AWS Graviton

The first Annapurna Labs silicon product launched under the AWS umbrella was the AWS Nitro hardware and supporting hypervisor in November 2017.{{cite journal |last1=Liguori |first1=A |title=The Nitro Project–Next Generation AWS Infrastructure |journal=Hot Chips: A Symposium on High Performance Chips |date=2018 |url=https://old.hotchips.org/hc31/HC31_T1_AWS_Nitro_Hot_Chips_20190818-2.pdf |publisher=Institute of Electrical and Electronics Engineers (IEEE) |access-date=13 October 2023}} Following on from Nitro, Annapurna began to develop general-purpose CPUs using its expertise.

The benefits AWS anticipated included:

• Offering more choice in terms of selection of EC2 instances for customers
• Targeting Arm-based applications
• Providing high availability and security, while reducing virtualization costs
• Offering decent server performance with lower prices for customers

The first Graviton processor reached these goals. Graviton2 now offers better performance compared to X86-64:

35% faster running Redis,{{Cite web|url=https://community.arm.com/arm-community-blogs/b/infrastructure-solutions-blog/posts/redis-on-aws-graviton2|title=Gain up to 35% performance benefits for deploying Redis on AWS Graviton2|date=2021-07-20|website=arm|language=en-US}} 30% faster running Apache Cassandra,{{Cite web|url=https://community.arm.com/arm-community-blogs/b/tools-software-ides-blog/posts/increase-price_2d00_performance-by-deploying-cassandra-on-aws-graviton2|title=Increase performance by up to 30% by deploying Apache Cassandra on AWS Graviton2|date=2021-08-18|website=arm|language=en-US}} and up to 117% higher throughput for MongoDB.{{Cite web|url=https://community.arm.com/arm-community-blogs/b/operating-systems-blog/posts/mongodb-performance-on-aws-with-the-arm-graviton2|title=MongoDB performance on Arm Neoverse based AWS Graviton2 processors|date=2021-06-09|website=arm|language=en-US}} In addition to higher performance, Graviton offers 70% lower power consumption {{Cite web|url=https://www.nec.com/en/press/202209/global_20220929_03.html|title=NTT DOCOMO and NEC Reduce Power Consumption for 5G SA Core by an Average of 72% using AWS Graviton2, followed by a Successful Onboarding of 5G SA Core on Hybrid Cloud|date=2022-11-29|website=nec|language=en-US}} and 20% lower price.{{Cite web|url=https://aws.amazon.com/ec2/graviton/|title=20% lower cost and up to 40% higher performance for M6g, C6g, and R6g instances over M5, C5, and R5 instances respectively|date=2022-03-03|website=amazon|language=en-US}}

{{Infobox CPU

|name= Graviton

|numcores= 16x Cortex A72

|created= {{start date and age|November 26, 2018}}

|size-from= 16 nm

|clock= 2.3 GHz

|l1cache= 80 KB per core (48 instructions + 32 data)

|l2cache= 8 MB total

|arch= AArch64

|instructions= AArch64

|extensions= Neon, crc, crypto

|successor= Graviton2

|support status = Supported

}}

The first Graviton CPU has 16 Cortex A72 cores, with ARMv8-A ISA including Neon, crc, crypto. The vCPUs are physical cores in a single NUMA domain, running at 2.3 GHz. It also includes hardware acceleration for floating-point math, SIMD, plus AES, SHA-1, SHA-256, GCM, and CRC-32 algorithms.{{Cite web|url=https://www.theregister.com/2018/11/27/amazon_aws_graviton_specs|title=

Amazon's homegrown 2.3GHz 64-bit Graviton processor was very nearly an AMD Arm CPU|date=2018-11-27|website=theregister|language=en-US}}

Only the A1 EC2 instance contains the first version of Graviton.{{Cite web|url=https://aws.amazon.com/ec2/instance-types/a1/|title=Amazon EC2 A1 Instances|date=2018-11-26|website=Amazon Web Services|language=en-US|access-date=2022-10-11}}

The Graviton2 CPU has 64 Neoverse N1 cores, with ARMv8.2-A ISA including 2×128 bit Neon, LSE, fp16, rcpc, dotprod, crypto. The vCPUs are physical cores in a single NUMA domain, running at 2.5 GHz.{{Cite web|url=https://github.com/aws/aws-graviton-getting-started/blob/main/README.md|title=Building for Graviton2 and Graviton3|date=2022-09-22|website=Amazon Web Services|language=en-US|access-date=2022-10-10}}

EC2 instances with Graviton2 CPU: M6g, M6gd, C6g, C6gd, C6gn, R6g, R6gd, T4g, X2gd, G5g, Im4gn, Is4gen, I4g.{{Cite web|url=https://aws.amazon.com/ec2/graviton/|title=ARM Processor - AWS Graviton|website=Amazon Web Services|language=en-US|access-date=2024-05-01}} One or more of these instances are available in 28 AWS regions.

The Graviton3 CPU has 64 Neoverse V1 cores, with ARMv8.4-A ISA including 4×128 bit Neon, 2×256 bit SVE, {{abbr|LSE|Large System Extensions}}, rng, bf16, {{abbr|int8|Integer matrix multiply}}, crypto. Organized in a single NUMA domain, all vCPUs are physical cores running at 2.6 GHz. Graviton3 has 8 DDR5-4800 memory channels.

