undefined | Better HN

0 pointsivankra23d ago0 comments

But memory bandwidth (bottleneck for LLM inference) is only marginally improved, 614 GB/s vs 546 GB/s for M4/M5 Max - where is this 4x improvement coming from?

I think I'll pass on upgrading.

0 comments

singhrac23d ago

It’s prompt processing so prefill - that’s compute bound not memory.

0x45723d ago

4x is on Time To First Token it's on the graph.

j / k navigate · click thread line to collapse