Speed test matrix
Cosmos-Reason2-8B on RTX 3090 24GB
vllm · 0.32.0 · BF165 contexts (0–128K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K
Hardware
- Type
- Discrete GPU
- GPU
- RTX 3090
- VRAM
- 24 GB
- CPU
- Intel Core i7-9700K (8C)
- RAM
- 32 GB
- OS
- Ubuntu 24.04.5
Engine
- Engine
- vllm
- Version
- 0.32.0
- Quantization
- BF16
- Backend
- cuda
- Repository
- github.com/vllm-project/vllm
- Build
- vllm/vllm-openai:v0.32.0-x86_64
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. KV cache: int4_per_token_head # 1x RTX 3090 24GB (Zagitta/gpu-unlock), 420 W limit, PCIe Gen3 x16 | Intel Core i7-9700K 8C | 32 GB DDR4 | Ubuntu 24.04.5, NVIDIA driver 580.178.04 vllm serve nvidia/Cosmos-Reason2-8B --max-model-len 139264 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only --kv-cache-dtype int4_per_token_head --attention-backend TRITON_ATTN
Prefill
prompt tok/s by context length8K context3,9322,099 ms TTFT
16K context2,7176,016 ms TTFT
32K context1,69019,262 ms TTFT
64K context957.067,831 ms TTFT
128K context510.0254,600 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 | 47.738 ms TTFT | 95.142 ms TTFT | 184.872 ms TTFT | 353.489 ms TTFT | 644.497 ms TTFT | Over capacity | Over capacity | Over capacity |
| 16K | 38.4109 ms TTFT | 72.8148 ms TTFT | 117.1225 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | Looped | Looped | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 24.1357 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 16.0Requests queued731 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. 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 | 3,932.0 | — | 2,099 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 2,717.0 | — | 6,016 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 1,690.0 | — | 19,262 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 957.0 | — | 67,831 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 510.0 | — | 254,600 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 47.7 | 47.8 | 38 ms | 20.9 ms | 1.0 | 0 | 1,432 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 95.1 | 47.7 | 42 ms | 21.0 ms | 2.0 | 0 | 2,852 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 184.8 | 46.4 | 72 ms | 21.5 ms | 4.0 | 0 | 5,544 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 353.4 | 44.6 | 89 ms | 22.4 ms | 8.0 | 0 | 10,603 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 644.4 | 40.9 | 97 ms | 24.4 ms | 16.0 | 0 | 19,328 | 30.0 s |
| Sustained decode | 0 | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 38.4 | 38.4 | 109 ms | 26.0 ms | 1.0 | 0 | 1,151 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 72.8 | 36.4 | 148 ms | 27.5 ms | 2.0 | 0 | 2,182 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 117.1 | 29.3 | 225 ms | 34.1 ms | 4.0 | 0 | 3,512 | 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 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 32K | 2 | Looped | — | — | — | — | — | 0 | 0 | — |
| 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 | 24.1 | 24.1 | 357 ms | 41.4 ms | 1.0 | 0 | 723 | 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 | 128K | 1 | Requests queued | 16.0 | 16.1 | 731 ms | 62.0 ms | 1.0 | 0 | 481 | 30.0 s |
| Sustained decode | 128K | 2 | Over capacity | — | — | — | — | — | — | — | — |
| 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 | — | — | — | — | — | — | — | — |
Requests queued: the server could not run all requests at once, so some waited in a queue. 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.