Select runtimes to compare side by side. Click chips below to toggle selection.
| Metric | ROCm | ExLlamaV2 | vLLM |
|---|---|---|---|
| Score | C56 | D47 | F35 |
| Type | |||
| Execution | aot | aot | jit |
| Interface | sdk | sdk | api |
| Cold Start | 200ms | 1000ms | 5000ms |
| Memory | 600MB | 300MB | 2000MB |
| Startup | 100ms | 200ms | 3000ms |
| Isolation | hardware | process | process |
| Maturity | stable | stable | production |
| Languages | C, C++, Python | Python, C++, CUDA | Python |
| License | MIT | MIT | Apache-2.0 |
| Links |