Test environment: single node · 36 CPU cores · 64 GB RAM · 2080 Ti (modified VRAM) · 22 GB GPU memory
This single-node evaluation adapts the workload design from the Ray Data multimodal AI benchmark. Daft's OOMs on the image and audio workloads may reflect the test machine's limited memory, while the older 2080 Ti GPU also limits throughput. Given current hardware and compute-budget constraints, we have not reproduced the original benchmark's full cluster-scale setup; these results apply only to the recorded single-node environment.
| Workload | Vane Data | Ray Data | Daft | vs Ray Data | vs Daft |
|---|---|---|---|---|---|
| Document | 86.09 sBatch size: 2560 | 127 sBatch size: 320 | 413 sBatch size: 20 | 32.2% lower | 79.2% lower |
| Image | 1147 sBatch size: 100 | 1767.11 sBatch size: 100 | OOM | 35.1% lower | OOM |
| Audio | 2312 sBatch size: 128 | 2363.08 sBatch size: 128 | OOM | 2.2% lower | OOM |
| Video | 7603 sBatch size: 32 | 6922 sBatch size: 32 | 8322 sBatch size: Not set | 9.8% higher | 8.6% lower |
Results from the same recorded single-node environment before per-engine batch-size tuning.
This single-node evaluation adapts the workload design from the Ray Data multimodal AI benchmark. Daft's OOMs on the image and audio workloads may reflect the test machine's limited memory, while the older 2080 Ti GPU also limits throughput. Given current hardware and compute-budget constraints, we have not reproduced the original benchmark's full cluster-scale setup; these results apply only to the recorded single-node environment.