Speed test matrix
Qwen3.8-27B-W3Rot-INT3-Paiton-RDNA4 on Radeon AI Pro R9700 32GB
vllm · 0.29.0 · W3A43 contexts (0–228K) × 2 concurrency levels (1–8) · prefill at 8K, 64K, 128K
Hardware
- Type
- Discrete GPU
- GPU
- Radeon AI Pro R9700
- VRAM
- 32 GB
- CPU
- AMD Ryzen Threadripper 3960X 24-Core Processor
- RAM
- 251 GB
- OS
- Linux
Engine
- Engine
- vllm
- Version
- 0.29.0
- Quantization
- W3A4
Run
- Served model
- Qwen3.8
- Revision
- main
- Benchmark tool
- llm-inference-bench v0.7.7
- Decode test
- Fixed duration
- Prefill mode
- Measured during decode warm-up
- Time per measurement
- 30 s
- Max output
- 8,192 tokens
Prefill
prompt tok/s by context length8K context3,9732,062 ms TTFT
64K context3,40818,898 ms TTFT
128K context2,80645,849 ms TTFT
Sustained decode
total tok/s · rows are context length, columns are simultaneous requests| Context | 1 | 8 |
|---|---|---|
| 0 | 125.788 ms TTFT | 543.7308 ms TTFT |
| 128K | 103.81,707 ms TTFT | Over capacity |
| 228K | 87.32,597 ms TTFT | Over capacity |
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 | 3,973.0 | — | 2,062 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 3,408.0 | — | 18,898 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 2,806.0 | — | 45,849 ms | — | — | 1 | — | — |
| Sustained decode | 0 | 1 | Valid | 125.7 | 124.6 | 88 ms | 8.0 ms | 1.0 | 0 | 3,769 | 30.0 s |
| Sustained decode | 0 | 8 | Valid | 543.7 | 67.6 | 308 ms | 14.8 ms | 8.0 | 0 | 16,307 | 30.0 s |
| Sustained decode | 128K | 1 | Valid | 103.8 | 102.8 | 1,707 ms | 9.7 ms | 1.0 | 0 | 3,111 | 30.0 s |
| Sustained decode | 128K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 228K | 1 | Valid | 87.3 | 86.4 | 2,597 ms | 11.6 ms | 1.0 | 0 | 2,616 | 30.0 s |
| Sustained decode | 228K | 8 | Over capacity | — | — | — | — | — | — | — | — |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.