NVIDIAAI GPU · SXM5
H100 SXM
NVIDIA H100 SXM is a Hopper data-center accelerator with 80 GB HBM3 and 3.35 TB/s of memory bandwidth.
Specifications
Dense throughput, no sparsity| Architecture | Hopper |
|---|---|
| Form factor | SXM5 |
| Process | TSMC 4N |
| FP16 / BF16 tensor (dense) | 989.4 TFLOPS |
| FP8 tensor (dense) | 1,979 TFLOPS |
| FP4 tensor (dense) | — |
| Memory | 80 GB HBM3 |
| Memory bandwidth | 3.35 TB/s |
| Power | Up to 700 W |
| Launch | Sep 2022 |
AI index
Tensor × bandwidth100 × √((989.4 ÷ 989.4) × (3.35 ÷ 3.35)) = 100.0
The AI index is the geometric mean of dense 16-bit tensor throughput and memory bandwidth, each relative to the H100 SXM. It is a spec-sheet comparison, not a measured benchmark.