"Image may differ from actual product"
Roll over image to zoom in Click on image to zoom
This item is a Special Order and may take additional shipping time.
Place your order with SAITECH, and we will make every effort to have the item delivered to you as soon as possible.
How We Process Orders
Fastest Shipping: To get your products to you as quickly as possible, most orders placed directly by one of our several leading distributors or by the product's manufacturer. As a result, no one can get your orders to you faster than we can!
Strong Industry Alliances: SAITECH maintains long-standing alliances with the industry's leading distributors and manufacturers. As a result, we are the top source for your computer, office and Server technology needs.
Lowest Possible Prices: Advanced electronic interfaces continually scan our distributors for their best current pricing and availability. While changes in availability between different distributors may result in some price fluctuations, it ensures that you will always receive the lowest possible prices on currently available products!
Realtime Inventory: Live inventory updates are not available for this product.
Description
The NVIDIA H100 NVL Tensor Core GPU is engineered for businesses and data centers that need high efficiency in AI, deep learning & HPC workloads. Delivering up to 5X faster Llama 2 70B performance than NVIDIA A100 systems with low latency and power efficiency. Offering 1.5X more performance than H100 PCIe, it is the most optimized GPU in the H100 series, featuring high compute density, NVLink architecture, and enhanced memory bandwidth for large-scale AI inference and deep learning training.
- FP64 -30 teraFLOPS
- FP64 Tensor Core - 60 teraFLOPS
- FP32 - 60 teraFLOPS
- TF32 Tensor Core - 835 teraFLOPS
- BFLOAT16 Tensor Core - 1,671 teraFLOPS
- FP16 Tensor Core - 1,671 teraFLOPS
- FP8 Tensor Core - 3,341 teraFLOPS
- INT8 Tensor Core - 3,341 teraFLOPS
- GPU Memory - 94GB
- GPU Memory Bandwidth - 3.9TB/s
- Decoders 7 NVDEC / 7 JPEG
- Max Thermal Design Power (TDP)- 350-400W (configurable)
- Form Factor - PCIe dual-slot air-cooled
- Interconnect - NVIDIA NVLINK: 600GB/s
- PCIe Gen5 - 128GB/s
- Server Options Partner and NVIDIA
- NVIDIA Enterprise Included
Designed to operate in power-constrained data centers, the H100 NVL balances performance with energy efficiency, ensuring that high-density AI workloads remain cost-effective.
You may also like
Recently viewed
- Opens in a new window.