AWS unveils its own processors and they have nothing to do with x86 – Cloud – Hardware



[ad_1]

Amazon Web Services has revealed that it has designed and deployed its own processors in its cloud. They rely on the architecture Arm and not on Intel x86.

This is a huge vote of confidence in Arm Holdings, the UK company (formerly Acorn), which licenses CPU design instead of manufacturing silicon itself.

It also defies as never before the established thirty-six-year-old hegemony, creating more choice and competition for users.

Fully arm-equipped processors dominate mobile devices, a field in which Intel has had a bitter failure. Arm won by offering a combined speed and low energy consumption, both achieved without generating much heat.

These qualities are welcome everywhere and some small cloud operators have tried hand waiters, as well as some supercomputers. But servers equipped with arms have not met with sufficient success to raise such devices over promising sights.

But it's more than a quirk now, because the size and influence of AWS gives tangible, large-scale credibility to servers powered by Arm.

Best of all, AWS made the servers easily accessible for existing applications.

"If your application is written in a scripting language, chances are you can just move it to an A1 instance and run it as it is," wrote AWS, Jeff Barr.

Barr added that the new silicon is ideal for scalable workloads where you can share the load on a group of smaller instances. This includes containerized microservices, web servers, development environments, and caching fleets. "

Containerized microservices are state-of-the-art technology at the heart of cloud development efforts

The processors are called Graviton and feed a new type of "A1" EC2 instance that can support 1 to 16 virtual processors. As shown in the AWS EC2 price list, the smallest instance A1 – the a1.medium – costs US $ 0.0255 per hour, which is more than some of the very small instances of the company with the Intel technology.

Literate people in The register took a look at the A1 instances and found that they were based on the Cortex-A72 chip. It is an Arm product for mobile devices that can have between one and four cores per processor, but can be clustered to provide more processors on a single system-on-a-chip. El Reg also found that each virtual CPU corresponds to a physical kernel.

Intel stabbing at the front

The ability of the A1 to run scripted applications is critical as it will allow developers to not justify the effort required to begin using A1 instances, either seriously or out of curiosity.

This is bad news for Intel, who rather enjoyed being a supplier of AWS and other large-scale clouds.

Intel now knows that if its Xeon range remains superior to Arm's efforts, it will change over time.

It has also been shown that AWS had enough power to undertake the resolutely non-trivial task of designing its own silicon, building it and transforming it into servers. Or, in other words, he can now run most of Intel's core skills.

AWS also showed that it could collaborate with the software industry to create a complete ecosystem: shortly after unveiling A1 and Graviton, Red Hat's chief architect, Jon Masters, has stated to Medium that he had been fortunate to use the new A1 instances and stated that they are more or less indistinguishable from any other server running Linux.

So, AWS has also covered this aspect of things.

And this before considering the impact that it could have on the "edge", the new hot computer space will allow us to see the resources of computation and storage sprout to any place within the reach of a network 5G .

In general, armrests are expected to work very well on the edge. And AWS now has the opportunity, thanks to the flagship ESXi virtualization product from their recently transferred VMware partner, VM, which allows them to manage workloads where they are needed.

What is today a very important day is the day when the x86 hegemony has been challenged like never before.

[ad_2]
Source link