Speed test matrix

Nanbeige4.3-33B-FP8 on CMP 170HX 64GB

vllm · 0.32.0rc1.dev249+g18305948a · FP85 contexts (0–104K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 104K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 30, 2026, 11:58 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: 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 primitive-ai/Nanbeige4.3-33B-FP8 --max-model-len 114688 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code

Prefill

prompt tok/s by context length

8K context1,8514,366 ms TTFT

16K context1,6299,801 ms TTFT

32K context1,38522,930 ms TTFT

64K context1,06259,635 ms TTFT

104K context819.0125,637 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	35.385 ms TTFT	68.9119 ms TTFT	135.5123 ms TTFT	261.6191 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
16K	31.8206 ms TTFT	58.6303 ms TTFT	102.3415 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	29.0362 ms TTFT	50.2492 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	24.6509 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
104K	20.2Requests queued1,002 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	1,851.0	—	4,366 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,629.0	—	9,801 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,385.0	—	22,930 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,062.0	—	59,635 ms	—	—	1	—	—
Prefill	104K	1	Valid	819.0	—	125,637 ms	—	—	1	—	—
Sustained decode	0	1	Valid	35.3	35.4	85 ms	28.3 ms	1.0	0	1,058	30.0 s
Sustained decode	0	2	Valid	68.9	34.5	119 ms	29.0 ms	2.0	0	2,066	30.0 s
Sustained decode	0	4	Valid	135.5	34.0	123 ms	29.4 ms	4.0	0	4,064	30.0 s
Sustained decode	0	8	Valid	261.6	32.9	191 ms	30.4 ms	8.0	0	7,848	30.0 s
Sustained decode	0	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	31.8	31.9	206 ms	31.4 ms	1.0	0	953	30.0 s
Sustained decode	16K	2	Valid	58.6	29.4	303 ms	34.0 ms	2.0	0	1,758	30.0 s
Sustained decode	16K	4	Valid	102.3	25.6	415 ms	39.1 ms	4.0	0	3,068	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	29.0	29.1	362 ms	34.4 ms	1.0	0	870	30.0 s
Sustained decode	32K	2	Valid	50.2	25.2	492 ms	39.7 ms	2.0	0	1,506	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	24.6	24.7	509 ms	40.5 ms	1.0	0	737	30.0 s
Sustained decode	64K	2	Over capacity	—	—	—	—	—	—	—	—
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	104K	1	Requests queued	20.2	20.5	1,002 ms	48.8 ms	1.0	0	605	30.0 s
Sustained decode	104K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	104K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	104K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	104K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	104K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	104K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	104K	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.