Speed test matrix

Llama-3.3-70B-Instruct-FP8-dynamic on CMP 170HX 64GB ×2

vllm · 0.32.0rc1.dev249+g18305948a · FP8 dynamic5 contexts (0–120K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 120K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 03:08 UTC

Hardware

Type: Discrete GPU
GPU: CMP 170HX
GPU count: 2
VRAM: 64 GB (2 × 32 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 dynamic
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. Layout: PP2 (stock vLLM flag)
# 2x 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 RedHatAI/Llama-3.3-70B-Instruct-FP8-dynamic --max-model-len 131072 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --pipeline-parallel-size 2

Prefill

prompt tok/s by context length

8K context1,3556,055 ms TTFT

16K context1,35611,961 ms TTFT

32K context1,25025,824 ms TTFT

64K context1,02063,119 ms TTFT

120K context758.0159,148 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	18.5151 ms TTFT	36.7193 ms TTFT	60.6183 ms TTFT	119.2236 ms TTFT	234.7315 ms TTFT	Over capacity	Over capacity	Over capacity
16K	17.2213 ms TTFT	28.4311 ms TTFT	51.3445 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	16.1340 ms TTFT	26.6413 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	14.3477 ms TTFT	23.7647 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	11.8Requests queued1,067 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	1,355.0	—	6,055 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,356.0	—	11,961 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,250.0	—	25,824 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,020.0	—	63,119 ms	—	—	1	—	—
Prefill	120K	1	Valid	758.0	—	159,148 ms	—	—	1	—	—
Sustained decode	0	1	Valid	18.5	18.5	151 ms	54.0 ms	1.0	0	555	30.0 s
Sustained decode	0	2	Valid	36.7	18.3	193 ms	54.5 ms	2.0	0	1,100	30.0 s
Sustained decode	0	4	Valid	60.6	15.2	183 ms	65.9 ms	4.0	0	1,816	30.0 s
Sustained decode	0	8	Valid	119.2	14.9	236 ms	67.0 ms	8.0	0	3,574	30.0 s
Sustained decode	0	16	Valid	234.7	14.7	315 ms	68.1 ms	16.0	0	7,038	30.0 s
Sustained decode	0	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	17.2	17.2	213 ms	58.1 ms	1.0	0	516	30.0 s
Sustained decode	16K	2	Valid	28.4	14.2	311 ms	70.5 ms	2.0	0	850	30.0 s
Sustained decode	16K	4	Valid	51.3	12.8	445 ms	78.0 ms	4.0	0	1,539	30.0 s
Sustained decode	16K	8	Over capacity	—	—	—	—	—	—	—	—
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	16.1	16.2	340 ms	61.9 ms	1.0	0	484	30.0 s
Sustained decode	32K	2	Valid	26.6	13.3	413 ms	75.0 ms	2.0	0	798	30.0 s
Sustained decode	32K	4	Over capacity	—	—	—	—	—	—	—	—
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	14.3	14.4	477 ms	69.7 ms	1.0	0	430	30.0 s
Sustained decode	64K	2	Valid	23.7	11.9	647 ms	84.2 ms	2.0	0	710	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	Requests queued	11.8	12.0	1,067 ms	83.0 ms	1.0	0	355	30.0 s
Sustained decode	120K	2	Over capacity	—	—	—	—	—	—	—	—
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	—	—	—	—	—	—	—	—

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.