Speed test matrix

Hy-MT2-7B-FP8 on RTX 3090 24GB

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

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 02:06 UTC

Hardware

Type: Discrete GPU
GPU: RTX 3090
VRAM: 24 GB
CPU: Intel Core i7-9700K (8C)
RAM: 32 GB
OS: Ubuntu 24.04.5

Engine

Engine: vllm
Version: 0.32.0
Quantization: FP8
Backend: cuda
Repository: github.com/vllm-project/vllm
Build: vllm/vllm-openai:v0.32.0-x86_64

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 RTX 3090 24GB (Zagitta/gpu-unlock), 420 W limit, PCIe Gen3 x16 | Intel Core i7-9700K 8C | 32 GB DDR4 | Ubuntu 24.04.5, NVIDIA driver 580.178.04
vllm serve tencent/Hy-MT2-7B-FP8 --max-model-len 114688 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --enforce-eager

Prefill

prompt tok/s by context length

8K context6,5741,253 ms TTFT

16K context5,9282,755 ms TTFT

32K context4,8006,779 ms TTFT

64K context3,39519,125 ms TTFT

104K context2,48042,504 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	31.8105 ms TTFT	60.91,419 ms TTFT	120.6108 ms TTFT	239.1114 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
16K	32.0141 ms TTFT	59.0131 ms TTFT	117.5182 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	32.0141 ms TTFT	60.2191 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	31.9242 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
104K	31.9373 ms TTFT	Over capacity	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	6,574.0	—	1,253 ms	—	—	1	—	—
Prefill	16K	1	Valid	5,928.0	—	2,755 ms	—	—	1	—	—
Prefill	32K	1	Valid	4,800.0	—	6,779 ms	—	—	1	—	—
Prefill	64K	1	Valid	3,395.0	—	19,125 ms	—	—	1	—	—
Prefill	104K	1	Valid	2,480.0	—	42,504 ms	—	—	1	—	—
Sustained decode	0	1	Valid	31.8	31.9	105 ms	31.4 ms	1.0	0	955	30.0 s
Sustained decode	0	2	Valid	60.9	30.4	1,419 ms	32.9 ms	2.0	0	1,826	30.0 s
Sustained decode	0	4	Valid	120.6	30.1	108 ms	33.2 ms	4.0	0	3,616	30.0 s
Sustained decode	0	8	Valid	239.1	29.9	114 ms	33.5 ms	8.0	0	7,168	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	32.0	32.0	141 ms	31.3 ms	1.0	0	960	30.0 s
Sustained decode	16K	2	Valid	59.0	29.5	131 ms	33.9 ms	2.0	0	1,770	30.0 s
Sustained decode	16K	4	Valid	117.5	29.4	182 ms	34.1 ms	4.0	0	3,525	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	32.0	32.0	141 ms	31.3 ms	1.0	0	958	30.0 s
Sustained decode	32K	2	Valid	60.2	30.0	191 ms	33.3 ms	2.0	0	1,804	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	31.9	32.0	242 ms	31.3 ms	1.0	0	958	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	Valid	31.9	31.9	373 ms	31.3 ms	1.0	0	956	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	—	—	—	—	—	—	—	—

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