Speed test matrix

granite-4.2-8b-fp8 on RTX 3090 24GB

vllm · 0.32.0 · FP85 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, 15:41 UTC

Hardware

Type: Discrete GPU
GPU: RTX 3090
VRAM: 24 GB
CPU: AMD Ryzen 9 7900X (12C)
RAM: 32 GB
OS: Ubuntu 24.04.5

Engine

Engine: vllm
Version: 0.32.0
Quantization: FP8
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 3090 24GB (Zagitta/gpu-unlock), 350 W limit (VBIOS max), PCIe Gen4 x16 | AMD Ryzen 9 7900X 12C | 32 GB DDR5 | Ubuntu 24.04.5, NVIDIA driver 595.99.02
vllm serve ibm-granite/granite-4.2-8b-fp8 --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 context5,3801,531 ms TTFT

16K context4,7023,470 ms TTFT

32K context3,6778,831 ms TTFT

64K context2,56925,212 ms TTFT

120K context1,67672,396 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	82.726 ms TTFT	162.128 ms TTFT	314.131 ms TTFT	594.634 ms TTFT	1,091.351 ms TTFT	Over capacity	Over capacity	Over capacity
16K	73.242 ms TTFT	141.459 ms TTFT	245.986 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	65.964 ms TTFT	124.392 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	55.1113 ms TTFT	98.9157 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	42.8228 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	5,380.0	—	1,531 ms	—	—	1	—	—
Prefill	16K	1	Valid	4,702.0	—	3,470 ms	—	—	1	—	—
Prefill	32K	1	Valid	3,677.0	—	8,831 ms	—	—	1	—	—
Prefill	64K	1	Valid	2,569.0	—	25,212 ms	—	—	1	—	—
Prefill	120K	1	Valid	1,676.0	—	72,396 ms	—	—	1	—	—
Sustained decode	0	1	Valid	82.7	82.9	26 ms	12.1 ms	1.0	0	2,480	30.0 s
Sustained decode	0	2	Valid	162.1	81.4	28 ms	12.3 ms	2.0	0	4,862	30.0 s
Sustained decode	0	4	Valid	314.1	79.1	31 ms	12.6 ms	4.0	0	9,420	30.0 s
Sustained decode	0	8	Valid	594.6	75.3	34 ms	13.3 ms	8.0	0	17,832	30.0 s
Sustained decode	0	16	Valid	1,091.3	69.8	51 ms	14.3 ms	16.0	0	32,736	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	73.2	73.2	42 ms	13.7 ms	1.0	0	2,194	30.0 s
Sustained decode	16K	2	Valid	141.4	70.8	59 ms	14.1 ms	2.0	0	4,242	30.0 s
Sustained decode	16K	4	Valid	245.9	61.7	86 ms	16.2 ms	4.0	0	7,376	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	65.9	65.9	64 ms	15.2 ms	1.0	0	1,976	30.0 s
Sustained decode	32K	2	Valid	124.3	62.2	92 ms	16.1 ms	2.0	0	3,728	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	55.1	55.1	113 ms	18.1 ms	1.0	0	1,652	30.0 s
Sustained decode	64K	2	Valid	98.9	49.5	157 ms	20.2 ms	2.0	0	2,968	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	Valid	42.8	42.8	228 ms	23.4 ms	1.0	0	1,283	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	—	—	—	—	—	—	—	—

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