NVIDIA Data Center GPU
The world's most advanced AI and HPC GPUs. Powering the AI revolution with unprecedented performance for training and inference.
The world's most advanced AI and HPC GPUs. Powering the AI revolution with unprecedented performance for training and inference.
产品图片需上传至: images/products/nvidia/
The NVIDIA H100 Tensor Core GPU is the flagship of the NVIDIA data center platform. Built on the Hopper architecture, it delivers unprecedented performance for AI training and inference, with up to 4x faster AI training than the previous generation.
| Architecture | NVIDIA Hopper | GPU Memory | 80GB HBM3 |
|---|---|
| Memory Bandwidth | 3.35 TB/s |
| Form Factor | SXM5 / PCIe Gen5 |
| TDP | 700W (SXM5) / 350W (PCIe) |
| Tensor Cores | Fourth-generation |
| Transformer Engine | Yes - 4x faster AI training |
The NVIDIA H200 is the next-generation AI superchip with 141GB of HBM3e memory. It delivers up to 1.8x faster LLM inference than H100 and is designed for the most demanding generative AI workloads.
| Architecture | NVIDIA Hopper |
|---|---|
| GPU Memory | 141GB HBM3e |
| Memory Bandwidth | 4.8 TB/s |
| Form Factor | SXM5 |
| TDP | 700W |
| vs H100 Inference | Up to 1.8x faster |
产品图片需上传至: images/products/nvidia/
产品图片需上传至: images/products/nvidia/
The NVIDIA A100 Tensor Core GPU delivers unprecedented acceleration for AI, data analytics, and HPC. The A100 enables flexible multi-instance GPU (MIG) partitioning for optimal resource utilization.
| Architecture | NVIDIA Ampere |
|---|---|
| GPU Memory | 40GB / 80GB HBM2e |
| Memory Bandwidth | 2.0 TB/s (80GB) |
| Form Factor | SXM4 / PCIe Gen4 |
| TDP | 400W (SXM4) / 300W (PCIe) |
| MIG Support | Up to 7 GPU instances |
Choose the right GPU for your AI and HPC workloads
Contact us for competitive pricing on NVIDIA data center GPUs.