Loading the catalog…
H100 NVL (NVIDIA)
Blurred values are subscriber-only. Sign in to start a free trial — 2 full pages a day — or subscribe for unlimited access.
| Memory Bandwidth Tb S | 3.9 |
|---|---|
| Process Node | TSMC 4N |
| Fp8 Tflops Dense | 1670.5 |
| Interconnect | NVIDIA NVLink: 600GB/s, PCIe Gen5: 128GB/s |
| Fp32 Tflops | 60 |
| Tdp W | 400 |
| Memory Type | HBM3 |
| Gpu Memory Gb | 188 |
| Accelerator Form Factor | PCIe |
| Architecture | Hopper |
| Fp16 Bf16 Tflops Dense | 835.5 |
| Fp16 Bf16 Tflops Sparse | 1671 |
| Int8 Tops Dense | 1670.5 |
| Fp64 Tflops | |
| Fp8 Tflops Sparse | |
| Int8 Tops Sparse | |
| Media Engines | |
| Product Name | |
| Tf32 Tflops Dense | |
| Tf32 Tflops Sparse |
|---|
| SKU | H100 NVL |
|---|
| Product | |
|---|---|
| Target LLM scale | |
| Memory (H100 NVL prose) | |
| Performance vs A100 (Llama 2 70B) | |
| NVIDIA AI Enterprise subscription | |
| BFLOAT16 Tensor Core* | |
| INT8 Tensor Core* | |
| Multi-Instance GPUs | |
| Server Options | |
| NVIDIA Enterprise | |
| Transistors |
