AWS is now renting the 64-bit Army Army service editors at the Museum • The Register


re: Invent Amazon is designing its own 64-bit Army Editors, named Graviton, and is currently selling on AWS.

Just in time for the year each year: A conference conference in Las Vegas, today's family revealed its A1 family of EC2 settings, and comes from a1.medium with 1 vCPU, 2GB of RAM, and up to 3.5Gbps EBS and a bandwidth network of 10Gbps, which costs $ 0.0255 per hour on request, and & # 39; Up to 1.4xlarge with 16 vCPUs and 32GB of RAM, and spend $ 0.408 per hour.

This can work out cheaper than a comparative T session in terms of CPU; However, there are different definitions between examples. You can compare and compare yourself here. Amazon claims that A1 issues up to 45 percent are cheaper than their x86 virtual devices, depending on the adjustment.

In any case, you should not look directly at the raw price: you will test your software stack on the A1 VMs for performance and conformance before making the architecture, due to the differences between the Army and x86.

The Graviton AArch64 CPUs, which run Amazon Linux 2, Red Hat Enterprise Linux 7.6, are not Ubuntu, and are available in Eastern Europe (North Virginia and Ohio), the US West (Oregon), and Europe (Ireland) in different formats from special guest requests. Imagine a scale platform, web hosting, and so on.

If you run written material in mobile scripts or interpreting languages, you may simply have to transfer this code to A1 machines and give it. If not, then scratch out an Armv8-A device, get pre-installed packages from the preferred Linux distro, or try previous binaries, depending on their accessibility.

"If you do not look closely, you may not first recognize that you were using an Army server. That's a design," Jon Masters " to find the Hat Hat Red Hat who was working on receiving RHEL supported by A1 settings. "I've been playing with Amazon A1 settings for a while and I've been swept away with the quality of engineering and team abilities. "

What's in …?

If you've been looking at what this came true, well, back in 2014, we warned you that Amazon was a. move by using weapons teams for cloud computing facilities. There was a big step in that question to & # 39; Raising Annapurna Labs for just one hundred million bucks in 2015. It is a Biz and Army designer and tiny system designer, and initially established Internet of Things projects. Then, AWS revealed that it used normal chips designed by Annapurna with the nest called Nitro. These networking and storage actions are reduced by EC2 financing, and & # 39; allowing x86 accommodation devices to focus on & # 39; running users' workloads as long as the packages and data were in place; move from Nitro ASIC.

Now, the tidy design team has used this knowledge to keep and edit Graviton Army editors in the AWS scout to place a consumer code on general funding tools. As you would expect from the Amazon very secret, there is not much more in the way in which technical details are currently available. We have organized PR AWS people for more information.

In the meantime, look at it / proc / cpuinfo on the A1 course, the ID number is 0xD08 CPU, which suggests that the Graviton is based on the Cortex-A72 blueprints at Arm 2015. Also, we note that there are one vCPU maps to one body corporate. The 16 VCPU settings are order in four quad-basic clusters with 2MB of L2 shared split cylinders, and 32KB of L1 data data, and 48KB of L1 management archive, each croft.

Last month, Army said that hosting a million Army Army data center had a power in 2018, and most of our collaborative systems would like us; Amazon was predicted to form a large proportion of that number. And Amazon also offers Epyc AMD's processes in its EC2 scam. At the same time, Microsoft Azure wants to have a minimum of 50-per-cent Arm powered.

Everything has more emphasis on Intel. The x86 Chipzilla CPUs comprise more than 90 per cent of the world data center; Canned a market, according to analysis. Epyc, Graviton, Cavium, and ThunderX2, and some other competitors, are likely to be deleted at that section at Intel. Pressure (Millions) ®

Source link