Speed test matrix

TwIL-8B on CMP 170HX 64GB

vllm · 0.32.0rc1.dev249+g18305948a · BF165 contexts (0–120K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 120K

by @Sandbergllm-inference-bench v0.8.0Measured Nov 5, 2026, 09:58 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: BF16
Backend: cuda
Repository: github.com/vllm-project/vllm
Build: ghcr.io/morrowmake/vllm-cmp170hx@sha256:5b45c4d6feafefe2eba962375b0a9a0b64fd22b9e05abb4d1966f14a32a2db79

Run

Served model: bench
Revision: 58d49f0d9909e62f9c2fb09e99fc5c76f3f2d630
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 webAI-Official/TwIL-8B --revision 58d49f0d9909e62f9c2fb09e99fc5c76f3f2d630 --max-model-len 131072 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code

Prefill

prompt tok/s by context length

8K context6,6071,241 ms TTFT

16K context5,6822,853 ms TTFT

32K context4,5307,122 ms TTFT

64K context3,17120,300 ms TTFT

120K context2,07858,014 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	64.355 ms TTFT	126.372 ms TTFT	246.681 ms TTFT	469.2117 ms TTFT	866.3159 ms TTFT	1,393.6220 ms TTFT	Over capacity	Over capacity
16K	56.4169 ms TTFT	102.7267 ms TTFT	173.2271 ms TTFT	262.6462 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	50.1249 ms TTFT	85.1362 ms TTFT	129.4517 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	41.2429 ms TTFT	62.0696 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	31.2833 ms TTFT	41.71,158 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity

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	6,607.0	—	1,241 ms	—	—	1	—	—
Prefill	16K	1	Valid	5,682.0	—	2,853 ms	—	—	1	—	—
Prefill	32K	1	Valid	4,530.0	—	7,122 ms	—	—	1	—	—
Prefill	64K	1	Valid	3,171.0	—	20,300 ms	—	—	1	—	—
Prefill	120K	1	Valid	2,078.0	—	58,014 ms	—	—	1	—	—
Sustained decode	0	1	Valid	64.3	64.5	55 ms	15.5 ms	1.0	0	1,929	30.0 s
Sustained decode	0	2	Valid	126.3	63.5	72 ms	15.8 ms	2.0	0	3,790	30.0 s
Sustained decode	0	4	Valid	246.6	62.1	81 ms	16.1 ms	4.0	0	7,396	30.0 s
Sustained decode	0	8	Valid	469.2	59.4	117 ms	16.8 ms	8.0	0	14,072	30.0 s
Sustained decode	0	16	Valid	866.3	55.4	159 ms	18.1 ms	16.0	0	25,984	30.0 s
Sustained decode	0	32	Valid	1,393.6	44.9	220 ms	22.3 ms	32.0	0	41,812	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	56.4	56.5	169 ms	17.7 ms	1.0	0	1,690	30.0 s
Sustained decode	16K	2	Valid	102.7	51.5	267 ms	19.4 ms	2.0	0	3,080	30.0 s
Sustained decode	16K	4	Valid	173.2	43.4	271 ms	23.0 ms	4.0	0	5,197	30.0 s
Sustained decode	16K	8	Valid	262.6	32.9	462 ms	30.4 ms	8.0	0	7,872	30.0 s
Sustained decode	16K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	1	Valid	50.1	50.2	249 ms	19.9 ms	1.0	0	1,503	30.0 s
Sustained decode	32K	2	Valid	85.1	42.7	362 ms	23.4 ms	2.0	0	2,552	30.0 s
Sustained decode	32K	4	Valid	129.4	32.4	517 ms	30.9 ms	4.0	0	3,880	30.0 s
Sustained decode	32K	8	Over capacity	—	—	—	—	—	—	—	—
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	41.2	41.3	429 ms	24.2 ms	1.0	0	1,234	30.0 s
Sustained decode	64K	2	Valid	62.0	31.1	696 ms	32.2 ms	2.0	0	1,860	30.0 s
Sustained decode	64K	4	Over capacity	—	—	—	—	—	—	—	—
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	120K	1	Valid	31.2	31.4	833 ms	31.9 ms	1.0	0	937	30.0 s
Sustained decode	120K	2	Valid	41.7	20.9	1,158 ms	47.8 ms	2.0	0	1,252	30.0 s
Sustained decode	120K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	128	Over capacity	—	—	—	—	—	—	—	—

Over capacity: not run because this many requests at this context would exceed the server's KV cache.