AMD and Intel Simplify AI Development on x86 CPUs Using PyTorch

Rival chipmakers combine x86 PyTorch optimisation efforts via oneDNN

By GPU Data Hub Desk · GPU Data Hub·Editorial standards·20 Oct 2026, 19:32

AMD and Intel Simplify AI Development on x86 CPUs Using PyTorch

AMD and Intel are collaborating to unify their optimization work for PyTorch workloads executing on x86 processors. The joint effort is being channelled through the open-source oneDNN library, which is governed by the Unified Acceleration Foundation. This shared framework aims to simplify deployment and improve application performance across both chipmakers' hardware platforms.

Original source

This summary was written by the GPU Data Hub desk from reporting published by HPCwire on 20 Oct 2026, 19:32. Read the full article at the original publisher.

Read Original Source →

Companies

AMD Intel Unified Acceleration Foundation

Topics

CPUs PyTorch oneDNN x86 architecture open source

LinkedIn X

More from GPU Data Hub

Latest →

AMD's FSR 4.1 now works on Radeon RX 6000 cards, Xbox Series X and S, and Ryzen AI Max, AI 9, and Z2 SoCs

Advanced Micro Devices has expanded the availability of its FSR 4.1 upscaling software across a broader selection of silicon. The technology is now functional on legacy Radeon RX 6000 discrete graphics processing units alongside Xbox Series X and S home consoles. Support has also been integrated into recent Ryzen AI Max, AI 9, and Z2 system-on-chip families.

Semiconductors·4h ago·Tom's Hardware

xMems solid-state cooling chips for data centers and Edge devices go into production

Component manufacturer xMEMS has commenced commercial production of solid-state cooling silicon intended for data centres and edge computing deployments. The devices utilise micro-vibrations to manage thermal dissipation across high-performance hardware.

Semiconductors·6h ago·Data Center Dynamics

Mysterious AMD Zen 6 APU with 24GB of LPDDR5 memory appears in benchmarks

Benchmark listings have revealed an unreleased AMD Zen 6 accelerated processing unit paired with 24 gigabytes of high-speed LPDDR5 memory. The early engineering unit incorporates six processor cores and integrated graphics silicon running at 2.4 GHz, though targeted end-market platforms remain unspecified.

Semiconductors·7h ago·Tom's Hardware

NVIDIA Runs Record Fault-Tolerant Quantum Simulation at Jülich Supercomputing Centre

Nvidia has achieved a landmark quantum computing simulation using JUPITER, Europe's inaugural exascale supercomputer located at the Jülich Supercomputing Centre. The project enables scientists to analyse and validate magic state cultivation techniques essential for building fault-tolerant quantum systems.

Semiconductors·9h ago·HPCwire

Modal taps Nvidia Vera CPUs after firm's internal tests show 2x throughput over Intel and AMD chips

Modal has integrated Nvidia's Vera central processing units after proprietary benchmarking indicated a twofold gain in throughput compared with comparable x86 processors from Intel and AMD. The performance advantage was specifically recorded during the execution of artificial intelligence agent workloads. The cloud platform is deploying the silicon to improve execution speeds across its operational infrastructure.

Semiconductors·10h ago·Tom's Hardware

Intel asks FCC for waiver to buy foreign robots

Intel has formally petitioned the US Federal Communications Commission for regulatory relief regarding the acquisition of automated guided vehicles produced overseas. The chipmaker fears that extensive import prohibitions could prevent it from servicing or expanding its internal fleet of factory robotics. Such restrictions pose potential operational risks to its domestic fabrication facility automation.

Semiconductors·10h ago·Tom's Hardware