Speed test matrix
gemma-4-12B-it-qat-w4a16-ct on RTX 3090 24GB
vllm · 0.32.0 · W4A166 contexts (0–248K) × 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
- W4A16
- 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), 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 google/gemma-4-12B-it-qat-w4a16-ct --max-model-len 262144 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only
Prefill
prompt tok/s by context length8K context3,7522,136 ms TTFT
16K context3,1864,983 ms TTFT
32K context2,53512,463 ms TTFT
64K context1,78535,308 ms TTFT
128K context1,121112,367 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 | 82.336 ms TTFT | 158.229 ms TTFT | 293.443 ms TTFT | 512.969 ms TTFT | 820.497 ms TTFT | 1,200.2156 ms TTFT | 1,550.1Requests queued189 ms TTFT | Over capacity |
| 16K | 69.384 ms TTFT | Looped | Looped | Looped | Looped | Over capacity | Over capacity | Over capacity |
| 32K | Looped | Looped | Looped | Looped | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 56.8308 ms TTFT | Looped | 210.5595 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 47.0697 ms TTFT | Looped | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 248K | 34.7Requests queued1,386 ms TTFT | 67.8Requests queued1,867 ms TTFT | 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 | 3,752.0 | — | 2,136 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 3,186.0 | — | 4,983 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 2,535.0 | — | 12,463 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 1,785.0 | — | 35,308 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 1,121.0 | — | 112,367 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 82.3 | 82.8 | 36 ms | 12.1 ms | 1.0 | 0 | 2,470 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 158.2 | 79.9 | 29 ms | 12.5 ms | 2.0 | 0 | 4,746 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 293.4 | 74.5 | 43 ms | 13.4 ms | 4.0 | 0 | 8,800 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 512.9 | 65.9 | 69 ms | 15.2 ms | 8.0 | 0 | 15,384 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 820.4 | 53.6 | 97 ms | 18.6 ms | 16.0 | 0 | 24,608 | 30.0 s |
| Sustained decode | 0 | 32 | Valid | 1,200.2 | 40.7 | 156 ms | 24.6 ms | 32.0 | 0 | 36,000 | 30.0 s |
| Sustained decode | 0 | 64 | Requests queued | 1,550.1 | 32.6 | 189 ms | 30.7 ms | 49.9 | 0 | 46,506 | 30.0 s |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 69.3 | 69.5 | 84 ms | 14.4 ms | 1.0 | 0 | 2,077 | 30.0 s |
| Sustained decode | 16K | 2 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 16K | 4 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 16K | 8 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 16K | 16 | Looped | — | — | — | — | — | 0 | 0 | — |
| 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 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 32K | 8 | Looped | — | — | — | — | — | 0 | 0 | — |
| 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 | 56.8 | 56.9 | 308 ms | 17.6 ms | 1.0 | 0 | 1,703 | 30.0 s |
| Sustained decode | 64K | 2 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 64K | 4 | Valid | 210.5 | 52.7 | 595 ms | 19.0 ms | 4.0 | 0 | 6,312 | 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 | Valid | 47.0 | 47.2 | 697 ms | 21.2 ms | 1.0 | 0 | 1,411 | 30.0 s |
| Sustained decode | 128K | 2 | Looped | — | — | — | — | — | 0 | 0 | — |
| 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 | Requests queued | 34.7 | 35.8 | 1,386 ms | 28.0 ms | 1.0 | 0 | 1,042 | 30.0 s |
| Sustained decode | 248K | 2 | Requests queued | 67.8 | 35.5 | 1,867 ms | 28.2 ms | 1.9 | 0 | 2,034 | 30.0 s |
| 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. 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.