystem Advisory Group and marks a significant step in x86 CPU optimization for AI frameworks.
(Source：Phoronix)