Speed test matrix

granite-4.2-8b-nvfp4 on RTX 3090 24GB

vllm · 0.32.0 · NVFP45 contexts (0–80K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 80K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 29, 2026, 08:26 UTCEdited Oct 29, 2026, 12:41 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: NVFP4
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 ibm-granite/granite-4.2-8b-nvfp4 --max-model-len 90112 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code

Prefill

prompt tok/s by context length

8K context5,8421,411 ms TTFT

16K context4,9853,273 ms TTFT

32K context3,9598,202 ms TTFT

64K context2,76723,412 ms TTFT

80K context2,40533,644 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	118.928 ms TTFT	222.629 ms TTFT	410.227 ms TTFT	708.529 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
16K	88.066 ms TTFT	140.787 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	69.8117 ms TTFT	101.8164 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	49.1287 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
80K	42.9334 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	5,842.0	—	1,411 ms	—	—	1	—	—
Prefill	16K	1	Valid	4,985.0	—	3,273 ms	—	—	1	—	—
Prefill	32K	1	Valid	3,959.0	—	8,202 ms	—	—	1	—	—
Prefill	64K	1	Valid	2,767.0	—	23,412 ms	—	—	1	—	—
Prefill	80K	1	Valid	2,405.0	—	33,644 ms	—	—	1	—	—
Sustained decode	0	1	Valid	118.9	120.0	28 ms	8.3 ms	1.0	0	3,566	30.0 s
Sustained decode	0	2	Valid	222.6	113.3	29 ms	8.8 ms	2.0	0	6,676	30.0 s
Sustained decode	0	4	Valid	410.2	105.3	27 ms	9.5 ms	4.0	0	12,306	30.0 s
Sustained decode	0	8	Valid	708.5	92.6	29 ms	10.8 ms	8.0	0	21,255	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	88.0	88.2	66 ms	11.3 ms	1.0	0	2,639	30.0 s
Sustained decode	16K	2	Valid	140.7	70.6	87 ms	14.2 ms	2.0	0	4,220	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	69.8	70.0	117 ms	14.3 ms	1.0	0	2,093	30.0 s
Sustained decode	32K	2	Valid	101.8	51.0	164 ms	19.6 ms	2.0	0	3,052	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	49.1	49.2	287 ms	20.3 ms	1.0	0	1,472	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	Valid	42.9	43.1	334 ms	23.2 ms	1.0	0	1,288	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	—	—	—	—	—	—	—	—

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