ROMED8-2T vs ROMED8-2T
| ROMED8-2T | ROMED8-2T | Difference | |
|---|---|---|---|
| Single Precision | 19101 | 19101 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Half Precision | 26235 | 26235 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Quantized | 12684 | 12684 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Geekbench AI 1.2.0 | Geekbench AI 1.2.0 | ||
System Information
| ROMED8-2T | ROMED8-2T | |
|---|---|---|
| AI Framework | ONNX | ONNX |
| AI Backend | DirectML | DirectML |
| AI Device | NVIDIA GeForce RTX 3090 | NVIDIA GeForce RTX 3090 |
| Operating System | Microsoft Windows 10 Pro (64-bit) | Microsoft Windows 10 Pro (64-bit) |
| Model | ROMED8-2T | ROMED8-2T |
| Processor |
AMD EPYC 7F52 @ 3.50 GHz
1 Processor, 16 Cores, 32 Threads |
AMD EPYC 7F52 @ 3.50 GHz
1 Processor, 16 Cores, 32 Threads |
| Processor ID | AuthenticAMD Family 23 Model 49 Stepping 0 | AuthenticAMD Family 23 Model 49 Stepping 0 |
| Processor Codename | ||
| Processor Package | ||
| Motherboard | ASRockRack ROMED8-2T | ASRockRack ROMED8-2T |
| Northbridge | ||
| Southbridge | ||
| BIOS | American Megatrends Inc. P3.80 | American Megatrends Inc. P3.80 |
| Memory | 256.00 GB | 256.00 GB |
AI Performance
| ROMED8-2T | ROMED8-2T | Difference | |
|---|---|---|---|
| Image Classification (SP) | 7825 | 7825 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Classification (HP) | 10221 | 10221 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Classification (Q) | 5617 | 5617 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Segmentation (SP) | 9788 | 9788 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Segmentation (HP) | 11404 | 11404 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Segmentation (Q) | 7615 | 7615 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Pose Estimation (SP) | 168720 | 168720 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Pose Estimation (HP) | 317250 | 317250 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Pose Estimation (Q) | 127163 | 127163 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Object Detection (SP) | 9462 | 9462 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Object Detection (HP) | 12128 | 12128 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Object Detection (Q) | 6711 | 6711 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Face Detection (SP) | 25654 | 25654 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Face Detection (HP) | 32478 | 32478 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Face Detection (Q) | 18153 | 18153 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Depth Estimation (SP) | 54677 | 54677 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Depth Estimation (HP) | 91057 | 91057 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Depth Estimation (Q) | 33932 | 33932 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Style Transfer (SP) | 246960 | 246960 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Style Transfer (HP) | 506695 | 506695 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Style Transfer (Q) | 179478 | 179478 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Super-Resolution (SP) | 30499 | 30499 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Super-Resolution (HP) | 39104 | 39104 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Image Super-Resolution (Q) | 18726 | 18726 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Text Classification (SP) | 2186 | 2186 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Text Classification (HP) | 2591 | 2591 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Text Classification (Q) | 1230 | 1230 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Machine Translation (SP) | 2290 | 2290 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Machine Translation (HP) | 2269 | 2269 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
| Machine Translation (Q) | 1160 | 1160 | 100.0% |
| ROMED8-2T | |||
| ROMED8-2T | |||
