Speed test matrix
Qwen3-Coder-30B-A3B-Instruct-AWQ-4bit on RTX 3090 24GB
vllm · 0.32.0 · AWQ 4-bit6 contexts (0–136K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K
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 4-bit
- 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), 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 cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-4bit --max-model-len 147456 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --kv-cache-dtype int4_per_token_head --attention-backend TRITON_ATTN
Prefill
prompt tok/s by context length8K context3,5312,337 ms TTFT
16K context2,2277,342 ms TTFT
32K context1,27525,531 ms TTFT
64K context686.094,695 ms TTFT
128K context358.0362,642 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 | 122.530 ms TTFT | 213.245 ms TTFT | 363.344 ms TTFT | 591.649 ms TTFT | 933.673 ms TTFT | Over capacity | Over capacity | Over capacity |
| 16K | 73.1101 ms TTFT | Looped | 194.6223 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | 51.7179 ms TTFT | 86.4267 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 32.3346 ms TTFT | 52.6518 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 18.4Requests queued688 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 136K | 17.4Requests queued732 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 | 3,531.0 | — | 2,337 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 2,227.0 | — | 7,342 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 1,275.0 | — | 25,531 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 686.0 | — | 94,695 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 358.0 | — | 362,642 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 122.5 | 124.6 | 30 ms | 8.0 ms | 1.0 | 0 | 3,673 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 213.2 | 108.6 | 45 ms | 9.2 ms | 2.0 | 0 | 6,396 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 363.3 | 92.7 | 44 ms | 10.8 ms | 4.0 | 0 | 10,896 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 591.6 | 76.0 | 49 ms | 13.2 ms | 8.0 | 0 | 17,744 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 933.6 | 60.5 | 73 ms | 16.5 ms | 16.0 | 0 | 28,000 | 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 | 73.1 | 73.4 | 101 ms | 13.6 ms | 1.0 | 0 | 2,193 | 30.0 s |
| Sustained decode | 16K | 2 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 16K | 4 | Valid | 194.6 | 49.2 | 223 ms | 20.3 ms | 4.0 | 0 | 5,836 | 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 | Valid | 51.7 | 51.8 | 179 ms | 19.3 ms | 1.0 | 0 | 1,550 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 86.4 | 43.3 | 267 ms | 23.1 ms | 2.0 | 0 | 2,592 | 30.0 s |
| 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 | 32.3 | 32.4 | 346 ms | 30.9 ms | 1.0 | 0 | 969 | 30.0 s |
| Sustained decode | 64K | 2 | Valid | 52.6 | 26.3 | 518 ms | 38.0 ms | 2.0 | 0 | 1,578 | 30.0 s |
| 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 | 18.4 | 18.5 | 688 ms | 54.1 ms | 1.0 | 0 | 552 | 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 | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 1 | Requests queued | 17.4 | 17.5 | 732 ms | 57.1 ms | 1.0 | 0 | 522 | 30.0 s |
| Sustained decode | 136K | 2 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 4 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 136K | 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.