Scott Wasson
@scottwasson.bsky.social
Founder, The Tech Report and Ars Technica. Formerly AMD Radeon and Intel GPU groups. Now product management at MLCommons. Opinions are my own.
Yeah. It’s a funny thing to combine them.
August 1, 2025 at 2:32 PM
Yeah. It’s a funny thing to combine them.
More info here:
mlcommons.org/2025/07/mlpe...
mlcommons.org/2025/07/mlpe...
MLCommons Releases MLPerf Client v1.0: A New Standard for AI PC and Client LLM Benchmarking - MLCommons
MLCommons Releases MLPerf Client v1.0 with Expanded Models, Prompts, and Hardware Support, Standardizing AI PC Performance.
mlcommons.org
July 30, 2025 at 3:39 PM
More info here:
mlcommons.org/2025/07/mlpe...
mlcommons.org/2025/07/mlpe...
Don’t have a lift, so going with a shop.
May 17, 2025 at 6:18 PM
Don’t have a lift, so going with a shop.
Just got a stage-1 tune for my mk7 GTI. Need a beefier clutch to handle the torque.
May 17, 2025 at 4:22 PM
Just got a stage-1 tune for my mk7 GTI. Need a beefier clutch to handle the torque.
We have four new benchmarks this round as we work to keep pace with AI innovation:
-Llama 3.1 405B, our largest model yet
-Llama 2 70B interactive, with lower latency targets
-RGAT using a graph neural network
-Automotive PointPainting for edge devices
-Llama 3.1 405B, our largest model yet
-Llama 2 70B interactive, with lower latency targets
-RGAT using a graph neural network
-Automotive PointPainting for edge devices
April 3, 2025 at 4:19 PM
We have four new benchmarks this round as we work to keep pace with AI innovation:
-Llama 3.1 405B, our largest model yet
-Llama 2 70B interactive, with lower latency targets
-RGAT using a graph neural network
-Automotive PointPainting for edge devices
-Llama 3.1 405B, our largest model yet
-Llama 2 70B interactive, with lower latency targets
-RGAT using a graph neural network
-Automotive PointPainting for edge devices
Progress over the past year includes a doubling of the median score per round for our Llama 2 70B LLM test, while the top score is over 3X faster now. This is the server scenario.
April 3, 2025 at 4:19 PM
Progress over the past year includes a doubling of the median score per round for our Llama 2 70B LLM test, while the top score is over 3X faster now. This is the server scenario.