NVIDIA A100 PCIe with a maximum of 8 units, 4U rack-mounted GPGPU model.
This is a super high-density GPGPU model that can accommodate up to 8 NVIDIA A100 PCIe in a 4U rack mount chassis. This product features two AMD 2nd generation EPYC processors, enabling high multi-threaded performance and high-speed network bandwidth with PCIe 4.0 connection, allowing for the construction of a high-density GPU compute node that significantly reduces TCO.
Inquire About This Product
basic information
The NVIDIA A100 Tensor Core GPU, developed as a successor to the Volta generation for data center GPUs aimed at AI and HPC applications, adopts the 7nm Ampere architecture with manufacturing process rules. It implements the third generation of Tensor cores and newly supports BFloat16 and Tensor Float 32, achieving 20 times the performance in FP32 calculations and INT8 deep learning inference processing compared to the previous generation Volta, enhancing both training and inference. Additionally, with the Multi-instance GPU (MIG) feature, a single A100 can be treated as up to seven GPUs, allowing multiple users to utilize independent instances.
Price range
P5
Delivery Time
Applications/Examples of results
AI, Deep Learning, HPC
Detailed information
-
NVIDIA A100 PCIe
catalog（1）
Download All Catalogs
Company information
Japan Computing Systems (JCS) has consistently cultivated cutting-edge technological capabilities as a supplier of core hardware (such as rack-mounted server systems and disk array systems) that combines high speed and stability to support the IT industry, as well as a solution provider in the IT field. In January 1996, it became a Microsoft Official Dealer (MOD) and a certified OEM manufacturer for Intel Japan. Our original brand, the Vintage series, is highly regarded for its reliability and the quality of support, backed by our experience and technological expertise, boasting an overwhelming track record of deliveries to major companies in the IT sector, as well as various government ministries and universities.