B200 Attention Kernel from Scratch to Near-SOTA in 60 Diagrams

·Hacker News··

B200 attention from scratch in 60 diagrams: 14 CUDA/PTX kernels from naive to near-SOTA, plus a video-generation capstone.

Read full article →

Related Articles

OpenAI’s o1 correctly diagnosed 67% of ER patients vs. 50-55% by triage doctors
donsupreme · Hacker News · 4mo ago
Accelerating Gemma 4: faster inference with multi-token prediction drafters
amrrs · Hacker News · 3mo ago
A couple million lines of Haskell: Production engineering at Mercury
unignorant · Hacker News · 4mo ago
Using “underdrawings” for accurate text and numbers
samcollins · Hacker News · 4mo ago
LLM Judges Verify Presence, Not Absence: Omission Blindness in AI Clinical Notes
sbulaev · Hacker News · 4h ago