Loading...
Loading...
141GB HBM3e | 4.8TB/s Bandwidth | Hopper | 700W TDP | PCIe 5.0
The NVIDIA H200 SXM is a next-generation AI accelerator featuring 141GB of HBM3e memory with 4.8TB/s bandwidth, delivering a significant uplift over the H100 for large language model inference. Built on the Hopper architecture, it is optimized for generative AI and high-performance computing.
Architecture
Hopper (GH200)
Memory
141 GB HBM3e
Memory Bandwidth
4.8 TB/s
Full hardware specifications for the NVIDIA H200 SXM
Architecture
Hopper (GH200)
Memory
141 GB HBM3e
Memory Bandwidth
4.8 TB/s
TDP
700W
Interface
PCIe 5.0 / SXM5
FP16 Performance
989 TFLOPS
80GB HBM3 | 3.35TB/s Bandwidth | Hopper | 700W TDP | NVLink 4.0
Hopper (GH100)
192GB HBM3e | 8TB/s Bandwidth | Blackwell | 1000W TDP | NVLink 5.0
Blackwell (GB200)
48GB GDDR6 | 864GB/s Bandwidth | Ada Lovelace | 350W TDP | PCIe 4.0 x16
Ada Lovelace (AD102)
80GB HBM2e | 2TB/s Bandwidth | Ampere | 400W TDP | NVLink 3.0
Ampere (GA100)