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, 09:07 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: int4_per_token_head
# 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 int4_per_token_head --attention-backend TRITON_ATTN

Prefill

prompt tok/s by context length

8K context1,6115,089 ms TTFT

16K context1,05615,350 ms TTFT

32K context625.051,659 ms TTFT

64K context344.0186,975 ms TTFT

120K context420.0287,337 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	90.6118 ms TTFT	170.136 ms TTFT	303.750 ms TTFT	508.468 ms TTFT	782.4112 ms TTFT	1,305.8183 ms TTFT	Over capacity	Over capacity
16K	50.4195 ms TTFT	78.6282 ms TTFT	108.7407 ms TTFT	129.1554 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	34.4262 ms TTFT	50.4359 ms TTFT	64.5544 ms TTFT	72.0786 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
64K	21.1Requests queued496 ms TTFT	29.0Requests queued654 ms TTFT	35.5931 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	3.1Requests queued23,085 ms TTFT	16.5Requests queued1,167 ms TTFT	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	1,611.0	—	5,089 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,056.0	—	15,350 ms	—	—	1	—	—
Prefill	32K	1	Valid	625.0	—	51,659 ms	—	—	1	—	—
Prefill	64K	1	Valid	344.0	—	186,975 ms	—	—	1	—	—
Prefill	120K	1	Valid	420.0	—	287,337 ms	—	—	1	—	—
Sustained decode	0	1	Valid	90.6	92.1	118 ms	10.9 ms	1.0	0	2,719	30.0 s
Sustained decode	0	2	Valid	170.1	87.0	36 ms	11.5 ms	2.0	0	5,102	30.0 s
Sustained decode	0	4	Valid	303.7	78.5	50 ms	12.7 ms	4.0	0	9,108	30.0 s
Sustained decode	0	8	Valid	508.4	66.8	68 ms	15.0 ms	8.0	0	15,248	30.0 s
Sustained decode	0	16	Valid	782.4	52.4	112 ms	19.1 ms	16.0	0	23,473	30.0 s
Sustained decode	0	32	Valid	1,305.8	43.6	183 ms	23.0 ms	32.0	0	39,168	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	50.4	50.6	195 ms	19.8 ms	1.0	0	1,512	30.0 s
Sustained decode	16K	2	Valid	78.6	39.5	282 ms	25.3 ms	2.0	0	2,358	30.0 s
Sustained decode	16K	4	Valid	108.7	27.3	407 ms	36.6 ms	4.0	0	3,256	30.0 s
Sustained decode	16K	8	Valid	129.1	16.2	554 ms	61.7 ms	8.0	0	3,872	30.0 s
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	34.4	34.5	262 ms	29.0 ms	1.0	0	1,033	30.0 s
Sustained decode	32K	2	Valid	50.4	25.2	359 ms	39.7 ms	2.0	0	1,510	30.0 s
Sustained decode	32K	4	Valid	64.5	16.2	544 ms	61.8 ms	4.0	0	1,932	29.9 s
Sustained decode	32K	8	Valid	72.0	9.0	786 ms	110.7 ms	8.0	0	2,160	30.0 s
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	21.1	21.1	496 ms	47.3 ms	1.0	0	632	30.0 s
Sustained decode	64K	2	Requests queued	29.0	14.6	654 ms	68.6 ms	2.0	0	869	30.0 s
Sustained decode	64K	4	Valid	35.5	8.9	931 ms	112.6 ms	4.0	0	1,064	29.9 s
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	3.1	12.7	23,085 ms	79.0 ms	0.7	0	94	29.9 s
Sustained decode	120K	2	Requests queued	16.5	8.4	1,167 ms	119.3 ms	1.9	0	493	30.0 s
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.