AMD 与 Intel 合作统一 x86 CPU 的 PyTorch 优化
AMD, Intel Join Forces on x86 CPU AI Development < Semiconductor < 기사본문 - The Elec Inc.
AMD 与 Intel 于 10 月 20 日宣布，将各自在 PyTorch 中的 x86 CPU 优化工作合并到 oneDNN 库，以减少 AI 生态碎片化，方便开发者在两家 CPU 上构建、维护和部署 AI 应用。oneDNN 是面向 x86 CPU 的开源深度学习加速库，最初由 Intel 开发，现由 Linux 基金会下的 UXL Foundation 管理。双方还在合作推进 x86 指令集扩展 ACE，用于加速 AI 计算中的矩阵乘法，该技术将通过后续 oneDNN 更新集成进 PyTorch。
正文
当前显示原文，中文全文译文尚未就绪。
가
The body content of the article will be changed to this letter size.
스크롤 이동 상태바
AMD and Intel are collaborating to simplify AI development. (Photo: Intel)
AMD and Intel are collaborating to simplify artificial intelligence development for x86 central processing units (CPUs). The companies announced on Oct. 20 local time that they will consolidate their x86 CPU optimization efforts within PyTorch into the oneAPI Deep Neural Network Library (oneDNN).
The initiative aims to reduce fragmentation in the AI ecosystem and make optimization technologies more accessible. It will allow developers to build, maintain and deploy AI applications more easily across AMD and Intel CPUs.
PyTorch is a collection of Python-based software libraries and tools for deep learning that helps developers build and deploy AI models quickly and efficiently.
oneDNN is an open-source deep learning acceleration library that serves as an acceleration engine for x86 CPUs within PyTorch. It helps PyTorch maximize hardware performance and improve computational efficiency. Originally developed by Intel, oneDNN is now managed by the UXL Foundation under the Linux Foundation.
AMD and Intel are also collaborating on AI Compute Extensions (ACE), a next-generation extension to the x86 instruction set architecture (ISA). ACE is designed to accelerate matrix multiplication, a fundamental operation in AI computing. The technology will be integrated into PyTorch through future oneDNN updates.
JEONG IL JOO 1week@thelec.kr
My comment