Speed test matrix
granite-4.2-8b on RTX PRO 6000 Blackwell Server Edition 96GB
sglang · 0.5.22.dev220+gfa7784d14 · BF165 contexts (8K–123K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 123K, 128K
Hardware
- Type
- Discrete GPU
- GPU
- RTX PRO 6000 Blackwell Server Edition
- VRAM
- 96 GB
- OS
- linux
Engine
- Engine
- sglang
- Version
- 0.5.22.dev220+gfa7784d14
- Quantization
- BF16
- Backend
- cuda
- Repository
- github.com/sgl-project/sglang
- Build
- cuda-13.2
- Commit
fa7784d140ee
Run
- Served model
- granite-4.2-8b
- Revision
- main
- Benchmark tool
- llm-inference-bench v0.7.6
- Decode test
- Fixed duration
- Prefill mode
- Measured during decode warm-up
- Time per measurement
- 30 s
- Max output
- 4,096 tokens
Launch command
"$HOME/env_rtxpro6k-ibm-granite-sglang-fa7784d140ee/bin/sglang" serve --model-path ibm-granite/granite-4.2-8b --served-model-name granite-4.2-8b --tp 1 --reasoning-parser auto --tool-call-parser auto --context-length 131072 --mem-fraction-static 0.79 --enable-metrics --host 0.0.0.0 --port 8000 --api-key <redacted> --dtype bfloat16 --attention-backend flashinfer --bf16-gemm-backend torch
Prefill
prompt tok/s by context length8K context17.7K463 ms TTFT
16K context15.7K1,035 ms TTFT
32K context12.9K2,492 ms TTFT
64K context9,3776,863 ms TTFT
123K context6,24619,777 ms TTFT
128K context6,08021,144 ms TTFT
Sustained decode
total tok/s · rows are context length, columns are simultaneous requests| Context | 1 | 2 | 4 | 8 | 16 | 32 | 64 | 128 |
|---|---|---|---|---|---|---|---|---|
| 8K | 73.431 ms TTFT | 141.752 ms TTFT | 282.059 ms TTFT | 527.388 ms TTFT | 866.9160 ms TTFT | Over capacity | Over capacity | Over capacity |
| 16K | 69.141 ms TTFT | 133.659 ms TTFT | 265.091 ms TTFT | 474.7144 ms TTFT | 730.7297 ms TTFT | Over capacity | Over capacity | Over capacity |
| 32K | 61.761 ms TTFT | 119.8102 ms TTFT | 237.9153 ms TTFT | 395.5307 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity |
| 64K | 51.0104 ms TTFT | 99.5184 ms TTFT | 197.2309 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity |
| 123K | 38.6201 ms TTFT | 75.6371 ms TTFT | Over capacity | Over capacity | Over capacity | Over capacity | Over capacity | 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 | 17.7K | — | 463 ms | — | — | 1 | — | — |
| Prefill | 16K | 1 | Valid | 15.7K | — | 1,035 ms | — | — | 1 | — | — |
| Prefill | 32K | 1 | Valid | 12.9K | — | 2,492 ms | — | — | 1 | — | — |
| Prefill | 64K | 1 | Valid | 9,377.0 | — | 6,863 ms | — | — | 1 | — | — |
| Prefill | 123K | 1 | Valid | 6,246.0 | — | 19,777 ms | — | — | 1 | — | — |
| Prefill | 128K | 1 | Valid | 6,080.0 | — | 21,144 ms | — | — | 1 | — | — |
| Sustained decode | 8K | 1 | Valid | 73.4 | 73.5 | 31 ms | 13.6 ms | 1.0 | 0 | 2,203 | 30.0 s |
| Sustained decode | 8K | 2 | Valid | 141.7 | 71.0 | 52 ms | 14.1 ms | 2.0 | 0 | 4,248 | 30.0 s |
| Sustained decode | 8K | 4 | Valid | 282.0 | 70.8 | 59 ms | 14.1 ms | 4.0 | 0 | 8,460 | 30.0 s |
| Sustained decode | 8K | 8 | Valid | 527.3 | 66.2 | 88 ms | 15.1 ms | 8.0 | 0 | 15,816 | 30.0 s |
| Sustained decode | 8K | 16 | Valid | 866.9 | 54.5 | 160 ms | 18.3 ms | 16.0 | 0 | 26,000 | 30.0 s |
| Sustained decode | 8K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 8K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 8K | 128 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 16K | 1 | Valid | 69.1 | 69.2 | 41 ms | 14.5 ms | 1.0 | 0 | 2,072 | 30.0 s |
| Sustained decode | 16K | 2 | Valid | 133.6 | 66.9 | 59 ms | 14.9 ms | 2.0 | 0 | 4,008 | 30.0 s |
| Sustained decode | 16K | 4 | Valid | 265.0 | 66.5 | 91 ms | 15.0 ms | 4.0 | 0 | 7,948 | 30.0 s |
| Sustained decode | 16K | 8 | Valid | 474.7 | 59.5 | 144 ms | 16.8 ms | 8.0 | 0 | 14,240 | 30.0 s |
| Sustained decode | 16K | 16 | Valid | 730.7 | 45.9 | 297 ms | 21.8 ms | 16.0 | 0 | 21,920 | 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 | 61.7 | 61.8 | 61 ms | 16.2 ms | 1.0 | 0 | 1,852 | 30.0 s |
| Sustained decode | 32K | 2 | Valid | 119.8 | 60.0 | 102 ms | 16.7 ms | 2.0 | 0 | 3,592 | 30.0 s |
| Sustained decode | 32K | 4 | Valid | 237.9 | 59.7 | 153 ms | 16.8 ms | 4.0 | 0 | 7,136 | 30.0 s |
| Sustained decode | 32K | 8 | Valid | 395.5 | 49.6 | 307 ms | 20.2 ms | 8.0 | 0 | 11,864 | 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 | 51.0 | 51.0 | 104 ms | 19.6 ms | 1.0 | 0 | 1,530 | 30.0 s |
| Sustained decode | 64K | 2 | Valid | 99.5 | 49.8 | 184 ms | 20.1 ms | 2.0 | 0 | 2,984 | 30.0 s |
| Sustained decode | 64K | 4 | Valid | 197.2 | 49.4 | 309 ms | 20.2 ms | 4.0 | 0 | 5,912 | 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 | 123K | 1 | Valid | 38.6 | 38.6 | 201 ms | 25.9 ms | 1.0 | 0 | 1,156 | 30.0 s |
| Sustained decode | 123K | 2 | Valid | 75.6 | 37.9 | 371 ms | 26.4 ms | 2.0 | 0 | 2,268 | 30.0 s |
| Sustained decode | 123K | 4 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 123K | 8 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 123K | 16 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 123K | 32 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 123K | 64 | Over capacity | — | — | — | — | — | — | — | — |
| Sustained decode | 123K | 128 | Over capacity | — | — | — | — | — | — | — | — |
Over capacity: not run because this many requests at this context would exceed the server's KV cache.