Speed test matrix

Qwen3.6-27B-FP8 on CMP 170HX 64GB

vllm · 0.32.0rc1.dev249+g18305948a · FP86 contexts (0–248K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 30, 2026, 05:30 UTC

Hardware

Type: Discrete GPU
GPU: CMP 170HX
VRAM: 64 GB
CPU: 2x Intel Xeon E5-2630 v3
RAM: 64 GB
OS: Ubuntu 26.04.1

Engine

Engine: vllm
Version: 0.32.0rc1.dev249+g18305948a
Quantization: FP8
Backend: cuda
Repository: github.com/vllm-project/vllm
Build: ghcr.io/morrowmake/vllm-cmp170hx@sha256:5b45c4d6feafefe2eba962375b0a9a0b64fd22b9e05abb4d1966f14a32a2db79

Run

Served model: bench
Revision: main
Benchmark tool: llm-inference-bench v0.8.0
Decode test: Fixed duration
Prefill mode: Measured during decode warm-up
Time per measurement: 30 s
Max output: 8,192 tokens

Launch command

# Untuned out-of-the-box baseline. Spec decode: none
# 1x NVIDIA CMP 170HX 64GB (cmpunlocker), 180 W limit, PCIe Gen2 x4 | 2x Intel Xeon E5-2630 v3 | 64 GB DDR4 | Ubuntu 26.04.1, NVIDIA driver 610.57.04 | vLLM: Morrowmake vllm-cmp170hx image
vllm serve Qwen/Qwen3.6-27B-FP8 --max-model-len 262144 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only

Prefill

prompt tok/s by context length

8K context1,6814,844 ms TTFT

16K context1,6299,893 ms TTFT

32K context1,53920,850 ms TTFT

64K context1,37546,553 ms TTFT

128K context1,134112,802 ms TTFT

Sustained decode

total tok/s · rows are context length, columns are simultaneous requests

Context	1	2	4	8	16	32	64	128
0	41.4200 ms TTFT	80.9344 ms TTFT	156.6357 ms TTFT	297.2552 ms TTFT	507.7952 ms TTFT	823.91,641 ms TTFT	Over capacity	Over capacity
16K	39.6438 ms TTFT	74.8601 ms TTFT	137.6952 ms TTFT	239.11,662 ms TTFT	363.63,091 ms TTFT	Over capacity	Over capacity	Over capacity
32K	38.1716 ms TTFT	69.91,064 ms TTFT	122.91,662 ms TTFT	199.82,892 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
64K	35.4832 ms TTFT	61.71,163 ms TTFT	101.41,790 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	31.2988 ms TTFT	50.01,181 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
248K	23.7Requests queued2,681 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity

Requests queued: the server could not run all requests at once, so some waited in a queue. Over capacity: not run because this many requests at this context would exceed the server's KV cache.

TTFT is measured after the benchmark warms the server's prompt cache, so it mostly excludes prefill.

All measurements

Latencies are medians. Total tok/s is output across all simultaneous requests.

Phase	Context	Requests at once	Status	Total tok/s	Per request tok/s	TTFT	Time between tokens	Average active requests	Completed requests	Output tokens	Measured for
Prefill	8K	1	Valid	1,681.0	—	4,844 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,629.0	—	9,893 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,539.0	—	20,850 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,375.0	—	46,553 ms	—	—	1	—	—
Prefill	128K	1	Valid	1,134.0	—	112,802 ms	—	—	1	—	—
Sustained decode	0	1	Valid	41.4	41.5	200 ms	24.1 ms	1.0	0	1,242	30.0 s
Sustained decode	0	2	Valid	80.9	40.5	344 ms	24.7 ms	2.0	0	2,426	30.0 s
Sustained decode	0	4	Valid	156.6	39.2	357 ms	25.5 ms	4.0	0	4,696	30.0 s
Sustained decode	0	8	Valid	297.2	37.2	552 ms	26.9 ms	8.0	0	8,912	30.0 s
Sustained decode	0	16	Valid	507.7	31.8	952 ms	31.4 ms	16.0	0	15,232	30.0 s
Sustained decode	0	32	Valid	823.9	25.7	1,641 ms	38.8 ms	32.0	0	24,704	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	39.6	39.6	438 ms	25.2 ms	1.0	0	1,186	30.0 s
Sustained decode	16K	2	Valid	74.8	37.3	601 ms	26.8 ms	2.0	0	2,244	30.0 s
Sustained decode	16K	4	Valid	137.6	34.0	952 ms	29.4 ms	4.0	0	4,128	30.0 s
Sustained decode	16K	8	Valid	239.1	29.1	1,662 ms	34.4 ms	8.0	0	7,168	30.0 s
Sustained decode	16K	16	Valid	363.6	21.3	3,091 ms	47.0 ms	16.0	0	10,896	30.0 s
Sustained decode	16K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	1	Valid	38.1	38.1	716 ms	26.2 ms	1.0	0	1,141	30.0 s
Sustained decode	32K	2	Valid	69.9	34.7	1,064 ms	28.8 ms	2.0	0	2,094	30.0 s
Sustained decode	32K	4	Valid	122.9	30.1	1,662 ms	33.3 ms	4.0	0	3,685	30.0 s
Sustained decode	32K	8	Valid	199.8	23.7	2,892 ms	42.2 ms	8.0	0	5,992	30.0 s
Sustained decode	32K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	1	Valid	35.4	35.5	832 ms	28.1 ms	1.0	0	1,062	30.0 s
Sustained decode	64K	2	Valid	61.7	30.7	1,163 ms	32.6 ms	2.0	0	1,850	30.0 s
Sustained decode	64K	4	Valid	101.4	24.9	1,790 ms	40.2 ms	4.0	0	3,040	30.0 s
Sustained decode	64K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	1	Valid	31.2	31.3	988 ms	32.0 ms	1.0	0	935	30.0 s
Sustained decode	128K	2	Valid	50.0	25.0	1,181 ms	40.0 ms	2.0	0	1,498	30.0 s
Sustained decode	128K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	1	Requests queued	23.7	25.5	2,681 ms	39.2 ms	0.9	0	711	30.0 s
Sustained decode	248K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	128	Over capacity	—	—	—	—	—	—	—	—

Requests queued: the server could not run all requests at once, so some waited in a queue. Over capacity: not run because this many requests at this context would exceed the server's KV cache.