Speed test matrix
Qwen3.5-9B-AWQ-4bit on RTX 3060 12GB
vllm · 0.32.0 · W4A16 g326 contexts (0–248K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K
Hardware
- Type
- Discrete GPU
- GPU
- RTX 3060
- VRAM
- 12 GB
- CPU
- Intel Xeon E5-2680 v4 (14C)
- RAM
- 32 GB
- OS
- Ubuntu 24.04.5
Engine
- Engine
- vllm
- Version
- 0.32.0
- Quantization
- W4A16 g32
- 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 3060 12GB, stock 170 W limit, PCIe Gen3 x16 | Intel Xeon E5-2680 v4 14C | 32 GB DDR4-2400 | Ubuntu 24.04.5, NVIDIA driver 580.178.04 vllm serve cyankiwi/Qwen3.5-9B-AWQ-4bit --max-model-len 262144 --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 context1,4735,530 ms TTFT
16K context1,28712,521 ms TTFT
32K context1,06630,092 ms TTFT
64K context789.081,134 ms TTFT
128K context522.0245,157 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 | 49.293 ms TTFT | 95.7340 ms TTFT | 182.7293 ms TTFT | 335.1503 ms TTFT | 567.1924 ms TTFT | 775.51,575 ms TTFT | Over capacity | Over capacity |
| 16K | 44.31,746 ms TTFT | 80.82,614 ms TTFT | 137.14,270 ms TTFT | 215.27,629 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | 40.31,875 ms TTFT | 70.12,804 ms TTFT | 109.44,583 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 34.21,424 ms TTFT | 55.41,985 ms TTFT | 78.53,156 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 23.7Requests queued3,710 ms TTFT | 30.9Requests queued5,408 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 248K | Over capacity | 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 | 1,473.0 | — | 5,530 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 1,287.0 | — | 12,521 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 1,066.0 | — | 30,092 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 789.0 | — | 81,134 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 522.0 | — | 245,157 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 49.2 | 49.3 | 93 ms | 20.3 ms | 1.0 | 0 | 1,475 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 95.7 | 48.0 | 340 ms | 20.8 ms | 2.0 | 0 | 2,870 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 182.7 | 45.8 | 293 ms | 21.8 ms | 4.0 | 0 | 5,480 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 335.1 | 42.1 | 503 ms | 23.8 ms | 8.0 | 0 | 10,048 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 567.1 | 35.7 | 924 ms | 28.0 ms | 16.0 | 0 | 17,008 | 30.0 s |
| Sustained decode | 0 | 32 | Valid | 775.5 | 24.2 | 1,575 ms | 41.3 ms | 32.0 | 0 | 23,264 | 30.0 s |
| Sustained decode | 0 | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 44.3 | 44.4 | 1,746 ms | 22.5 ms | 1.0 | 0 | 1,329 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 80.8 | 39.5 | 2,614 ms | 25.3 ms | 2.0 | 0 | 2,424 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 137.1 | 32.1 | 4,270 ms | 31.2 ms | 4.0 | 0 | 4,112 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 215.2 | 23.1 | 7,629 ms | 43.3 ms | 8.0 | 0 | 6,456 | 30.0 s |
| 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 | 40.3 | 40.4 | 1,875 ms | 24.8 ms | 1.0 | 0 | 1,209 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 70.1 | 34.3 | 2,804 ms | 29.2 ms | 2.0 | 0 | 2,104 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 109.4 | 25.5 | 4,583 ms | 39.2 ms | 4.0 | 0 | 3,280 | 30.0 s |
| 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 | 34.2 | 34.3 | 1,424 ms | 29.1 ms | 1.0 | 0 | 1,027 | 30.0 s |
| Sustained decode | 64K | 2 | Valid | 55.4 | 27.4 | 1,985 ms | 36.5 ms | 2.0 | 0 | 1,662 | 30.0 s |
| Sustained decode | 64K | 4 | Valid | 78.5 | 18.8 | 3,156 ms | 53.2 ms | 4.0 | 0 | 2,352 | 30.0 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 | Requests queued | 23.7 | 26.5 | 3,710 ms | 37.7 ms | 0.9 | 0 | 710 | 30.0 s |
| Sustained decode | 128K | 2 | Requests queued | 30.9 | 18.5 | 5,408 ms | 54.3 ms | 1.7 | 0 | 926 | 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 | Over capacity | — | — | — | — | 0.7 | 0 | 0 | 30.0 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.