Speed test matrix
Fara1.5-4B-AutoRound-W4A16-RTN on RTX 3060 12GB
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 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
- 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. 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 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 --kv-cache-dtype int4_per_token_head --attention-backend TRITON_ATTN
Prefill
prompt tok/s by context length8K context2,3473,472 ms TTFT
16K context1,9578,237 ms TTFT
32K context1,48321,642 ms TTFT
64K context994.064,442 ms TTFT
128K context601.0212,786 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 | 83.185 ms TTFT | 159.399 ms TTFT | 294.2203 ms TTFT | 511.3358 ms TTFT | 802.3615 ms TTFT | 1,053.31,002 ms TTFT | 1,273.21,892 ms TTFT | Over capacity |
| 16K | 70.31,282 ms TTFT | 122.91,892 ms TTFT | 193.23,080 ms TTFT | 279.15,465 ms TTFT | 353.710,312 ms TTFT | Over capacity | Over capacity | Over capacity |
| 32K | 60.81,527 ms TTFT | 99.62,265 ms TTFT | 142.93,656 ms TTFT | 190.86,465 ms TTFT | 226.012,192 ms TTFT | Over capacity | Over capacity | Over capacity |
| 64K | 47.91,282 ms TTFT | 72.61,775 ms TTFT | 94.52,766 ms TTFT | 116.94,759 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 128K | 30.7Requests queued3,439 ms TTFT | 37.9Requests queued5,014 ms TTFT | 55.77,956 ms TTFT | 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 | 2,347.0 | — | 3,472 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 1,957.0 | — | 8,237 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 1,483.0 | — | 21,642 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 994.0 | — | 64,442 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 601.0 | — | 212,786 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 83.1 | 83.4 | 85 ms | 12.0 ms | 1.0 | 0 | 2,492 | 30.0 s |
| Sustained decode | 0 | 2 | Valid | 159.3 | 80.1 | 99 ms | 12.5 ms | 2.0 | 0 | 4,780 | 30.0 s |
| Sustained decode | 0 | 4 | Valid | 294.2 | 74.1 | 203 ms | 13.5 ms | 4.0 | 0 | 8,824 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 511.3 | 64.6 | 358 ms | 15.5 ms | 8.0 | 0 | 15,336 | 30.0 s |
| Sustained decode | 0 | 16 | Valid | 802.3 | 50.9 | 615 ms | 19.7 ms | 16.0 | 0 | 24,080 | 30.0 s |
| Sustained decode | 0 | 32 | Valid | 1,053.3 | 33.2 | 1,002 ms | 30.1 ms | 32.0 | 0 | 31,584 | 30.0 s |
| Sustained decode | 0 | 64 | Valid | 1,273.2 | 19.8 | 1,892 ms | 50.4 ms | 64.0 | 0 | 38,208 | 30.0 s |
| Sustained decode | 0 | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 70.3 | 70.4 | 1,282 ms | 14.2 ms | 1.0 | 0 | 2,107 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 122.9 | 60.6 | 1,892 ms | 16.5 ms | 2.0 | 0 | 3,686 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 193.2 | 46.2 | 3,080 ms | 21.7 ms | 4.0 | 0 | 5,791 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 279.1 | 31.3 | 5,465 ms | 32.0 ms | 8.0 | 0 | 8,368 | 30.0 s |
| Sustained decode | 16K | 16 | Valid | 353.7 | 17.7 | 10,312 ms | 56.6 ms | 16.0 | 0 | 10,608 | 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 | Valid | 60.8 | 60.9 | 1,527 ms | 16.4 ms | 1.0 | 0 | 1,822 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 99.6 | 49.0 | 2,265 ms | 20.4 ms | 2.0 | 0 | 2,988 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 142.9 | 33.9 | 3,656 ms | 29.5 ms | 4.0 | 0 | 4,284 | 30.0 s |
| Sustained decode | 32K | 8 | Valid | 190.8 | 21.0 | 6,465 ms | 47.6 ms | 8.0 | 0 | 5,720 | 30.0 s |
| Sustained decode | 32K | 16 | Valid | 226.0 | 11.0 | 12,192 ms | 90.6 ms | 16.0 | 0 | 6,768 | 29.9 s |
| Sustained decode | 32K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 32K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 64K | 1 | Valid | 47.9 | 48.1 | 1,282 ms | 20.8 ms | 1.0 | 0 | 1,437 | 30.0 s |
| Sustained decode | 64K | 2 | Valid | 72.6 | 35.9 | 1,775 ms | 27.8 ms | 2.0 | 0 | 2,178 | 30.0 s |
| Sustained decode | 64K | 4 | Valid | 94.5 | 22.8 | 2,766 ms | 43.9 ms | 4.0 | 0 | 2,832 | 30.0 s |
| Sustained decode | 64K | 8 | Valid | 116.9 | 13.5 | 4,759 ms | 74.3 ms | 8.0 | 0 | 3,504 | 30.0 s |
| 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 | 30.7 | 34.0 | 3,439 ms | 29.4 ms | 0.9 | 0 | 921 | 30.0 s |
| Sustained decode | 128K | 2 | Requests queued | 37.9 | 22.4 | 5,014 ms | 44.8 ms | 1.8 | 0 | 1,138 | 30.0 s |
| Sustained decode | 128K | 4 | Valid | 55.7 | 12.4 | 7,956 ms | 80.5 ms | 4.0 | 0 | 1,668 | 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 | Over capacity | — | — | — | — | 0.7 | 0 | 0 | 30.0 s |
| Sustained decode | 248K | 2 | Over capacity | — | — | — | — | 0.7 | 0 | 0 | 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. Over capacity: not run because this many requests at this context would exceed the server's KV cache.