Speed test matrix
Fara1.5-4B-AutoRound-W4A16-RTN on RTX 3090 24GB
vllm · 0.32.0 · AutoRound 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
- AMD Ryzen 9 7900X (12C)
- RAM
- 32 GB
- OS
- Ubuntu 24.04.5
Engine
- Engine
- vllm
- Version
- 0.32.0
- Quantization
- AutoRound 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), 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 LeaderboardModel1/Fara1.5-4B-AutoRound-W4A16-RTN --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 context10.6K772 ms TTFT
16K context9,9931,623 ms TTFT
32K context8,8863,634 ms TTFT
64K context7,1968,952 ms TTFT
128K context5,32224,182 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 | Looped | 343.032 ms TTFT | 638.852 ms TTFT | 1,109.192 ms TTFT | 1,763.2137 ms TTFT | Looped | 3,160.9449 ms TTFT | Over capacity |
| 16K | 161.279 ms TTFT | 289.7109 ms TTFT | 521.3177 ms TTFT | 829.0300 ms TTFT | 1,125.3584 ms TTFT | Over capacity | Over capacity | Over capacity |
| 32K | Looped | 250.197 ms TTFT | 432.0145 ms TTFT | 636.2238 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 125.3132 ms TTFT | 196.2164 ms TTFT | 318.1257 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 96.6354 ms TTFT | 137.1431 ms TTFT | 206.6637 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 248K | 67.4678 ms TTFT | 87.5852 ms TTFT | 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 | 10.6K | — | 772 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 9,993.0 | — | 1,623 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 8,886.0 | — | 3,634 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 7,196.0 | — | 8,952 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 5,322.0 | — | 24,182 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 0 | 2 | Valid | 343.0 | 172.9 | 32 ms | 5.8 ms | 2.0 | 0 | 10,290 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 638.8 | 161.7 | 52 ms | 6.2 ms | 4.0 | 0 | 19,160 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 1,109.1 | 141.2 | 92 ms | 7.1 ms | 8.0 | 0 | 33,272 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 1,763.2 | 112.8 | 137 ms | 8.9 ms | 16.0 | 0 | 52,897 | 30.0 s |
| Sustained decode | 0 | 32 | Looped | — | — | — | — | — | 0 | 0 | — |
| Sustained decode | 0 | 64 | Valid | 3,160.9 | 50.3 | 449 ms | 19.9 ms | 64.0 | 0 | 94,784 | 30.0 s |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 161.2 | 161.6 | 79 ms | 6.2 ms | 1.0 | 0 | 4,835 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 289.7 | 145.2 | 109 ms | 6.9 ms | 2.0 | 0 | 8,690 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 521.3 | 130.8 | 177 ms | 7.6 ms | 4.0 | 0 | 15,640 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 829.0 | 103.7 | 300 ms | 9.6 ms | 8.0 | 0 | 24,864 | 30.0 s |
| Sustained decode | 16K | 16 | Valid | 1,125.3 | 69.8 | 584 ms | 14.3 ms | 16.0 | 0 | 33,744 | 30.0 s |
| 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 | Valid | 250.1 | 125.3 | 97 ms | 8.0 ms | 2.0 | 0 | 7,502 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 432.0 | 108.3 | 145 ms | 9.2 ms | 4.0 | 0 | 12,956 | 30.0 s |
| Sustained decode | 32K | 8 | Valid | 636.2 | 79.6 | 238 ms | 12.6 ms | 8.0 | 0 | 19,080 | 30.0 s |
| 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 | 125.3 | 125.5 | 132 ms | 8.0 ms | 1.0 | 0 | 3,760 | 30.0 s |
| Sustained decode | 64K | 2 | Valid | 196.2 | 98.2 | 164 ms | 10.2 ms | 2.0 | 0 | 5,884 | 30.0 s |
| Sustained decode | 64K | 4 | Valid | 318.1 | 79.7 | 257 ms | 12.6 ms | 4.0 | 0 | 9,540 | 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 | 96.6 | 96.8 | 354 ms | 10.3 ms | 1.0 | 0 | 2,898 | 30.0 s |
| Sustained decode | 128K | 2 | Valid | 137.1 | 68.4 | 431 ms | 14.6 ms | 2.0 | 0 | 4,110 | 30.0 s |
| Sustained decode | 128K | 4 | Valid | 206.6 | 51.5 | 637 ms | 19.4 ms | 4.0 | 0 | 6,196 | 30.0 s |
| 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 | Valid | 67.4 | 67.5 | 678 ms | 14.8 ms | 1.0 | 0 | 2,023 | 30.0 s |
| Sustained decode | 248K | 2 | Valid | 87.5 | 43.6 | 852 ms | 22.9 ms | 2.0 | 0 | 2,624 | 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 | — | — | — | — | — | — | — | — |
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.