NVIDIA H200 NVL 141GB HBM3e PCIe 5.0
- Bus Interface: PCI-Express 5.0 x16
- RAM Size: 80 GB
- Slot Width: Full Size ATX / Dual-Slot
- Power Consumption: 350 Watt
The NVIDIA H200 NVL 141GB is a high-end computing accelerator card designed for demanding AI, HPC, and data center workloads. Based on the NVIDIA Hopper architecture and equipped with 141 GB of HBM3e ECC memory, it offers enormous memory bandwidth and computing power for large language models (LLMs), generative AI, machine learning, scientific simulations, and data-intensive applications.
The exceptionally large HBM3e memory enables the processing of extensive AI models and datasets directly on the GPU. Multiple H200 NVL GPUs can be interconnected via NVLink for appropriately scalable workloads. The passive cooling is designed for use in dedicated server and workstation systems with sufficient system airflow.
The H200 NVL has no display outputs and is designed as a compute accelerator for GPU computing. It is particularly well-suited for professional systems where maximum AI computing power, large GPU memory, and high memory bandwidth are paramount.
Package Contents:
• NVIDIA H200 NVL 141GB PCIe
| GPU Architecture | Hopper |
|---|---|
| GPU Name | GH100 |
| Bus Interface | PCI-Express 5.0 x16 |
|---|---|
| Slot Width | Full Size ATX / Dual-Slot |
| Cooling | Passiv (Kühlkörper) |
| Power Consumption | 350 Watt |
| Single Precision Performance (FP32) | 24.08 TFLOPS |
|---|---|
| FP64 (Double Precision Performance) | 12.04 TFLOPS |
| Number of Shading Units | 7296 |
| Memory Interface Connection | 5120 bit |
| RAM Type | HBM2e |
| RAM Size | 80 GB |
| Bandwith | 1280 GB/s |
| ECC-Support | Ja |
| CUDA | 11.6 |
|---|---|
| DirectX | 12 |
| OpenGL | 4.6 |
| OpenCL | 3.0 |
| Vulkan | 1.3 |
| NVLink (Scalable Link Interface) | Ja |
