Speed test matrix

Tess-4.5-27B-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, 04:19 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 vwdubb/Tess-4.5-27B-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 context2,1073,886 ms TTFT

16K context1,9888,128 ms TTFT

32K context1,86617,221 ms TTFT

64K context1,62939,331 ms TTFT

128K context1,30697,939 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	37.5175 ms TTFT	73.2265 ms TTFT	141.3367 ms TTFT	269.1627 ms TTFT	489.71,043 ms TTFT	672.71,524 ms TTFT	Over capacity	Over capacity
16K	36.0409 ms TTFT	68.5571 ms TTFT	126.3895 ms TTFT	219.91,529 ms TTFT	350.12,913 ms TTFT	Over capacity	Over capacity	Over capacity
32K	34.9654 ms TTFT	64.6962 ms TTFT	113.91,533 ms TTFT	185.72,692 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
64K	32.8866 ms TTFT	57.81,152 ms TTFT	95.11,731 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	29.3948 ms TTFT	47.51,327 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
248K	22.6Requests queued2,740 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	2,107.0	—	3,886 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,988.0	—	8,128 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,866.0	—	17,221 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,629.0	—	39,331 ms	—	—	1	—	—
Prefill	128K	1	Valid	1,306.0	—	97,939 ms	—	—	1	—	—
Sustained decode	0	1	Valid	37.5	37.5	175 ms	26.7 ms	1.0	0	1,124	30.0 s
Sustained decode	0	2	Valid	73.2	36.7	265 ms	27.3 ms	2.0	0	2,196	30.0 s
Sustained decode	0	4	Valid	141.3	35.4	367 ms	28.3 ms	4.0	0	4,240	30.0 s
Sustained decode	0	8	Valid	269.1	33.7	627 ms	29.7 ms	8.0	0	8,072	30.0 s
Sustained decode	0	16	Valid	489.7	30.7	1,043 ms	32.6 ms	16.0	0	14,688	30.0 s
Sustained decode	0	32	Valid	672.7	20.8	1,524 ms	48.1 ms	32.0	0	20,192	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	36.0	36.1	409 ms	27.7 ms	1.0	0	1,080	30.0 s
Sustained decode	16K	2	Valid	68.5	34.2	571 ms	29.3 ms	2.0	0	2,056	30.0 s
Sustained decode	16K	4	Valid	126.3	31.2	895 ms	32.0 ms	4.0	0	3,788	30.0 s
Sustained decode	16K	8	Valid	219.9	26.8	1,529 ms	37.3 ms	8.0	0	6,592	30.0 s
Sustained decode	16K	16	Valid	350.1	20.6	2,913 ms	48.5 ms	16.0	0	10,496	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	34.9	34.9	654 ms	28.6 ms	1.0	0	1,045	30.0 s
Sustained decode	32K	2	Valid	64.6	32.1	962 ms	31.2 ms	2.0	0	1,936	30.0 s
Sustained decode	32K	4	Valid	113.9	27.9	1,533 ms	35.9 ms	4.0	0	3,416	30.0 s
Sustained decode	32K	8	Valid	185.7	22.1	2,692 ms	45.3 ms	8.0	0	5,568	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	32.8	32.9	866 ms	30.4 ms	1.0	0	984	30.0 s
Sustained decode	64K	2	Valid	57.8	28.8	1,152 ms	34.8 ms	2.0	0	1,732	30.0 s
Sustained decode	64K	4	Valid	95.1	23.4	1,731 ms	42.8 ms	4.0	0	2,852	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	29.3	29.4	948 ms	34.1 ms	1.0	0	877	30.0 s
Sustained decode	128K	2	Valid	47.5	23.8	1,327 ms	42.1 ms	2.0	0	1,426	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	22.6	24.3	2,740 ms	41.1 ms	0.9	0	677	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.