Speed test matrix

DeepSeek-R1-Distill-Qwen-14B-FP8-dynamic on RTX 3090 24GB

vllm · 0.32.0 · FP8 dynamic5 contexts (0–80K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 80K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 02:47 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 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. KV cache: int4_per_token_head
# 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 RedHatAI/DeepSeek-R1-Distill-Qwen-14B-FP8-dynamic --max-model-len 90112 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code  --kv-cache-dtype int4_per_token_head --attention-backend TRITON_ATTN

Prefill

prompt tok/s by context length

8K context2,1373,857 ms TTFT

16K context1,47511,078 ms TTFT

32K context918.035,455 ms TTFT

64K context518.0125,433 ms TTFT

80K context424.0191,168 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	48.042 ms TTFT	94.346 ms TTFT	181.484 ms TTFT	338.486 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
16K	36.3127 ms TTFT	59.1179 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	29.1218 ms TTFT	42.5329 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	20.9Requests queued413 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
80K	18.3Requests queued561 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	2,137.0	—	3,857 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,475.0	—	11,078 ms	—	—	1	—	—
Prefill	32K	1	Valid	918.0	—	35,455 ms	—	—	1	—	—
Prefill	64K	1	Valid	518.0	—	125,433 ms	—	—	1	—	—
Prefill	80K	1	Valid	424.0	—	191,168 ms	—	—	1	—	—
Sustained decode	0	1	Valid	48.0	48.2	42 ms	20.8 ms	1.0	0	1,440	30.0 s
Sustained decode	0	2	Valid	94.3	47.4	46 ms	21.1 ms	2.0	0	2,830	30.0 s
Sustained decode	0	4	Valid	181.4	45.7	84 ms	21.9 ms	4.0	0	5,440	30.0 s
Sustained decode	0	8	Valid	338.4	42.9	86 ms	23.3 ms	8.0	0	10,144	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	36.3	36.3	127 ms	27.5 ms	1.0	0	1,088	30.0 s
Sustained decode	16K	2	Valid	59.1	29.6	179 ms	33.8 ms	2.0	0	1,772	30.0 s
Sustained decode	16K	4	Over capacity	—	—	—	—	—	—	—	—
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	29.1	29.2	218 ms	34.3 ms	1.0	0	874	30.0 s
Sustained decode	32K	2	Valid	42.5	21.3	329 ms	47.0 ms	2.0	0	1,274	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	Requests queued	20.9	20.9	413 ms	47.9 ms	1.0	0	626	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	80K	1	Requests queued	18.3	18.3	561 ms	54.5 ms	1.0	0	548	30.0 s
Sustained decode	80K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	80K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	80K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	80K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	80K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	80K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	80K	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.