Nvidia’s B200: Keeping the CUDA Juggernaut Rolling ft. Verda (formerly DataCrunch)

·Chips and Cheese··

Nvidia has dominated the GPU compute scene ever since it became mainstream. The company’s Blackwell B200 GPU is the next to take up the mantle of being the go-to compute GPU. Unlike prior generations, Blackwell can’t lean heavily on process node improvements. TSMC’s 4NP process likely provides something over the 4N process used in the older Hopper generation, but it’s unlikely to offer the same degree of improvement as prior full node shrinks. Blackwell therefore moves away from Nvidia’s tried-a...

Read full article →

Related Articles

Show HN: Tilde.run – Agent sandbox with a transactional, versioned filesystem
ozkatz · Hacker News · 2d ago
Computer Use is 45x more expensive than structured APIs
palashawas · Hacker News · 3d ago
Bun is being ported from Zig to Rust
SergeAx · Hacker News · 4d ago
Reverse-engineering the 1998 Ultima Online demo server
notsentient · Hacker News · 2d ago
RaTeX: KaTeX-compatible LaTeX rendering engine in pure Rust
atilimcetin · Hacker News · 4d ago