Speed test matrix

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

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 30, 2026, 20:39 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
# 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 90112 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code

Prefill

prompt tok/s by context length

8K context2,7752,954 ms TTFT

16K context2,2777,120 ms TTFT

32K context1,68319,174 ms TTFT

64K context1,10458,336 ms TTFT

80K context941.085,461 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	105.353 ms TTFT	194.463 ms TTFT	345.149 ms TTFT	591.969 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
16K	70.4174 ms TTFT	104.5199 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	52.6253 ms TTFT	70.4321 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	34.7488 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
80K	29.8574 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	2,775.0	—	2,954 ms	—	—	1	—	—
Prefill	16K	1	Valid	2,277.0	—	7,120 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,683.0	—	19,174 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,104.0	—	58,336 ms	—	—	1	—	—
Prefill	80K	1	Valid	941.0	—	85,461 ms	—	—	1	—	—
Sustained decode	0	1	Valid	105.3	106.9	53 ms	9.4 ms	1.0	0	3,157	30.0 s
Sustained decode	0	2	Valid	194.4	99.3	63 ms	10.1 ms	2.0	0	5,837	30.0 s
Sustained decode	0	4	Valid	345.1	89.2	49 ms	11.2 ms	4.0	0	10,352	30.0 s
Sustained decode	0	8	Valid	591.9	78.6	69 ms	12.7 ms	8.0	0	17,759	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	70.4	70.8	174 ms	14.1 ms	1.0	0	2,112	30.0 s
Sustained decode	16K	2	Valid	104.5	52.4	199 ms	19.1 ms	2.0	0	3,134	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	52.6	52.7	253 ms	19.0 ms	1.0	0	1,577	30.0 s
Sustained decode	32K	2	Valid	70.4	35.2	321 ms	28.4 ms	2.0	0	2,110	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	34.7	34.8	488 ms	28.7 ms	1.0	0	1,041	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	29.8	29.9	574 ms	33.4 ms	1.0	0	894	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.