AMD and Intel have reached a partnership, designating oneDNN as the default open-source library for x86 acceleration in PyTorch.

BlockBeats news, October 21: Intel announced that AMD and Intel have reached an agreement to make oneDNN the default open-source library for x86 acceleration in PyTorch. Through this collaboration, both parties will simplify the AI development process for developers on x86 CPUs and further optimize AI computing support for related platforms.