Rapid growth in AI and machine learning innovations and workloads necessitates constant developments in both software and hardware infrastructure. Developers of TensorFlow* (which is the Google* end-to-end open source, machine learning framework) and Intel® oneAPI Deep Neural Network Library (oneDNN) have been collaborating closely to enable users to fully use new hardware features and accelerators, with a focus on x86 architecture. This talk covers recent projects such as int8^† and bfloat16^‡ vectorization support that brings custom oneDNN operations to basic TensorFlow software and the upcoming Intel® XPU device plug-in for TensorFlow.

^†Intel® Advanced Vector Extensions 512 (Intel® AVX-512) with Vector Neural Network Instructions (VNNI)
^‡Intel AVX512 for bfloat16

Speaker

Penporn Koanantakool is a senior software engineer at Google. She leads the TensorFlow performance optimization collaboration with Intel. Penporn holds a PhD in computer science from the University of California, Berkeley, and a bachelor of engineering degree in computer engineering from Kasetsart University, Thailand.