The fourth-generation AWS Graviton4 processor, now generally available, is a high-performance CPU based on Arm Neoverse-V2 cores and offers significant improvements over its predecessor.
It features 96 cores and delivers an average of 30% higher computing performance and 40% faster performance for databases and Java software.
If you’re wondering how it compares to competitors from Intel and AMD (and its predecessors), Phoronix has the answer. The site benchmarked five AWS instances in the “16xlarge” configuration, each with 64 vCPUs and 512 GB of memory. Graviton4 (r8g. 16xlarge) was compared to Graviton3 (r7g. 16xlarge), Graviton2 (r6g. 16xlarge), AMD EPYC 9R14 (r7a. 16xlarge), and Intel Xeon 8488C (r7i. 16xlarge). The instances were tested on Ubuntu 24.04 with the Linux Kernel 6.8 and the stock GCC 13.2 compiler.
Impressive performance
The Graviton4 performed impressively across the full range of tests, and in the geometric mean of all test results it came in second place, behind AMD’s EPYC processor. You can view all the test results here.
The results were consistent with Amazon’s own claims about its processor: Graviton4 had a slight edge of around 5% over the Intel Xeon processor, though AMD’s EPYC processor still had a lead of around 25% in overall performance.
Michael Larabel, Phoronix Founder, noted: “We’re at the stage where the Neoverse V2 cores with Graviton4 effectively match Intel Sapphire Rapids core-for-core and, depending on the workload, also match or surpass AMD 4th Gen EPYC. Of course, Graviton4 just went GA and will be going up against AMD EPYC Turin and Intel Xeon 6 Granite Rapids in the coming months.”
While Intel and AMD will continue to lead the way in terms of raw performance, Graviton4 isn’t far behind. It currently leads in ARM64 server performance, and given the generational leap AWS made with its latest processor, it will be interesting to see what improvements it brings with Graviton5.