AINewsnow

The more aggressive matmul kernel lost to the register budget

This story is from 2026-09-13. It is preserved in the archive; the latest stories are on the live feed.

The WebGPU matmul sweep started with a naive kernel, then added 16 by 16 workgroup tiling and a 4 by 4 output block per thread. At a 2048 cubed matrix size, the measured time moved from 47.24 ms for the naive kernel to 17.23 ms for tiling and 9.12 ms for the 4 by 4 blocked kernel. The blocked versi…

Read the full story at DEV Community — Machine Learning ↗

Timeline · 1 report

  1. 2026-09-13 04:30 · DEV Community — Machine Learning
    The more aggressive matmul kernel lost to the register budget

More stories

  1. AI's role in building AI surging? Anthropic says Claude now leads 26% of its R&D — Mint AI
  2. Anthropic, OpenAI, SpaceXAI, Google sued over call to ‘pace’ AI development — Politico Technology
  3. Google Joins OpenAI, Anthropic, Meta in Disclosing AI Hacks — Bloomberg AI
  4. Introducing Kimi K3 on Amazon Bedrock — AWS Machine Learning Blog
  5. Optimizing agent system prompts with Amazon Bedrock AgentCore — AWS Machine Learning Blog
  6. Introducing Amazon SageMaker HyperPod Inference Gateway — AWS Machine Learning Blog
  7. Introducing Astra for Law — OpenAI News
  8. Novo Nordisk Will Use Anthropic’s Claude for Drug Research — Wall Street Journal Technology

Get the daily brief of stories like this at 6:30 every morning →