VOOZH about

URL: https://www.phoronix.com/review/aws-m8a-m8g-m8i-benchmarks/10

⇱ AMD EPYC Turin vs. Intel Xeon 6 Granite Rapids vs. Graviton4 Benchmarks With AWS M8 Instances Review - Phoronix


👁 Phoronix

AMD EPYC Turin vs. Intel Xeon 6 Granite Rapids vs. Graviton4 Benchmarks With AWS M8 Instances

Written by Michael Larabel in Cloud on 24 October 2025 at 11:00 AM EDT. Page 10 of 10. 5 Comments.

Across dozens of different workloads, the 5th Gen AMD EPYC M8a instance was most commonly delivering the best performance and value over M8g Graviton4 and M8i Granite Rapids. In many workloads it wasn't even a remotely close comparison.

Even for simple single-threaded Python benchmarks, AMD EPYC Turin with Zen 5 cores was delivering the fastest performance over Graniton4 and Granite Rapids.

From AI and HPC workloads to video encoding/transcoding, code compilation, and traditional LAMP tasks, the 5th Gen AMD EPYC processors powering the AWS M8a instances were delivering the best performance overall against AWS Graviton4 and Intel Xeon 6 Granite Rapids. Even though Graviton4 offers the lowest cost per hour of operation, AMD EPYC Turin was routinely delivering much better performance-per-dollar across most of the workloads tested.

The m8a.4xlarge instance was the front-runner over 90% of the time followed by Graviton4 and then M8i Granite Rapids picking up just five first place finishes.

When taking the geometric mean of all the benchmarks that ran successfully on all three M8 compute instance types, the AMD-powered m8a.4xlarge came out to delivering 1.6x the performance of the M8i.4xlarge Granite Rapids instance and 2.26x the performance of Graviton4. The AMD EPYC Turin instance does cost 1.15x that of Granite Rapids on an hourly on-demand basis but it's well worth it for the significantly better performance. Similarly the Graviton4 leads in the lowest hourly price at 73% the cost of the EPYC Turin instance, but for most workloads M8a was delivering both the best performance and performance-per-dollar in the AWS cloud.

Those wanting to go through all 145 benchmark results can do so via this result file, including performance-per-dollar metrics for each benchmark result in full.

If you enjoyed this article consider joining Phoronix Premium to view this site ad-free, multi-page articles on a single page, and other benefits. PayPal or Stripe tips are also graciously accepted. Thanks for your support.

Michael Larabel is the principal author of Phoronix.com and founded the site in 2004 with a focus on enriching the Linux hardware experience. Michael has written more than 20,000 articles covering the state of Linux hardware support, Linux performance, graphics drivers, and other topics. Michael is also the lead developer of the Phoronix Test Suite, Phoromatic, and OpenBenchmarking.org automated benchmarking software. He can be followed via Twitter, LinkedIn, or contacted via MichaelLarabel.com.