Amazon bets on ARM Graviton systems to enhance and lower the price of its cloud services

Improving and reducing the costs of cloud services is a priority for Amazon and for this reason they are committed to creating systems based on ARM Graviton

It's funny how ARM is creeping into server systems and Data Centers without saying much. The company has ARM Graviton processors, a very interesting solution for large server systems. Amazon seems to know and is betting heavily on implementing this on a large scale. However this processor seems to be an updated version of an AMD Opteron, specifically the Opteron A1100. Although it seems impossible Amazon says that these are more efficient than any x86 from AMD or Intel.

ARM Graviton will save costs on Amazon's cloud services

Curiously, the e-commerce giant assures that Graviton will reduce the costs of the cloud service by 45%. It should be noted that Graviton is based on 64-bit Neoverse cores in 16nm lithography. It is based on the Cosmos architecture. Drew Henry, Senior Vice President of ARM has said that Graviton is an Israeli-designed chip with a 72-bit Cortex-A64 core working at a frequency of up to 2.3GHz.

INTERNATIONAL DRAW FOR AN AORUS GTX 1080 Ti GRAPHIC CARD

Logically Amazon will not withdraw all its systems based on solutions from Intel and AMD. As they have said the ARM solution will help in scaling applications. This will allow users to share the load through smaller sets of instances. It will allow you to develop microservices in containers, web servers, development environments or caching fleets.

In addition, Amazon will be able to license the ARM plans through the Annapurna system. In addition, this will allow them to customize and / or adjust the designs of the chips. They could even work with GlobalFoundries or TSMC to make these new chips.

It is not the only front they are working on as they are developing ASICs for AI Inference, called Inferentia. It is intended to scale to hundreds of trillion operations per second with greatly reduced cloud operating costs.

Source: Fudzilla