AWS recently announced the general availability (GA) of general-purpose M8a instances powered by the 5th Generation AMD EPYC (codename Turin) processors with a maximum frequency of 4.5GHz, suitable for applications that benefit from high performance and high throughput such as gaming, financial applications, ML inferencing, video encoding, application servers, simulation modeling, mid-size data stores, application development environments, and caching fleets.
The release of M8a is another addition to Amazon EC2’s instances portfolio, alongside the recent release of the memory-optimized R8i and R8i-flex instances and the compute-optimized C8i and C8i-flex EC2 Instances. These instances are a follow-up to the M7a instance from two years ago, with, according to AWS, 30% higher performance and up to 19% better price-to-performance.
Betty Zheng, a Senior Developer Advocate at AWS, writes:
For workloads with high I/O requirements, M8a instances provide up to 75 Gbps of networking bandwidth and 60 Gbps of Amazon Elastic Block Store (Amazon EBS) bandwidth, a 50% improvement over the previous generation. These enhancements support modern applications that rely on rapid data transfer and low-latency network communication.
M8a instances offer 10 virtualized sizes and two bare-metal options (metal-24xl and metal-48xl) to suit a range of deployment needs, from small applications to large workloads. Built on the AWS Nitro System, they provide low virtualization overhead, consistent performance, and enhanced security. With sixth-generation AWS Nitro Cards for I/O acceleration, M8a instances support up to 192 vCPU and 768 GiB RAM.
(Source: AWS News Blog)
Ryan Sagare commented in a LinkedIn post on the performance of the M8a instances:
3rd-party tests show AMD EPYC 9005 Series (“Turin”)-powered M8a instances dominated — winning 133 of 145 benchmarks vs. M8i and M8g. That’s 91%!
In addition, Niall Mullen, Senior Director, Cloud Infrastructure Engineering, Netflix, commented in an AMD blog post:
5th Gen EPYC delivered best in class performance and became a core part of Netflix Infrastructure enabling rapid scalability for Live events. AMD EPYC-powered EC2 M8a furthers that performance, delivering new features like AVX-512 to scale our encoding, personalization, and machine learning workloads across the globe.
However, not all industry watchers are entirely enthusiastic about the ever-expanding portfolio. The Snark bot from lastweekinaws commented on Bluesky:
AWS announces M8a instances: 30% faster than M7a! Because apparently M7g, M7i, M7a-large-with-extra-cheese weren’t confusing enough. Now with 45% more marketing percentages and 100% more pricing calculator headaches!
Currently, Amazon EC2 M8a instances are available in the US East (Ohio), US West (Oregon), and Europe (Spain) AWS Regions. Organizations can purchase M8a instances as On-Demand, Savings Plan, and Spot instances. Furthermore, M8a instances are also available in Dedicated Instances and Dedicated Hosts.
