Edge Computing Cards
EdgeCom70-C75
EdgeCom70-C75
Front
Features and Benefits
Performance Features- Supports INT8 peak computing performance of 96 TOPS
Supports FP16/BF16 peak computing performance of 48 TOPS- Supports mixedprecision computing
Hardware decoding for 96channel HD video @25fps- Hardware encoding for 36channel HD video @25fps
- Decoding resolution for video and image up to 8K
- Compatible with various servers
Overview
The EdgeCom70-C72 Series AI Computing Card is designed for AI inference. It efficiently adapts to mainstream AI algorithms, enabling video structuring, facial recognition, behavior analysis, condition monitoring and other applications, empowering smart city, smart transportation, smart energy, smart finance, smart telecom, and smart industry.
Technical Parameters
| Main Processor | |
|---|---|
| CPU | 24-core A53, 169,280 DMIPS, 2.3GHz |
| AI Computing Power | |
| INT8 | 96 TOPS |
| FP16 / BF16 | 48 TFLOPS |
| FP32 | 6TFLOPS |
| Video/Image Codec | |
| Video Decoding | H.264 & H.265: 1080P @2400fps |
| Video Decoding Resolution | 8K / 4K / 1080P / 720P / D1 / CIF |
| Video Encoding | H.264 & H.265: 1080P @900fps |
| Video Encoding Resolution | 4K / 1080P / 720P / D1 / CIF |
| Image Codec | JPEG: 1200 images/s @1080P (interleave off) |
| Image Codec Max Resolution | 32768 * 32768 |
| Memory & Storage | |
| Memory | 48GB LPDDR4x 384bit 205GB/s |
| Software Features | |
| Deep Learning Frameworks | TensorFlow / PyTorch / Paddle / Caffe / ONNX / MXNet / DarkNet |
| Supported OS | Ubuntu / CentOS / Debian / UOS / Kylin |
| Supported Host CPUs | Compatible with Intel, AMD, Phytium, Hygon, Sunway, Loongson and other CPUs |
| Mechanical | |
| Dimensions | Half-height, half-length, single-slot |
| Electrical | |
| Wattage | 75W |
| Operating Environment | |
| Operating Temperature | 0℃ ~ 55℃ |
| Storage Temperature | 0℃ ~ 55℃ |
| Relative Humidity | 5% ~ 95% (non-condensing) |
| Standards and Certifications | |
| System Certification | ISO9001, ISO14001, ISO22163(IRIS) |
Resources
Ordering Information
| Model | Description |
|---|---|
| EdgeCom70-C72 | AI computing card, 72 TOPS, 4.5 TFLOPS AI computing power |