Speed test matrix
Qwen3-Coder-30B-A3B-Instruct-AWQ-4bit on RTX 3090 24GB
vllm · 0.32.0 · AWQ 4-bit4 contexts (0–64K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K
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
- 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: fp8 # 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 cyankiwi/Qwen3-Coder-30B-A3B-Instruct-AWQ-4bit --max-model-len 73728 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --kv-cache-dtype fp8
Prefill
prompt tok/s by context length8K context9,630857 ms TTFT
16K context7,3742,217 ms TTFT
32K context5,1676,299 ms TTFT
64K context3,13320,729 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 | 179.626 ms TTFT | 298.229 ms TTFT | 488.834 ms TTFT | 780.842 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 16K | 153.653 ms TTFT | 261.475 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 32K | Looped | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 108.2169 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
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 | 9,630.0 | — | 857 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 7,374.0 | — | 2,217 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 5,167.0 | — | 6,299 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 3,133.0 | — | 20,729 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 179.6 | 180.4 | 26 ms | 5.5 ms | 1.0 | 0 | 5,387 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 298.2 | 150.6 | 29 ms | 6.6 ms | 2.0 | 0 | 8,946 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 488.8 | 123.9 | 34 ms | 8.1 ms | 4.0 | 0 | 14,663 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 780.8 | 99.7 | 42 ms | 10.0 ms | 8.0 | 0 | 23,424 | 30.0 s |
| 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 | 153.6 | 153.8 | 53 ms | 6.5 ms | 1.0 | 0 | 4,608 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 261.4 | 131.1 | 75 ms | 7.6 ms | 2.0 | 0 | 7,840 | 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 | Looped | — | — | — | — | — | 0 | 0 | — |
| 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 | 64K | 1 | Valid | 108.2 | 108.4 | 169 ms | 9.2 ms | 1.0 | 0 | 3,245 | 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 | — | — | — | — | — | — | — | — |
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.