First native PyTorch-CUDA build for Windows on Arm lands as NVIDIA RTX Spark ships Oct 16
lmoroney · x · 2026-10-14
Ahead of RTX Spark laptops shipping Friday, Oct. 16 (up to 128GB unified memory, 1 petaflop of AI compute), NVIDIA says the community is releasing the first PyTorch-CUDA build for Windows on Arm — a Python 3.15 build native by default — plus a native Conda installer with 2,000+ packages, with 97% of the top 15,000 PyTorch libraries said to work. Real PyTorch work on WoA laptops is finally close to viable.
Also announced: Nemotron 3.5 Bolt 74B, purpose-built for agentic AI on RTX Spark, arrives Oct. 15; NVIDIA PAIR exits beta with llama.cpp support; top agent apps (GitHub Copilot, Perplexity, etc.) get streamlined setup, OpenShell and Microsoft Execution Containers (MXC) security, and hybrid local/cloud AI routing; Adobe Photoshop and Premiere are being rearchitected for up to 2x performance gains.
More from Infra
- AWS launches Lambda Web Functions with up to 128 concurrent requests per environment — tlakomy · 2026-10-14
- Bosch and Rittal show massive manifolds for MW liquid cooling and 800VDC — BenBajarin · 2026-10-14
- OCP Global Summit sponsor wall shows Nvidia, AMD, Google Cloud, Meta and hundreds more — firstadopter · 2026-10-14
- Strata Runs 100 tks in Agentic Workloads but Shows Looping and Port-Conflict Issues — GodComplecs · 2026-10-14
- Ben Lorica: local AI is a routing problem, not a privacy feature — bigdata · 2026-10-14
- Meta CFO warned Zuckerberg the superintelligence bet could end 'very, very bad' — robleclerc · 2026-10-14