Speed test matrix
gemma-4-31B-it-FP8-dynamic on CMP 170HX 64GB
vllm · 0.32.0rc1.dev249+g18305948a · FP8 dynamic6 contexts (0–248K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K
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 dynamic
- 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 RedHatAI/gemma-4-31B-it-FP8-dynamic --max-model-len 262144 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only
Prefill
prompt tok/s by context length8K context1,5195,246 ms TTFT
16K context1,25812,538 ms TTFT
32K context992.031,658 ms TTFT
64K context694.090,245 ms TTFT
128K context432.0289,691 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 | 31.088 ms TTFT | 60.4812 ms TTFT | 115.4175 ms TTFT | 211.6223 ms TTFT | 387.9307 ms TTFT | 557.2486 ms TTFT | Over capacity | Over capacity |
| 16K | 27.6229 ms TTFT | 52.7327 ms TTFT | 98.1455 ms TTFT | 167.1643 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | 25.7350 ms TTFT | 48.2480 ms TTFT | 87.0705 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 23.7647 ms TTFT | 43.4908 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 20.1Requests queued1,435 ms TTFT | 34.9Requests queued1,771 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 248K | Over capacity | 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 | 1,519.0 | — | 5,246 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 1,258.0 | — | 12,538 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 992.0 | — | 31,658 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 694.0 | — | 90,245 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 432.0 | — | 289,691 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 31.0 | 31.1 | 88 ms | 32.1 ms | 1.0 | 0 | 930 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 60.4 | 30.4 | 812 ms | 32.9 ms | 2.0 | 0 | 1,812 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 115.4 | 29.1 | 175 ms | 34.3 ms | 4.0 | 0 | 3,460 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 211.6 | 26.9 | 223 ms | 37.2 ms | 8.0 | 0 | 6,344 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 387.9 | 24.9 | 307 ms | 40.1 ms | 16.0 | 0 | 11,632 | 30.0 s |
| Sustained decode | 0 | 32 | Valid | 557.2 | 17.9 | 486 ms | 55.8 ms | 32.0 | 0 | 16,704 | 30.0 s |
| Sustained decode | 0 | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 27.6 | 27.6 | 229 ms | 36.2 ms | 1.0 | 0 | 829 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 52.7 | 26.4 | 327 ms | 37.8 ms | 2.0 | 0 | 1,581 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 98.1 | 24.5 | 455 ms | 40.8 ms | 4.0 | 0 | 2,940 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 167.1 | 20.9 | 643 ms | 47.8 ms | 8.0 | 0 | 5,008 | 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 | 25.7 | 25.8 | 350 ms | 38.8 ms | 1.0 | 0 | 771 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 48.2 | 24.2 | 480 ms | 41.3 ms | 2.0 | 0 | 1,446 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 87.0 | 21.8 | 705 ms | 45.8 ms | 4.0 | 0 | 2,608 | 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 | 23.7 | 23.8 | 647 ms | 42.1 ms | 1.0 | 0 | 710 | 30.0 s |
| Sustained decode | 64K | 2 | Valid | 43.4 | 21.8 | 908 ms | 45.8 ms | 2.0 | 0 | 1,300 | 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 | 128K | 1 | Requests queued | 20.1 | 20.7 | 1,435 ms | 48.4 ms | 1.0 | 0 | 601 | 30.0 s |
| Sustained decode | 128K | 2 | Requests queued | 34.9 | 18.2 | 1,771 ms | 54.8 ms | 1.9 | 0 | 1,046 | 30.0 s |
| Sustained decode | 128K | 4 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 128K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 128K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 128K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 128K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 128K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 1 | Over capacity | — | — | — | — | 0.6 | 0 | 0 | 30.0 s |
| Sustained decode | 248K | 2 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 4 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 248K | 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.