Speed test matrix
deepseek-r1-distill-qwen-14b-awq on RTX 3090 24GB
vllm · 0.32.0 · AWQ4 contexts (0–48K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 48K
Hardware
- Type
- Discrete GPU
- GPU
- RTX 3090
- VRAM
- 24 GB
- CPU
- AMD Ryzen 9 7900X (12C)
- RAM
- 32 GB
- OS
- Ubuntu 24.04.5
Engine
- Engine
- vllm
- Version
- 0.32.0
- Quantization
- AWQ
- 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 # 1x RTX 3090 24GB (Zagitta/gpu-unlock), 350 W limit (VBIOS max), PCIe Gen4 x16 | AMD Ryzen 9 7900X 12C | 32 GB DDR5 | Ubuntu 24.04.5, NVIDIA driver 595.99.02 vllm serve casperhansen/deepseek-r1-distill-qwen-14b-awq --max-model-len 57344 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code
Prefill
prompt tok/s by context length8K context3,1882,584 ms TTFT
16K context2,7765,883 ms TTFT
32K context2,21714,675 ms TTFT
48K context1,84426,425 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 | 80.039 ms TTFT | 153.744 ms TTFT | 291.633 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 16K | 61.763 ms TTFT | 102.189 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | 50.6105 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 48K | 42.6152 ms TTFT | Over capacity | Over capacity | Over capacity | 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 | 3,188.0 | — | 2,584 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 2,776.0 | — | 5,883 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 2,217.0 | — | 14,675 ms | — | — | 1 | — | — |
| Prefill | 48K | 1 | Valid | 1,844.0 | — | 26,425 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 80.0 | 80.4 | 39 ms | 12.4 ms | 1.0 | 0 | 2,399 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 153.7 | 77.6 | 44 ms | 12.9 ms | 2.0 | 0 | 4,610 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 291.6 | 74.1 | 33 ms | 13.5 ms | 4.0 | 0 | 8,748 | 30.0 s |
| Sustained decode | 0 | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 61.7 | 61.8 | 63 ms | 16.2 ms | 1.0 | 0 | 1,852 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 102.1 | 51.1 | 89 ms | 19.6 ms | 2.0 | 0 | 3,062 | 30.0 s |
| Sustained decode | 16K | 4 | Over capacity | — | — | — | — | — | — | — | — |
| 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 | Valid | 50.6 | 50.6 | 105 ms | 19.7 ms | 1.0 | 0 | 1,517 | 30.0 s |
| Sustained decode | 32K | 2 | Over capacity | — | — | — | — | — | — | — | — |
| 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 | 48K | 1 | Valid | 42.6 | 42.6 | 152 ms | 23.5 ms | 1.0 | 0 | 1,276 | 30.0 s |
| Sustained decode | 48K | 2 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 48K | 4 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 48K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 48K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 48K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 48K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 48K | 128 | Over capacity | — | — | — | — | — | — | — | — |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.