Top 10 Facts About Evaluating Graviton2 FAQ
FAQ
Foire aux questions
Yes, Intel Xeon is more performant and cost-effective than Graviton2 for many common workloads. Other claims about Intel performance in the AWS cloud are misleading. There are a few key technical differences affecting performance results. Newer generations of Intel processors in AWS instances beat those results and when per-core performance is compared, Intel leads for many common workloads.1
No, it’s a time-consuming, intense process. And it isn’t a one-time event—you will need to manage that code long-term after initial testing and validation you conduct during migration. Also, you’ll lose access to special instructions like Advanced Vector Extensions 512 and Deep Learning Boost that is only available on Intel instances. Application capabilities and performance gains that your IT organization or software vendor has made based on Intel-specific innovations aren’t available for the same application in Graviton2, or may not be supported by the ISV on ARM. Many enterprise applications aren’t certified for ARM, or if they’re used in Graviton2 instances, they ‘behave’ differently than on Intel architecture and familiar features aren’t enabled. Others are only certified on Intel architecture in the AWS cloud.
Yes. You can select the underlying processor hardware for all IaaS instances in EC2 and most PaaS instances in AWS.2
Selecting Intel architecture enables you to live migrate virtual machines (VM) between Intel-based servers without downtime. Live migration isn’t supported across architectures and requires the VM to go offline during migration. Add to that downtime and lost productivity the time and expense to your IT team to revalidate applications for different architectures and they’ll have to work through incompatibilities or suboptimizations that likely require time-consuming, expensive revalidation or troubleshooting.
Intel M6i instance has 3rd Gen Intel Xeon Scalable processors. They support up to 128 virtual central processing units (vCPUs) per instance, with a 4:1 ratio of memory to vCPU. The M6i instances support up to 40 gigabits per second (Gbps) bandwidth to Amazon Elastic Block Store in the largest size, and up to 10 Gbps in the four smallest sizes, which are more than double that of the previous-generation M5 instances. The M6i instances also support up to 50 Gbps networking, which is twice that of M5 instances.3
You can go to Contact Intel for support and more information.
Selecting Intel architecture enables you to live migrate virtual machines (VM) between Intel-based servers without downtime. Live migration isn’t supported across architectures and requires the VM to go offline during migration. Add to that downtime and lost productivity the time and expense to your IT team to revalidate applications for different architectures and they’ll have to work through incompatibilities or suboptimizations that likely require time-consuming, expensive revalidation or troubleshooting.
Intel M6i instance has 3rd Gen Intel Xeon Scalable processors. They support up to 128 virtual central processing units (vCPUs) per instance, with a 4:1 ratio of memory to vCPU. The M6i instances support up to 40 gigabits per second (Gbps) bandwidth to Amazon Elastic Block Store in the largest size, and up to 10 Gbps in the four smallest sizes, which are more than double that of the previous-generation M5 instances. The M6i instances also support up to 50 Gbps networking, which is twice that of M5 instances.3
You can go to Contact Intel for support and more information.