Speed test matrix

granite-4.2-30b-fp8 on CMP 170HX 64GB

vllm · 0.32.0rc1.dev249+g18305948a · 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, 16:20 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: FP8
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. KV cache: fp8
# 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-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 context2,0164,065 ms TTFT

16K context1,8088,964 ms TTFT

32K context1,56820,579 ms TTFT

64K context1,23252,230 ms TTFT

120K context896.0134,495 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	35.486 ms TTFT	70.059 ms TTFT	137.696 ms TTFT	267.1120 ms TTFT	510.3177 ms TTFT	Over capacity	Over capacity	Over capacity
16K	32.5149 ms TTFT	60.8240 ms TTFT	106.2290 ms TTFT	171.7436 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	30.0255 ms TTFT	53.9335 ms TTFT	86.2511 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	26.4465 ms TTFT	44.0561 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
120K	21.6Requests queued800 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	2,016.0	—	4,065 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,808.0	—	8,964 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,568.0	—	20,579 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,232.0	—	52,230 ms	—	—	1	—	—
Prefill	120K	1	Valid	896.0	—	134,495 ms	—	—	1	—	—
Sustained decode	0	1	Valid	35.4	35.5	86 ms	28.2 ms	1.0	0	1,062	30.0 s
Sustained decode	0	2	Valid	70.0	35.1	59 ms	28.5 ms	2.0	0	2,100	30.0 s
Sustained decode	0	4	Valid	137.6	34.5	96 ms	29.0 ms	4.0	0	4,128	30.0 s
Sustained decode	0	8	Valid	267.1	33.6	120 ms	29.8 ms	8.0	0	8,008	30.0 s
Sustained decode	0	16	Valid	510.3	32.2	177 ms	31.1 ms	16.0	0	15,296	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	32.5	32.6	149 ms	30.7 ms	1.0	0	975	30.0 s
Sustained decode	16K	2	Valid	60.8	30.4	240 ms	32.9 ms	2.0	0	1,824	30.0 s
Sustained decode	16K	4	Valid	106.2	26.6	290 ms	37.6 ms	4.0	0	3,184	30.0 s
Sustained decode	16K	8	Valid	171.7	21.5	436 ms	46.5 ms	8.0	0	5,152	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	30.0	30.1	255 ms	33.3 ms	1.0	0	901	30.0 s
Sustained decode	32K	2	Valid	53.9	27.0	335 ms	37.1 ms	2.0	0	1,614	30.0 s
Sustained decode	32K	4	Valid	86.2	21.6	511 ms	46.3 ms	4.0	0	2,584	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	26.4	26.5	465 ms	37.7 ms	1.0	0	791	30.0 s
Sustained decode	64K	2	Valid	44.0	22.0	561 ms	45.4 ms	2.0	0	1,318	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	21.6	21.8	800 ms	45.9 ms	1.0	0	648	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.