Speed test matrix

gemma-4-12B-it-FP8-Dynamic on RTX 3090 24GB

vllm · 0.32.0 · FP8 dynamic6 contexts (0–248K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 08:37 UTC

Hardware

Type: Discrete GPU
GPU: RTX 3090
VRAM: 24 GB
CPU: AMD Ryzen 9 7900X (12C)
RAM: 32 GB
OS: Ubuntu 24.04.5

Engine

Engine: vllm
Version: 0.32.0
Quantization: FP8 dynamic
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), 350 W limit (VBIOS max), PCIe Gen4 x16 | AMD Ryzen 9 7900X 12C | 32 GB DDR5 | Ubuntu 24.04.5, NVIDIA driver 595.99.02
vllm serve RedHatAI/gemma-4-12B-it-FP8-Dynamic --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 context3,3852,367 ms TTFT

16K context2,9305,417 ms TTFT

32K context2,30213,726 ms TTFT

64K context1,61838,953 ms TTFT

128K context1,019123,529 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	56.935 ms TTFT	110.746 ms TTFT	209.847 ms TTFT	381.158 ms TTFT	650.7121 ms TTFT	923.8Requests queued104 ms TTFT	Over capacity	Over capacity
16K	49.970 ms TTFT	Looped	189.7146 ms TTFT	Looped	Over capacity	Over capacity	Over capacity	Over capacity
32K	46.5120 ms TTFT	92.1156 ms TTFT	Looped	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	Looped	84.8293 ms TTFT	Looped	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	Looped	Looped	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
248K	28.9Requests queued1,011 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. Looped: the model kept repeating the same text, so the benchmark discarded this measurement. 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	3,385.0	—	2,367 ms	—	—	1	—	—
Prefill	16K	1	Valid	2,930.0	—	5,417 ms	—	—	1	—	—
Prefill	32K	1	Valid	2,302.0	—	13,726 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,618.0	—	38,953 ms	—	—	1	—	—
Prefill	128K	1	Valid	1,019.0	—	123,529 ms	—	—	1	—	—
Sustained decode	0	1	Valid	56.9	57.1	35 ms	17.5 ms	1.0	0	1,707	30.0 s
Sustained decode	0	2	Valid	110.7	55.7	46 ms	17.9 ms	2.0	0	3,321	30.0 s
Sustained decode	0	4	Valid	209.8	53.0	47 ms	18.9 ms	4.0	0	6,292	30.0 s
Sustained decode	0	8	Valid	381.1	48.5	58 ms	20.6 ms	8.0	0	11,432	30.0 s
Sustained decode	0	16	Valid	650.7	42.0	121 ms	23.8 ms	16.0	0	19,520	30.0 s
Sustained decode	0	32	Requests queued	923.8	36.3	104 ms	27.5 ms	26.5	0	27,691	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	49.9	50.0	70 ms	20.0 ms	1.0	0	1,496	30.0 s
Sustained decode	16K	2	Looped	—	—	—	—	—	0	0	—
Sustained decode	16K	4	Valid	189.7	47.6	146 ms	21.0 ms	4.0	0	5,688	30.0 s
Sustained decode	16K	8	Looped	—	—	—	—	—	0	0	—
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	46.5	46.5	120 ms	21.5 ms	1.0	0	1,394	30.0 s
Sustained decode	32K	2	Valid	92.1	46.0	156 ms	21.7 ms	2.0	0	2,762	30.0 s
Sustained decode	32K	4	Looped	—	—	—	—	—	0	0	—
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	Looped	—	—	—	—	—	0	0	—
Sustained decode	64K	2	Valid	84.8	42.4	293 ms	23.6 ms	2.0	0	2,542	30.0 s
Sustained decode	64K	4	Looped	—	—	—	—	—	0	0	—
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	Looped	—	—	—	—	—	0	0	—
Sustained decode	128K	2	Looped	—	—	—	—	—	0	0	—
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	28.9	29.4	1,011 ms	34.0 ms	1.0	0	868	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	—	—	—	—	—	—	—	—