Oracle Cloud is Now ‘ARM-ed’ with Ampere Altra® Servers
The world is transitioning from the complex instruction set computing (CISC) architecture, popularly known as the x86, towards the simplified and futuristic reduced instruction set computing (RISC) architecture, namely, the ARM-based instances. And, having its implementation globally across legions of devices — particularly about 160 billion chips deployed in nearly every mobile phone, several high-end laptops (Apple’s MacBook, Microsoft’s Surface, etc.) and servers — all vouch for the growing acknowledgment of ARM’s dominance in cleaner and efficient performance, and power consumption. And, consequently, of the promise of a technologically advanced, yet sustainable future which Oracle has time and again emphasized.
Taking these prospects into account, Oracle on Tuesday, 25th of May 2021, announced the new Oracle Cloud Infrastructure (OCI) Ampere A1 Compute platform, powered by the Ampere’s Altra® cloud-native processors. Simply put, enterprises can now subscribe to ARM-based compute instances — with the indistinguishable level of scalability, availability, security, and more in the x86 architecture variant — on the Oracle Cloud, for seamlessly developing and running ARM-based applications, all while reducing the costs on performance, and energy consumption. We will go into the specifications, but let us set the stage first, with a short introduction on Ampere Computing.
Ampere Computing, like any other ARM-based chipset manufacturer, has procured the design from Arm Ltd., and worked on to create a family of cloud-native servers named Altra®, in proximity with Oracle itself.
- Ready for deployment in an ultra-performance bare metal shape, via 160 single-threaded cores @ up to 3.0 Ghz, and 1 TB of Memory. Every core, by design, is single-threaded with its individual 64 KB L1 I-cache, and 64 KB L1 D-cache, along with a massive 1 MB L2 D-cache, precisely and consistently delivering predictable performance.
- Deployment-ready in a flexible VM shape, with Oracle’s exclusive and flexible architecture for seamlessly provisioning between 1-80 cores, and 1-64 GB of RAM per core.
- With a complete integration with GitHub, Jenkins, and GitLab, for fostering effortless development and deployment in the Ampere A1-powered compute/containers. Enterprises can also deploy their ARM-optimized applications on Dedicated Region Cloud@Customer, right behind their firewall.
- Oracle also has a Free Tier option for Ampere A1 Compute platform, with added free-of-charge resources accessible for selected ISVs, developers, and educational institutions, under the newly set up ‘Arm Accelerator’ package.
According to the popular SpecFP and SpecInt benchmarks, the OCI’s Ampere A1 platform aces the price-performance on cloud. The results can be checked out on Oracle’s notification page, or can otherwise be referred to, below (summarized, of course).
- The OCI Ampere A1, compared to other ARM-based server instances, is 2 and a half times higher in the price performance scale. Whereas, in comparison to Intel Skylake/AMD Naples, A1 turned out to be 4-times higher.
- The H.264 codec in video encoding used in assessing the performance on CPU-intensive workloads revealed that the OCI Ampere A1 has 22% higher price performance ratio against the similar AMD’s Milan instance, while hitting a whopping 98% higher price performance score against the similar Intel’s Ice Lake compute.
For pricing, the OCI Ampere A1 platform is indeed the first ‘penny core’ cloud-native server initiative, in the public cloud and dedicated region, offered at $0.01 for a core per hour, and $0.0015 for every GB of memory, per hour.