PyTorchCon 2026 highlights: MoE comm hiding, CUDA graphs, 10x data loaders
PyTorch · x · 2026-10-18
PyTorch officially previews five highlighted talks for PyTorch Conference North America 2026, picked by Ecosystem Working Group co-chair Chris Hoge (NVIDIA):
- The Ecosystem Landscape: how projects apply for ecosystem status and lifecycle management
- Hiding expert-parallel communication in Mixture-of-Experts training
- Parameterized dynamic-shaped CUDA graphs
- Building fused compute-communication kernels with PyTorch Symmetric Memory
- Scaling PyTorch data loaders by 10x
Registration link in the original post.
More from Infra
- AgentCave puts your scattered Macs and GPU boxes in one virtual room — noahsolomon · 2026-10-18
- Real AI demand estimated at ~$189B in 2026, over half from OpenAI and Anthropic — builderjaydub · 2026-10-18
- Twinax cable shipments in AI backplanes hit 85,200 km/year at 224G — moon-length in 4.5 years — jwt0625 · 2026-10-18
- DeepSeek-V4.1-Flash-1017 GGUF quants appear on Hugging Face with 473GiB Q8_0 — realmrfakename · 2026-10-18
- DeepSeek-4.1-Flash Runs on 4x RTX Pro at 300+ TPS With NVMe Offload — TheZachMueller · 2026-10-18
- SGLang Summit sponsors jump from 11 to 36 in one week, 2,000+ attendees expected — ying11231 · 2026-10-18