Build a Unified Heterogeneous AI Computing Ecosystem for PyTorch
Talkat KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 · 2026-09-07..09-09·
summary
Describes efforts to standardize heterogeneous AI hardware support for PyTorch by the TAC Accelerator Integration Working Group (co-chaired by Huawei and Intel). Covers their hardware onboarding guidelines, cross-repository CI testing, device-aware test suites, and platform incubation workflows, and outlines a roadmap toward device-agnostic APIs and a broader multi-backend test matrix to enable vendors and developers to integrate hardware more easily.
AI-written summary · about
related talks
06mKeynote: Intel and PyTorch: Enabling AI Everywhere with Ubiquitous Hardware and Open...
25mRace to Superintelligence: Building Low-Friction AI Infrastructure for a Heterogen...
06mThe Building Blocks of Agentic Al
16mEdge AI needs to be for everyone: breaking barriers to adoption
18mtorch.accelerator: A Unified, Device-Agnostic Runtime API for Stream-Based Accelerators
37mEnabling AI For Everyone - Hiai Open Platform Introduction
13mLightning Talk: Challenges and Standardization in PyTorch Ecosystem...
08mUnlocking AI at Scale for PyTorch Developers