With parallel sampling at n=8, a 2K-token prompt and 500 output tokens each, how many KV blocks does copy-on-write save?
Count the blocks with and without sharing and the saving looks large. The answer that scores also knows when the copy fires, what a misaligned prompt does, and the workload where the saving all but disappears.
Updated Oct 2026 · Worked answers on inference mechanisms, with calculations, implementation details, and follow-up questions.
Count the blocks with and without sharing and the saving looks large. The answer that scores also knows when the copy fires, what a misaligned prompt does, and the workload where the saving all but disappears.
At batch 1 the step reads every weight once to produce one token per sequence, so arithmetic intensity sits near 1 FLOP per byte and the tensor cores idle.