Speed test matrix
Phi-4-reasoning-plus-FP8 on CMP 170HX 64GB
vllm · 0.32.0rc1.dev249+g18305948a · FP83 contexts (0–24K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 24K
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 nvidia/Phi-4-reasoning-plus-FP8 --max-model-len 32768 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code
Prefill
prompt tok/s by context length8K context3,2102,616 ms TTFT
16K context2,8525,754 ms TTFT
24K context2,5889,445 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 | 67.466 ms TTFT | 131.8116 ms TTFT | 255.6125 ms TTFT | 474.5129 ms TTFT | 857.6208 ms TTFT | 1,423.6265 ms TTFT | Over capacity | Over capacity |
| 16K | 59.9145 ms TTFT | 109.8189 ms TTFT | 179.4290 ms TTFT | 226.6447 ms TTFT | 325.0707 ms TTFT | Over capacity | Over capacity | Over capacity |
| 24K | 56.6188 ms TTFT | 101.0290 ms TTFT | 154.4430 ms TTFT | 177.8666 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
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 | 3,210.0 | — | 2,616 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 2,852.0 | — | 5,754 ms | — | — | 1 | — | — |
| Prefill | 24K | 1 | Valid | 2,588.0 | — | 9,445 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 67.4 | 67.6 | 66 ms | 14.8 ms | 1.0 | 0 | 2,022 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 131.8 | 66.1 | 116 ms | 15.1 ms | 2.0 | 0 | 3,952 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 255.6 | 64.4 | 125 ms | 15.5 ms | 4.0 | 0 | 7,668 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 474.5 | 60.2 | 129 ms | 16.6 ms | 8.0 | 0 | 14,232 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 857.6 | 54.7 | 208 ms | 18.3 ms | 16.0 | 0 | 25,728 | 30.0 s |
| Sustained decode | 0 | 32 | Valid | 1,423.6 | 45.7 | 265 ms | 21.9 ms | 32.0 | 0 | 42,720 | 30.0 s |
| Sustained decode | 0 | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 59.9 | 60.1 | 145 ms | 16.7 ms | 1.0 | 0 | 1,797 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 109.8 | 55.0 | 189 ms | 18.2 ms | 2.0 | 0 | 3,292 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 179.4 | 45.0 | 290 ms | 22.2 ms | 4.0 | 0 | 5,381 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 226.6 | 28.4 | 447 ms | 35.2 ms | 8.0 | 0 | 6,792 | 30.0 s |
| Sustained decode | 16K | 16 | Valid | 325.0 | 20.4 | 707 ms | 49.1 ms | 16.0 | 0 | 9,744 | 30.0 s |
| Sustained decode | 16K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 24K | 1 | Valid | 56.6 | 56.8 | 188 ms | 17.6 ms | 1.0 | 0 | 1,697 | 30.0 s |
| Sustained decode | 24K | 2 | Valid | 101.0 | 50.7 | 290 ms | 19.7 ms | 2.0 | 0 | 3,030 | 30.0 s |
| Sustained decode | 24K | 4 | Valid | 154.4 | 38.8 | 430 ms | 25.8 ms | 4.0 | 0 | 4,632 | 30.0 s |
| Sustained decode | 24K | 8 | Valid | 177.8 | 22.3 | 666 ms | 44.8 ms | 8.0 | 0 | 5,328 | 30.0 s |
| Sustained decode | 24K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 24K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 24K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 24K | 128 | Over capacity | — | — | — | — | — | — | — | — |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.