Compared to Graviton2, Graviton3 provides up to 25% better compute performance, up to 2× higher floating-point performance, up to 2× faster cryptographic workload performance, up to 3× better performance for machine learning workloads including support for bfloat16, and 50% more memory bandwidth. Graviton3-based instances use up to 60% less energy for the same performance than comparable EC2 instances.{{Cite web|url=https://www.aboutamazon.com/news/company-news/2021-letter-to-shareholders|title=Amazon 2021 Letter to Shareholders|date=2022-04-14|website=AboutAmazon|language=en-US|access-date=2022-11-16}}

Graviton3E is a higher power version of Graviton3.{{Cite web|url=https://aws.amazon.com/blogs/aws/new-amazon-ec2-instance-types-in-the-works-c7gn-r7iz-and-hpc7g/|title=New Amazon EC2 Instance Types In the Works|date=2022-11-28|website=AWS News Blog|language=en-US|access-date=2022-11-29}}

EC2 instances with Graviton3 CPU: C7g, M7g, R7g; with local disk: C7gd, M7gd, R7gd.

EC2 instances with Graviton3E CPU: C7gn, HPC7g.

The Graviton4 CPU has 96 Neoverse V2 cores, with ARMv9.0-A ISA{{Cite web|url=https://aws.amazon.com/blogs/aws/join-the-preview-for-new-memory-optimized-aws-graviton4-powered-amazon-ec2-instances-r8g/|title=Join the preview for new memory-optimized, AWS Graviton4-powered Amazon EC2 instances (R8g)|date=2023-11-28|language=en-US|access-date=2023-11-28}} plus the SVE2-crypto{{Cite web|url=https://github.com/aws/aws-graviton-getting-started/blob/main/README.md#building-for-graviton|title=AWS Graviton Technical Guide|date=2025-02-25|language=en-US|access-date=2025-02-25}} extension. It has 2 MB of L2 cache per core (192 MB total), and 12 DDR5-5600 memory channels. Graviton4 supports Arm's Branch Target Identification (BTI).

Amazon claims that Graviton4 is up to 40% faster for databases, 30% faster for web applications, and 45% faster for large Java applications than the Graviton3.

EC2 instances with Graviton4 CPU: R8g,{{Cite web|url=https://aws.amazon.com/about-aws/whats-new/2023/11/amazon-ec2-r8g-instances-aws-graviton4-processors-preview/|title=Announcing new Amazon EC2 R8g instances powered by AWS Graviton4 processors (Preview)|date=2023-11-28}} X8g,{{Cite web|url=https://aws.amazon.com/about-aws/whats-new/2024/09/amazon-ec2-x8g-instances/|title=Introducing Amazon EC2 X8g Instances|date=2024-10-14}} C8g,{{Cite web|url=https://aws.amazon.com/about-aws/whats-new/2024/09/amazon-ec2-c8g-m8g-instances/|title=Introducing Amazon EC2 C8g and M8g Instances|date=2024-10-14}} M8g,{{Cite web|url=https://aws.amazon.com/about-aws/whats-new/2024/09/amazon-ec2-c8g-m8g-instances/|title=Introducing Amazon EC2 C8g and M8g Instances|date=2024-10-14}} I8g.{{Cite web |title=Announcing Amazon EC2 I8g instances - AWS |url=https://aws.amazon.com/about-aws/whats-new/2024/12/amazon-ec2-i8g-instances/ |access-date=2025-01-05 |website=Amazon Web Services, Inc. |language=en-US}}

• Timeline of Amazon Web Services
• Fujitsu Fugaku – a supercomputer using the A64FX CPU implementing ARMv8.2-A
• Ampere Altra – a 64-bit ARM datacenter processor family implementing ARMv8.2-A deployed on Google Cloud Platform, Microsoft Azure, and Oracle Cloud

{{reflist}}

• {{Official website|https://aws.amazon.com/ec2/graviton/}}
• AWS: [https://github.com/aws/aws-graviton-getting-started/blob/main/README.md AWS Graviton Technical Guide]
• List of Independent Software Vendor (ISV) products publicly listing arm64/Graviton support: [https://github.com/aws/aws-graviton-getting-started/blob/main/isv.md]
• Dev.to: [https://dev.to/aws-builders/large-system-extensions-for-aws-graviton-processors-3eci Large System Extensions for AWS Graviton Processors]
• Arm: [https://community.arm.com/arm-community-blogs/b/infrastructure-solutions-blog/posts/redis-on-aws-graviton2 Gain up to 35% performance benefits for deploying Redis on AWS Graviton2]
• Arm: [https://community.arm.com/arm-community-blogs/b/infrastructure-solutions-blog/posts/apache-kafka-benchmarks-on-aws-graviton2 Gain up to 30% Cost-Performance benefits for Apache Kafka on AWS Graviton2 Processors]
• AWS: [https://aws.amazon.com/blogs/aws/aws-lambda-functions-powered-by-aws-graviton2-processor-run-your-functions-on-arm-and-get-up-to-34-better-price-performance/ AWS Lambda Functions Powered by AWS Graviton2 Processor – Run Your Functions on Arm and Get Up to 34% Better Price Performance]
• AWS: [https://aws.amazon.com/blogs/aws/announcing-aws-graviton2-support-for-aws-fargate-get-up-to-40-better-price-performance-for-your-serverless-containers/ Announcing AWS Graviton2 Support for AWS Fargate – Get up to 40% Better Price-Performance for Your Serverless Containers]

{{Application ARM-based chips|state=expanded}}

Category:ARM-based systems on chips

Category:Amazon (company) hardware

Category:Amazon Web Services