Speed test matrix
Qwen3-1.7B-FP8 on CMP 170HX 64GB
vllm · 0.32.0rc1.dev249+g18305948a · FP83 contexts (0–32K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K
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 Qwen/Qwen3-1.7B-FP8 --max-model-len 40960 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code
Prefill
prompt tok/s by context length8K context18K455 ms TTFT
16K context16.1K1,012 ms TTFT
32K context13K2,496 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 | Looped | 452.887 ms TTFT | 769.875 ms TTFT | 1,243.5113 ms TTFT | 1,879.1167 ms TTFT | Looped | 3,788.8387 ms TTFT | Over capacity |
| 16K | Looped | 290.9213 ms TTFT | 416.3266 ms TTFT | 536.9372 ms TTFT | 634.6622 ms TTFT | Over capacity | Over capacity | Over capacity |
| 32K | 147.4251 ms TTFT | 204.1335 ms TTFT | 264.1461 ms TTFT | 315.1774 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
Looped: the model kept repeating the same text, so the benchmark discarded this measurement. 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 | 18K | — | 455 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 16.1K | — | 1,012 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 13K | — | 2,496 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 0 | 2 | Valid | 452.8 | 260.4 | 87 ms | 3.9 ms | 2.0 | 2 | 13,584 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 769.8 | 204.0 | 75 ms | 4.9 ms | 4.0 | 0 | 23,096 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 1,243.5 | 169.3 | 113 ms | 5.9 ms | 8.0 | 0 | 37,304 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 1,879.1 | 132.5 | 167 ms | 7.5 ms | 16.0 | 0 | 56,368 | 30.0 s |
| Sustained decode | 0 | 32 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 0 | 64 | Valid | 3,788.8 | 68.1 | 387 ms | 14.7 ms | 64.0 | 0 | 113,710 | 30.0 s |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 16K | 2 | Valid | 290.9 | 147.1 | 213 ms | 6.8 ms | 2.0 | 0 | 8,726 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 416.3 | 105.2 | 266 ms | 9.5 ms | 4.0 | 0 | 12,488 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 536.9 | 67.8 | 372 ms | 14.8 ms | 8.0 | 0 | 16,104 | 30.0 s |
| Sustained decode | 16K | 16 | Valid | 634.6 | 39.9 | 622 ms | 25.1 ms | 16.0 | 0 | 19,040 | 30.0 s |
| Sustained decode | 16K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 1 | Valid | 147.4 | 148.4 | 251 ms | 6.7 ms | 1.0 | 0 | 4,420 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 204.1 | 102.8 | 335 ms | 9.7 ms | 2.0 | 0 | 6,122 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 264.1 | 66.4 | 461 ms | 15.1 ms | 4.0 | 0 | 7,922 | 30.0 s |
| Sustained decode | 32K | 8 | Valid | 315.1 | 39.6 | 774 ms | 25.3 ms | 8.0 | 0 | 9,448 | 30.0 s |
| Sustained decode | 32K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 128 | Over capacity | — | — | — | — | — | — | — | — |
Looped: the model kept repeating the same text, so the benchmark discarded this measurement. Over capacity: not run because this many requests at this context would exceed the server's KV cache.