Speed test matrix

granite-4.2-3b-nvfp4 on RTX 3060 12GB

vllm · 0.32.0 · NVFP45 contexts (0–120K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 120K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 15:53 UTC

Hardware

Type: Discrete GPU
GPU: RTX 3060
VRAM: 12 GB
CPU: Intel Xeon E5-2680 v4 (14C)
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. KV cache: fp8
# 1x RTX 3060 12GB, stock 170 W limit, PCIe Gen3 x16 | Intel Xeon E5-2680 v4 14C | 32 GB DDR4-2400 | Ubuntu 24.04.5, NVIDIA driver 580.178.04
vllm serve ibm-granite/granite-4.2-3b-nvfp4 --max-model-len 131072 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --kv-cache-dtype fp8

Prefill

prompt tok/s by context length

8K context2,7972,930 ms TTFT

16K context2,2757,127 ms TTFT

32K context1,66919,336 ms TTFT

64K context1,08459,406 ms TTFT

120K context672.0179,535 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	111.057 ms TTFT	211.767 ms TTFT	389.949 ms TTFT	679.865 ms TTFT	1,146.0110 ms TTFT	Over capacity	Over capacity	Over capacity
16K	87.8157 ms TTFT	158.4193 ms TTFT	203.5307 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	72.1215 ms TTFT	125.5314 ms TTFT	135.9475 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	53.3384 ms TTFT	88.7534 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	36.4Requests queued591 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	2,797.0	—	2,930 ms	—	—	1	—	—
Prefill	16K	1	Valid	2,275.0	—	7,127 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,669.0	—	19,336 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,084.0	—	59,406 ms	—	—	1	—	—
Prefill	120K	1	Valid	672.0	—	179,535 ms	—	—	1	—	—
Sustained decode	0	1	Valid	111.0	111.7	57 ms	9.0 ms	1.0	0	3,329	30.0 s
Sustained decode	0	2	Valid	211.7	107.2	67 ms	9.3 ms	2.0	0	6,350	30.0 s
Sustained decode	0	4	Valid	389.9	99.8	49 ms	10.0 ms	4.0	0	11,696	30.0 s
Sustained decode	0	8	Valid	679.8	88.5	65 ms	11.3 ms	8.0	0	20,389	30.0 s
Sustained decode	0	16	Valid	1,146.0	75.6	110 ms	13.2 ms	16.0	0	34,383	30.0 s
Sustained decode	0	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	87.8	88.0	157 ms	11.4 ms	1.0	0	2,633	30.0 s
Sustained decode	16K	2	Valid	158.4	79.5	193 ms	12.6 ms	2.0	0	4,750	30.0 s
Sustained decode	16K	4	Valid	203.5	51.2	307 ms	19.5 ms	4.0	0	6,104	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	72.1	72.3	215 ms	13.8 ms	1.0	0	2,162	30.0 s
Sustained decode	32K	2	Valid	125.5	63.0	314 ms	15.9 ms	2.0	0	3,766	30.0 s
Sustained decode	32K	4	Valid	135.9	34.1	475 ms	29.3 ms	4.0	0	4,075	30.0 s
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	53.3	53.5	384 ms	18.7 ms	1.0	0	1,597	30.0 s
Sustained decode	64K	2	Valid	88.7	44.5	534 ms	22.4 ms	2.0	0	2,662	30.0 s
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	120K	1	Requests queued	36.4	36.7	591 ms	27.2 ms	1.0	0	1,092	30.0 s
Sustained decode	120K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	120K	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.