The NVIDIA A800 is the China based market version of the NVIDIA A100 GPU. It was created to comply with U.S market and export. It uses Ampere architecture and HBM2 memory generation. It supports NVLink for multi GPU utilization, allowing a large memory pool by having a clsuter of multiple A800 GPUs. This GPU is best for AI & Machine Learning workloads, and High performance computing. It does not provide display port features, hence in order to display it needs a companion GPU such as the NVIDIA RTX400 Ada or other GPUs with disply ports. Compared to memory GDDR6/GDDR7 architecture, HBM2 provides ultra-high bandwidth and wide buses, optimizing parallel throughput technology. Key Features: NVIDIA A800 40GB Brand NVIDIA Model # A800 GPU Memory 40GB Memory Generation HBM2 CUDA Cores 6,912 Memory Bandwidth 155.2GB/s Interface PCIe 4.0 x16 Tensor Cores 432 Architecture NVIDIA Ampere Power 240W Active Dimensions Dual Slot FH Display Capabilities Requires companion GPU RTX 4000 ADA, RTX A4000 MPN 900-51001-2200-000