←back to thread

Llama 3.1 405B now runs at 969 tokens/s on Cerebras Inference

(cerebras.ai)
426 points benchmarkist | 1 comments | 19 Nov 24 00:15 UTC | HN request time: 0.222s | source
Show context
zackangelo ◴[19 Nov 24 02:11 UTC] No.42179476[source]▶
This is astonishingly fast. I’m struggling to get over 100 tok/s on my own Llama 3.1 70b implementation on an 8x H100 cluster.

I’m curious how they’re doing it. Obviously the standard bag of tricks (eg, speculative decoding, flash attention) won’t get you close. It seems like at a minimum you’d have to do multi-node inference and maybe some kind of sparse attention mechanism?

replies(9): >>42179489 #>>42179493 #>>42179501 #>>42179503 #>>42179754 #>>42179794 #>>42180035 #>>42180144 #>>42180569 #
1. boroboro4 ◴[19 Nov 24 03:08 UTC] No.42179754[source]▶
There are two big tricks: their chips are enormous, and they use sram as their memory, which is vastly faster than hbm ram being used by GPUs. In fact this is main reason it’s so fast. Groq has the speed because of the same reason.