AWS Graviton and Arm Neoverse Enable the Best Price-performance in Amazon EC2
Based on Arm Neoverse V1, the AWS Graviton3 processor provides up 25% better compute performance, 2x faster cryptographic performance, and up to 3x better performance for ML workloads compared to AWS Graviton2.
Arm and AWS have partnered to enable the best price performance and CPU silicon innovation in the Amazon cloud. The AWS Graviton3 and AWS Graviton3E processors use the Arm Neoverse V1 core. And the AWS Graviton2 processor uses Arm Neoverse N1 cores to deliver its computational horsepower.
Tens of thousands of customers, including 48 of the top 50 Amazon EC2 customers, use AWS Graviton processors for their workloads. Snap, Twitter, Netflix, Epic Games, Splunk, Nielsen, Redbox, Hotelbeds, Honeycomb.io and others are experiencing the improved performance and cost savings AWS Graviton-based instances using Arm technology can deliver.
Looking to Get Started on Arm?
The AWS Graviton Fast Start program provides step-by-step directions for porting your workloads to Arm in as little as four days.
The AWS Graviton Advantage
Don MacAskill (SmugMug & Flickr), Liz Fong-Jones (Honeycomb.io) and Dave Brown (Amazon) share real world experiences about running workloads on AWS Graviton processors based on Arm Neoverse.
Introducing AWS Graviton3
With support for bfloat16, AWS Graviton3 processors deliver up to 3x better performance for ML workloads. Graviton3 is also the first CPU in the cloud to feature the latest DDR5 memory that provides 50% more memory bandwidth than DDR4.
*Compared to AWS Graviton2For scientific, machine learning, gaming and media encoding workloads. And up to 2x faster performance for cryptographic workloads than Graviton2.
Now available, the Hpc7g instance family delivers the best price-performance in Amazon EC2 for HPC workloads. Read the AWS Hpc7g blog.
EC2 Instances Powered by AWS Graviton
General Purpose (M7g and M6g/M6gd)
Built for general-purpose workloads such as application servers, mid-sized data stores, and microservices, and cluster computing.
Burstable General Purpose (T4g)
(Free Trial) Built for a broad range of burstable general-purpose workloads such as large-scale micro-services, small and medium databases, and virtual desktops.
Compute Optimized (C7g/C7gn and C6g/C6gd)
Built for compute-intensive applications such as high-performance computing, video encoding, gaming, and CPU-based machine learning inference acceleration.
Enhanced Networking (M7g and M6g/M6gd)
C7gn instances use AWS Nitro v5 with EFA for 200Gbps networking support.
Memory Optimized (R7g and R6g/R6gd)
Built for general-purpose workloads such as application servers, mid-sized data stores, and microservices, and cluster computing.
Enhanced Memory (X2gd)
Build for memory-intensive workloads (like R6g), but with the lowest cost per GiB of memory in Amazon EC2.
Storage Optimized (Im4gn and Is4gen)
Best price performance for storage-intensive workloads and lowest cost per TB of SSD storage in Amazon EC2.
GPU Accelerated (G5g)
Best price performance for Android game streaming and ML inference.
Use Cases for AWS Graviton-based Instances
Based on Arm Neoverse, these are some of the more common developer use cases for AWS Graviton.
Continuous Integration/Continuous Deployment |
|
---|---|
Users scaling up with GitLab and GitLab Runner on Graviton2 can achieve cost gains of 23% and performance gains of 36%. | |
Self-hosted runners add Arm support to CircleCI | |
Announcing general availability of Graviton2 CPU support for TravisCI |
More AWS Graviton-based Products and Services
Works on Arm
The Works on Arm initiative offers free-of-cost access to platforms and instances across the ecosystem and is a valuable resource for developers innovating on Arm.
Resources
Free Trial
Industry News
White Paper
- Benchmarking Apache Kafka for Cost-Performance on Amazon Web Service
- Guidelines for deploying NGINX Plus on AWS
Blogs
- Specialized processing is driving the era of Arm-based cloud computing, enabling AWS Graviton innovation
- Spark on AWS Graviton2: Machine Learning with MLlib
- Spark on AWS Graviton2: Real-Time Analysis using Spark Streaming
- Parsec add a new level of security for AWS IoT Greengrass
- Arm Blueprint:The cloud gets customized