Speed test matrix
Tess-4.5-27B-FP8 on CMP 170HX 64GB
vllm · 0.32.0rc1.dev249+g18305948a · FP86 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
- 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: MTP, 2 tokens lookahead
# 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 vwdubb/Tess-4.5-27B-FP8 --max-model-len 262144 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only --speculative-config {"method":"mtp","num_speculative_tokens":2}Prefill
prompt tok/s by context length8K context2,0623,972 ms TTFT
16K context1,9558,267 ms TTFT
32K context1,82617,593 ms TTFT
64K context1,60040,038 ms TTFT
128K context1,28099,945 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 | 59.0188 ms TTFT | 110.0374 ms TTFT | 214.5452 ms TTFT | 318.8706 ms TTFT | 579.51,165 ms TTFT | 724.51,770 ms TTFT | Over capacity | Over capacity |
| 16K | 41.9645 ms TTFT | 77.91,130 ms TTFT | 128.62,138 ms TTFT | 175.03,515 ms TTFT | 237.15,767 ms TTFT | Over capacity | Over capacity | Over capacity |
| 32K | 35.1797 ms TTFT | 66.51,325 ms TTFT | 91.32,426 ms TTFT | 122.73,948 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 24.71,019 ms TTFT | 46.31,639 ms TTFT | 55.52,900 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 16.72,305 ms TTFT | 33.43,790 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 248K | 8.4Requests queued3,623 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 | 2,062.0 | — | 3,972 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 1,955.0 | — | 8,267 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 1,826.0 | — | 17,593 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 1,600.0 | — | 40,038 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 1,280.0 | — | 99,945 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 59.0 | 60.3 | 188 ms | 16.6 ms | 1.0 | 0 | 1,771 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 110.0 | 56.0 | 374 ms | 17.8 ms | 2.0 | 0 | 3,299 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 214.5 | 54.4 | 452 ms | 18.4 ms | 4.0 | 0 | 6,431 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 318.8 | 40.5 | 706 ms | 24.7 ms | 8.0 | 0 | 9,564 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 579.5 | 36.7 | 1,165 ms | 27.3 ms | 16.0 | 0 | 17,361 | 30.0 s |
| Sustained decode | 0 | 32 | Valid | 724.5 | 23.0 | 1,770 ms | 43.5 ms | 32.0 | 0 | 21,720 | 30.0 s |
| Sustained decode | 0 | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 41.9 | 42.7 | 645 ms | 23.4 ms | 1.0 | 0 | 1,255 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 77.9 | 39.6 | 1,130 ms | 25.3 ms | 2.0 | 0 | 2,337 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 128.6 | 32.2 | 2,138 ms | 31.0 ms | 4.0 | 0 | 3,857 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 175.0 | 22.7 | 3,515 ms | 44.1 ms | 8.0 | 0 | 5,238 | 29.9 s |
| Sustained decode | 16K | 16 | Valid | 237.1 | 14.4 | 5,767 ms | 69.6 ms | 16.0 | 0 | 7,098 | 29.9 s |
| Sustained decode | 16K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 1 | Valid | 35.1 | 35.5 | 797 ms | 28.2 ms | 1.0 | 0 | 1,050 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 66.5 | 33.7 | 1,325 ms | 29.7 ms | 2.0 | 0 | 1,991 | 29.9 s |
| Sustained decode | 32K | 4 | Valid | 91.3 | 23.3 | 2,426 ms | 42.9 ms | 4.0 | 0 | 2,735 | 30.0 s |
| Sustained decode | 32K | 8 | Valid | 122.7 | 15.5 | 3,948 ms | 64.6 ms | 8.0 | 0 | 3,679 | 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 | — | — | — | — | — | — | — | — |
| Sustained decode | 64K | 1 | Valid | 24.7 | 25.1 | 1,019 ms | 39.8 ms | 1.0 | 0 | 738 | 29.9 s |
| Sustained decode | 64K | 2 | Valid | 46.3 | 23.8 | 1,639 ms | 42.1 ms | 2.0 | 0 | 1,387 | 30.0 s |
| Sustained decode | 64K | 4 | Valid | 55.5 | 14.3 | 2,900 ms | 69.9 ms | 4.0 | 0 | 1,661 | 29.9 s |
| 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 | Valid | 16.7 | 17.0 | 2,305 ms | 58.8 ms | 1.0 | 0 | 502 | 30.0 s |
| Sustained decode | 128K | 2 | Valid | 33.4 | 16.5 | 3,790 ms | 61.4 ms | 2.0 | 0 | 1,000 | 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 | Requests queued | 8.4 | 9.3 | 3,623 ms | 107.6 ms | 0.9 | 0 | 249 | 29.8 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.