Optimize PyTorch Performance on the Latest Intel CPUs and GPU

Alex Sin, Intel
Intel Logo Blue

Learn how the latest Intel optimizations extend stock PyTorch on Intel hardware, including the Intel® Xeon® CPU Max Series (formerly codenamed Sapphire Rapids) and Intel® Data Center GPU Max Series (formerly codenamed Ponte Vecchio).

PyTorch is a favorite among AI developers and researchers because it’s easy to learn, interfaces with Python, is easy to debug, can distribute tasks among multiple CPUs and GPUs, and has a lot of extensions.

Sweet, right?

This session introduces the Intel® Extension for PyTorch—part of the Intel® Optimization for PyTorch—which extends the stock sci-computing framework with optimizations for extra performance on Intel® architecture.

Topics covered:

  • An overview of the Intel optimizations, including installation and performance boost metrics
  • The newest features, including Intel® AVX-512 Vector Neural Network Instructions (VNNI), Intel® Advanced Matrix Extensions (Intel® AMX), ease-of-use python API, vectorization, parallelism, quantization, operator fusion, constant folding, and runtime extension
  • A live demo showcasing usage and performance boosts on both CPUs and GPUs.

Sign up today.