Mesure split ref: prefill HTP 55 / decode CPU 15 Qwen3.5-4B, end-to-end ~13s

This commit is contained in:
Richard Loyer 2026-05-24 18:08:39 +02:00
parent 1b64bbb7eb
commit eb72df4767
1 changed files with 3 additions and 0 deletions

View File

@ -66,3 +66,6 @@ GATED_DELTA_NET ABSENT du profil HTP = GDN 100% CPU confirme. HTP top: CONT(copi
## Note: gen lente cli = decode sur HTP (pas le fix)
-dev HTP0 garde decode sur NPU = 3-4 t/s + reasoning 200tok = 0.6 t/s percu. Vrai: prefill HTP 58 / decode CPU 13 (-dev none). cli mono-device, ne splitte pas. Cible = JNI prefill HTP -> bascule decode CPU. 0.6 t/s = artefact bench, pas perf.
## Mesure split reference (Qwen3.5-4B-Q4_0, GDN-HTP fixe)
prefill HTP=55.5 / decode CPU=15.3 t/s. ~512tok prefill 9s + 64tok 4s = 13s reponse. cli lent=mono-device; JNI bascule prefill HTP->decode CPU. B-2 etape1 bit-correct livre.