AWS and Arm’s Graviton Processor Can Help Customers Decarbonize Compute

Graviton3, the latest generation of the innovative Graviton processor – developed in a lighthouse partnership between Arm and AWS – is capable of providing customers with the ability to reduce their costs and associated carbon emissions.  The tech organizations’ collaborative approach has the potential to support future computing needs and accelerate decarbonization goals, at a time when business performance and cost management matters more than ever before.

AWS Graviton3 is built using a custom-designed Arm architecture using Arm Neoverse cores, optimized for delivering high performance and energy efficiency. AWS Graviton3 provides up to 25 percent better compute performance compared to its predecessor. The latest evolution of a custom-built, innovative processor aims to serve as a foundational first step for customers looking to reduce the impact of underlying compute and work to decarbonize.

As highlighted by the International Energy Agency (IEA), with increasing temperatures placing added strain on global energy use, the need to reduce carbon emissions across all aspects of society has never been more vital. While energy efficiency is always a functional requirement of technological innovation, the need for this to be front and center is now greater than ever.

Unlocking sizeable compute carbon savings at scale

Latest experiments conducted by AWS and Arm indicate that the incremental compute carbon savings made possible by the Graviton3 processor could be highly impactful at scale. Graviton processors, specifically designed to expedite cloud workloads, are widely available across most AWS regions since first being introduced into the marketplace in 2018. Graviton now supports over 50,000 customers of all sizes, from start-ups to SMEs to the world’s largest enterprises.

In order to qualify the business case and potential reduction in the carbon intensity of running a compute workload on Graviton3 compared to non-Graviton instances, Arm ran a silicon design verification computation in conjunction with AWS. The data from tests on the Graviton3 processor indicates that up to a predicted 67.6 percent lower workload carbon intensity (carbon emissions per simulation) can be achieved via the most energy efficient generation of processors powering the cloud.

The time, cost, and carbon emissions savings observed during this exercise indicate the possible gains when Graviton3 processors are put to use in Arm’s day-to-day HPC usage policy and design principles.  Arm’s ability to implement this across their operations could achieve considerable absolute carbon emissions reductions at scale.

This latest stage in the multi-year, purpose-driven collaboration between AWS, Arm and partners represents an opportunity to reduce the carbon impact of the most fundamental layer of technology powering our world today. The partnership’s mission is to drive forward energy- and carbon-efficient innovations in data center design and usage by utilizing Graviton processors. Migrating applicable workloads to Graviton is a practical and immediate step that organizations of all sizes can now take as a first action towards decarbonizing their IT functionality.

Decarbonizing compute through processor innovation

Arm’s longstanding expertise in highly-efficient, low-power compute lies at the heart of this latest processor, which looks to serve as the architecture of choice for further decarbonization at scale through the cloud to AWS’ millions of customers worldwide. Graviton3’s processing power already plays a fundamental role in aiding many organizations’ strategic drive to meet vital carbon emissions reduction goals, while also unlocking business performance gains and cost savings. Leveraging the energy-efficient capabilities of AWS Graviton processors offers the best performance per watt of energy used in Amazon Elastic Compute Cloud (Amazon EC2).

Cybersecurity platform Tehama successfully migrated to Graviton within just 12 months, including the initial proof of concept and the time needed by both development and DevOps teams to migrate the applications. Tehama has highlighted the positive impact that migrating its whole technology stack to Graviton has had.

“By adopting Graviton we have observed a significant improvement in price-performance and a notable reduction to the estimated carbon footprint of the Tehama room workload, without any compromise to the security, simplicity, availability and capabilities our global customers have come to expect,” says Carson White, Tehama VP of Operations.

Global data and applications platform provider Snowflake also credits Graviton with optimizing its virtual warehouses for sustainability. Since transitioning to AWS Graviton, Snowflake has reduced its carbon footprint per Snowflake virtual warehouse credit by an estimated 57 percent, while further optimizing its cloud environment for cost and performance.

As demand increases globally for compute and machine learning (ML), continual innovation at the chip level is critical to ensuring that future workloads can be powered sustainably. AWS and Arm are committed to maximizing Graviton’s energy efficiency, as well as the ongoing development of artificial intelligence (AI), ML, the Internet of Things (IoT), data analytics and computing.  

Building strong future foundations

The time has now come to decarbonize compute at all stages from the foundations upwards. The adoption of Graviton processors by AWS’ global and diverse customer base demonstrates the very real potential that exists to actively reduce the carbon intensity of a diverse array of foundational and essential IT workloads, helping to meet customer demand – even as compute requirements continue to grow and evolve in the future.

The rich capabilities and possible applications of AWS Graviton are vast and far-reaching – sectors including automotive, renewables, smart grids, aerospace, electronics and electrification could all benefit from the latest evolution of this processing technology.

Arm Editorial Staff also contributed to this blog.