Taalas HC1: 17,000 tokens/sec on Llama 3.1 8B vs Nvidia H200's 233 tokens/sec. 73x faster at one-tenth the power. Each chip runs ONE model, hardwired into the transistors.

you are viewing a single comment's thread
view the rest of the comments
[–] 21 points 8 months ago

The HC1 chip doesn't load model weights from memory. It etches them directly into the transistors. Every weight becomes a physical circuit.

That's one way to avoid memory bandwidth constraints!

  • source