llamaperf

Qwen3.8 27B

on RTX 4090 48GB · vLLM · 262,144 ctx

Tone: positive
Sep 7, 2026
Throughput
68.4 t/s gen
Quant
FP8
KV cache
FP8 E4M3
MTP (Multi-Token Prediction)
on
VRAM reported
48 GB

Use cases

coding

Summary

User reports decode speeds on an RTX 4090 with custom-tuned FP8 kernels: no MTP about 19.7 t/s, MTP=3 about 49.7 t/s, MTP=5 about 53.2 t/s, and with custom kernels about 68.4 t/s. Setup uses MTP=5 speculative decoding. The numbers come from real Cursor sessions, not fixed-prompt A/B.