s
May 13, 2024, 4:08 a.m. |

Simon Willison's Weblog simonwillison.net

GPUs Go Brrr


Fascinating, detailed low-level notes on how to get the most out of NVIDIA's H100 GPUs (currently selling for around $40,000 a piece) from the research team at Stanford who created FlashAttention, among other things.

The swizzled memory layouts are flat-out incorrectly documented, which took considerable time for us to figure out.



Via Hacker News

ai figure flat gpus h100 hacker low memory notes nvidia research research team selling stanford team via

Doctoral Researcher (m/f/div) in Automated Processing of Bioimages

@ Leibniz Institute for Natural Product Research and Infection Biology (Leibniz-HKI) | Jena

Seeking Developers and Engineers for AI T-Shirt Generator Project

@ Chevon Hicks | Remote

Software Engineer for AI Training Data (School Specific)

@ G2i Inc | Remote

Software Engineer for AI Training Data (Python)

@ G2i Inc | Remote

Software Engineer for AI Training Data (Tier 2)

@ G2i Inc | Remote

Manager, Business Intelligence

@ Revlon | New York City, United States