Skip to content

npu: a launch's result is dequantised with one multiply per element, not one ldexp(): prefill 24.4-25.1 -> 29.8-30.6 tok/s on the served 27B pack (Refs #1990) - #2008

Merged
bong-water-water-bong merged 1 commit into
mainfrom
cl-npuperf/1990-deq-out
Nov 10, 2026
Merged

npu: a launch's result is dequantised with one multiply per element, not one ldexp(): prefill 24.4-25.1 -> 29.8-30.6 tok/s on the served 27B pack (Refs #1990)#2008
bong-water-water-bong merged 1 commit into
mainfrom
cl-npuperf/1990-deq-out

Commits

  1. Commits on Nov 10, 2026