PyTorch 2.15 Q&A set: NCCL rewrite now default, faster NVFP4 decode on Blackwell
PyTorch · x · 2026-11-03
PyTorch announced a live release Q&A for PyTorch 2.15 on Thursday, November 5 at 11 AM PT, with Meta's Tristan Rice, Natalia Gimelshein, Huy Do, and Jane Xu presenting and answering community questions, moderated by Chris Gottbrath.
Key changes in 2.15:
- The rewritten NCCL backend is now the default, with new one-sided transports via a NIXL backend
- Finer memory and communication control for FSDP2 and pipelining
- Faster 4-bit (NVFP4) LLM decoding on NVIDIA Blackwell GPUs
- FP8 and faster decode on Apple Silicon, plus broader platform support
More from Infra
- rauchg: stateless sandboxes are the wrong benchmark as Vercel Sandbox gets 2.4x faster — cramforce · 2026-11-03
- Microsoft Cloud hits $63B quarter at 30% growth as AI buildout keeps scaling — thedealdirector · 2026-11-03
- Open-source sparse attention: Qwen 27B with no accuracy loss, 10M-token context — Immediate_Sky_6566 · 2026-11-03
- A year of renting an H100 costs $37.6K as warnings grow about tightening compute supply — teortaxesTex · 2026-11-03
- Mistral Large 4 (1T params) runs locally on two M3 Ultras at 20 tok/s via MLX-VLM — andrejusb · 2026-11-03
- AEHR jumps 16% and LSCC 9% after hours as both chip stocks report earnings — BenBajarin · 2026-11-03