AFDCU-Z100-HG32A
DCU(Deep Computing Unit Depth Calculator) is an accelerator card specially designed for AI artificial intelligence and deep learning. DCU can also be applied to intensive numerical calculation.
The new generation of general-purpose processors and co-processors will be widely used in big data processing, artificial intelligence, commercial computing and other fields, as well as key industries such as telecommunications, finance, Internet, education, transportation, etc., to promote the application and promotion of high-end processors in China. The technology and development reserve fund can make forward-looking pre-research and fund reserve for the company's future technology and product development, optimize the capital structure and improve its ability to resist risks.
- DESCRIPTION
- SPECIFICATION
- BIOS / DRIVERS UPDATE
- REQUIREMENT
►Universal accelerator card architecture
►Open software ecology and platform
►Comprehensive hashrate coverage
►Advanced artificial intelligence
GPU Engine | HY DCU 8100 |
Process | 7nm FinFET |
CUDA | 4096(64CUs) |
Peak FP64 |
Up to 1.5GHz |
Peak FP32 |
Up to 1.7GHz |
GPU Video Memory |
32GB HBM2 |
GPU Memory Interface |
4096 bit |
GPU Memory Clock |
2.0 GHz |
GPU Memory Bandwidth |
1024 GB/s |
Max thermal design power (TDP) | 350W |
Interconnect CPU to GPU | PCIe Gen4 × 16 |
Interconnect GPU to GPU | × GMI × 2 Up to 184 GB/s |