Speed test matrix

granite-4.2-30b-nvfp4 on CMP 170HX 64GB

vllm · 0.32.0rc1.dev249+g18305948a · NVFP45 contexts (0–120K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 120K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 29, 2026, 05:14 UTC

Hardware

Type: Discrete GPU
GPU: CMP 170HX
VRAM: 64 GB
CPU: 2x Intel Xeon E5-2630 v3
RAM: 64 GB
OS: Ubuntu 26.04.1

Engine

Engine: vllm
Version: 0.32.0rc1.dev249+g18305948a
Quantization: NVFP4
Backend: cuda
Repository: github.com/vllm-project/vllm
Build: ghcr.io/morrowmake/vllm-cmp170hx@sha256:5b45c4d6feafefe2eba962375b0a9a0b64fd22b9e05abb4d1966f14a32a2db79

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 NVIDIA CMP 170HX 64GB (cmpunlocker), 180 W limit, PCIe Gen2 x4 | 2x Intel Xeon E5-2630 v3 | 64 GB DDR4 | Ubuntu 26.04.1, NVIDIA driver 610.57.04 | vLLM: Morrowmake vllm-cmp170hx image
vllm serve ibm-granite/granite-4.2-30b-nvfp4 --max-model-len 131072 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code

Prefill

prompt tok/s by context length

8K context1,6544,955 ms TTFT

16K context1,51310,711 ms TTFT

32K context1,35323,845 ms TTFT

64K context1,11357,848 ms TTFT

120K context850.0141,902 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	47.6105 ms TTFT	92.2123 ms TTFT	179.485 ms TTFT	341.6102 ms TTFT	565.1162 ms TTFT	Over capacity	Over capacity	Over capacity
16K	40.0166 ms TTFT	70.3241 ms TTFT	116.6339 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	34.7267 ms TTFT	56.3368 ms TTFT	83.9529 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	27.6503 ms TTFT	39.7600 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	20.1Requests queued987 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	1,654.0	—	4,955 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,513.0	—	10,711 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,353.0	—	23,845 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,113.0	—	57,848 ms	—	—	1	—	—
Prefill	120K	1	Valid	850.0	—	141,902 ms	—	—	1	—	—
Sustained decode	0	1	Valid	47.6	47.8	105 ms	20.9 ms	1.0	0	1,428	30.0 s
Sustained decode	0	2	Valid	92.2	46.3	123 ms	21.6 ms	2.0	0	2,766	30.0 s
Sustained decode	0	4	Valid	179.4	45.2	85 ms	22.1 ms	4.0	0	5,380	30.0 s
Sustained decode	0	8	Valid	341.6	43.2	102 ms	23.2 ms	8.0	0	10,248	30.0 s
Sustained decode	0	16	Valid	565.1	35.9	162 ms	27.9 ms	16.0	0	16,944	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	40.0	40.1	166 ms	25.0 ms	1.0	0	1,198	30.0 s
Sustained decode	16K	2	Valid	70.3	35.3	241 ms	28.3 ms	2.0	0	2,110	30.0 s
Sustained decode	16K	4	Valid	116.6	29.2	339 ms	34.2 ms	4.0	0	3,496	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	34.7	34.8	267 ms	28.7 ms	1.0	0	1,041	30.0 s
Sustained decode	32K	2	Valid	56.3	28.2	368 ms	35.4 ms	2.0	0	1,688	30.0 s
Sustained decode	32K	4	Valid	83.9	21.0	529 ms	47.7 ms	4.0	0	2,512	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	27.6	27.7	503 ms	36.1 ms	1.0	0	828	30.0 s
Sustained decode	64K	2	Valid	39.7	19.9	600 ms	50.3 ms	2.0	0	1,190	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	20.1	20.4	987 ms	49.1 ms	1.0	0	602	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.