NVIDIA announces “Grace”, the first data center processing unit to target Intel’s competition


NVIDIA today announced a Grace processor aimed at supporting datacenters, which is based on ARM x86, and kicks off to rival Intel.

Today, NVIDIA joined Apple to compete with Intel in the processor market. After Apple introduced the M1 processor, NVIDIA announced in an event held today about the Grace processor that is based on ARM architecture, but it will not come to computers anytime soon.

NVIDIA’s first processing unit has been dubbed “Grace” after the pioneering computer scientist Grace Hopper, as it is NVIDIA’s first data center processing unit, which targets the high loads of AI-focused computers, as well as language processing.

The processing unit supports ARM artificial intelligence cores, the processing unit is also integrated with NVIDIA graphics card technology, and Nvidia indicates that the “Grace” processor system comes with 10 times faster performance than the traditional NLP language processor, as the processor supports up to a trillion operations compared to x86 devices DGXTM.

Also, the Grace system will be based on NVIDIA’s NVLink technology, reaching speeds of 900 GB / s between processing units and the graphics card, and NVIDIA confirms that it has 30 times faster performance than the current leading servers, also with LPDDR5x memory, Grace devices will enjoy 10 times better efficiency and energy.



Please enter your comment!
Please enter your name here