A100 Tensor Core GPU
The NVIDIA A100 Tensor Core GPU provides excellent acceleration performance for AI, data analytics, and high-performance computing (HPC) applications at all scales, providing strong support for high-performance resilient data centers worldwide. As the engine of NVIDIA data center platform, compared with the previous generation NVIDIA Volta ™ Compared to it, A100 can improve performance by up to 20 times. A100 can be efficiently expanded and can also be divided into 7 independent GPU instances using multi instance GPU (MIG) technology, providing a unified platform to help elastic data centers dynamically adjust according to changing workload demands.
Details
NVIDIA A100 Tensor Core technology supports a wide range of mathematical precision and provides a single accelerator for each workload. The latest generation A100 80GB doubles GPU memory, providing 2TB/s of global ultra fast memory bandwidth, which can accelerate the processing of ultra large models and massive datasets. A100 is part of the complete NVIDIA data center solution, which consists of hardware, network, software, libraries, and NGC ™ The optimized AI models and applications in the directory are stacked together. As a powerful end-to-end AI and HPC platform suitable for data centers, A100 can help researchers obtain real results and deploy solutions on a large scale to production environments.
