Python

AMD and Intel are combining CPU optimizations for PyTorch in oneDNN

AMD and Intel are combining their CPU optimizations for PyTorch in the oneDNN library, creating a shared path for tuning software on both companies’ processors. AMD’s ZenDNN, a downstream oneDNN version tailored to its Zen processors, is moving into the open-source project. AMD also joined the UXL Foundation as a steering member, with an annual contribution of $20,000. The companies said the integration was recommended by the x86 Ecosystem Advisory Group, established in 2024.

AI-written summary, checked against the cited sources. How summaries work

Image: Phoronix

What we know

Confirmed by several sources

  • AMD and Intel are collaborating to bring AMD CPU optimizations into oneDNN for PyTorch. — Tom's Hardware, Phoronix
  • AMD maintained ZenDNN as a downstream, oneDNN-compatible library for its processor designs. — Tom's Hardware, Phoronix

Reported by one source

  • AMD joined the UXL Foundation as a steering member and will contribute $20,000 per year. — Tom's Hardware
  • The x86 Ecosystem Advisory Group recommended unifying the optimizations and was established in 2024. — Tom's Hardware

Coverage 2 publishers

  1. PhoronixFirst report

    Intel & AMD Now Collaborating On Software Optimizations With PyTorch/oneDNN

    Intel has long worked on tuning their oneDNN neural network library for their hardware and saw its adoption by PyTorch and other AI frameworks. AMD has maintained ZenDNN as their downstream fork that is oneDNN API compatible while catering…

  2. Tom's Hardware

    AMD and Intel will work together to create a single PyTorch code path for AI inference on CPUs

    Intel and AMD are integrating their CPU optimizations for PyTorch into a unified library, offering tuning for both AMD and Intel chips through the oneDNN library. AMD has maintained a downstream version of oneDNN called ZenDNN, which is now…

Earliest report first. Articles stay on their publishers’ sites; each link opens the original.