Speed test matrix

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

vllm · 0.32.0rc1.dev249+g18305948a · FP85 contexts (0–112K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 112K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 30, 2026, 06:25 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
# 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 122880 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code

Prefill

prompt tok/s by context length

8K context2,0034,093 ms TTFT

16K context1,8118,952 ms TTFT

32K context1,58820,315 ms TTFT

64K context1,26950,730 ms TTFT

112K context977.0115,196 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	38.885 ms TTFT	75.3105 ms TTFT	147.588 ms TTFT	282.4108 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
16K	33.7184 ms TTFT	60.3233 ms TTFT	102.3324 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
32K	29.8294 ms TTFT	49.8349 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	24.4549 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
112K	18.8Requests queued937 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	2,003.0	—	4,093 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,811.0	—	8,952 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,588.0	—	20,315 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,269.0	—	50,730 ms	—	—	1	—	—
Prefill	112K	1	Valid	977.0	—	115,196 ms	—	—	1	—	—
Sustained decode	0	1	Valid	38.8	38.9	85 ms	25.7 ms	1.0	0	1,164	30.0 s
Sustained decode	0	2	Valid	75.3	37.8	105 ms	26.5 ms	2.0	0	2,260	30.0 s
Sustained decode	0	4	Valid	147.5	37.0	88 ms	27.0 ms	4.0	0	4,424	30.0 s
Sustained decode	0	8	Valid	282.4	35.6	108 ms	28.1 ms	8.0	0	8,472	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	33.7	33.7	184 ms	29.6 ms	1.0	0	1,010	30.0 s
Sustained decode	16K	2	Valid	60.3	30.2	233 ms	33.1 ms	2.0	0	1,810	30.0 s
Sustained decode	16K	4	Valid	102.3	25.6	324 ms	39.0 ms	4.0	0	3,068	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	29.8	29.9	294 ms	33.4 ms	1.0	0	895	30.0 s
Sustained decode	32K	2	Valid	49.8	24.9	349 ms	40.1 ms	2.0	0	1,492	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	24.4	24.5	549 ms	40.8 ms	1.0	0	732	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	112K	1	Requests queued	18.8	18.9	937 ms	53.0 ms	1.0	0	563	30.0 s
Sustained decode	112K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	112K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	112K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	112K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	112K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	112K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	112K	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.