Tag
PyTorch Conference North America 2026 will feature a talk by Huawei's Zesheng Zong on achieving full API parity across heterogeneous accelerators, highlighting the adaptation of 600,000+ upstream test cases and the new Cross-Repository CI Relay for real-time bidirectional testing.
PyTorch Conference North America (Oct 20-21, San Jose) features a keynote by Priya Nagpurkar of IBM Research on IBM and Red Hat's contributions to the PyTorch ecosystem, including out-of-tree accelerator infrastructure, Spyre Accelerator enablement, and llm-d distributed inference work.
In the interview, OpenAI VP of Hardware Richard Ho looks back on his chip career — from Stanford and the D. E. Shaw Anton supercomputer to eight generations of Google TPUs — and reveals the team-building philosophy behind OpenAI's in-house accelerator, Jalapeno: team first, ideas second; a blank-sheet design approach; and, most importantly, the key engineering lesson of keeping hardware programmable and avoiding overfitting to specific models.
IBM and Red Hat contributors share a PyTorch community blog detailing how they built Torch Spyre on top of PyTorch's Cross-Repository CI Relay (CRCR), using an agentic test-selection pipeline and declarative YAML test-reuse framework to achieve L2 integration for out-of-tree accelerator backends.
ApplySeed is a product that helps startups streamline accelerator applications by filling out forms based on provided info or ChatGPT-generated content, aiming to secure clear Yes/No responses from funding programs.
This blog post explores two AI-augmented coding approaches—accelerators who maintain code understanding and vibecoders who delegate more to AI—and discusses their long-term implications for software maintenance and team dynamics.
A tweet thread highlights Sam Altman's lecture on startup fundraising and shares a verified spreadsheet of top accelerators for 2026.
OpenAI and Broadcom announced a multi-year strategic collaboration to co-develop and deploy 10 gigawatts of custom AI accelerators and networking systems, with deployment beginning in mid-2026 and completion by end of 2029. This partnership enables OpenAI to design accelerators that embed learnings from frontier model development directly into hardware.