Oracle Cloud, the US-based Cloud Service Provider (CSP), has made major announcements regarding its High-Performance Computing (HPC) offerings. The new HPC offerings by Oracle indicate that the company is now ready to take massive strides in the sector, as the HPC market has been picking up lately. The new offerings from Oracle are targeted at offering resource-intensive tasks such as 3D visualizations and model renderings.
Oracle has announced that the start of 2021 will witness the company offering HPC instances to its users. These instances will mark Oracle moving away from Intel X7 HPC processors to the 10-nm Ice Lake processors. Oracle expects to deliver a 30 percent improved performance with its 2021 HPC instances.
Here’s what Karan Batta, product management VP at Oracle Cloud, said about the upgrade “We’ve spent a lot of time with Intel on tuning the parts and tuning the platform to our customers’ needs. We are expecting for similar cost to get about 30% performance upgrades. Those are specific for workloads like CFD and CAE traditional HPC workloads. It’s compatible with our old generations, you can still migrate data between two generations. This will also be part of our flexible instances, so you’ll be able to pick cores and memory, whichever way you want, and essentially make up your own.”
This focus on HPC marks Oracle honing its offerings to better compete with the more established CSPs in the market, especially the HPC sector. Batta said that Oracle is trying to create a dent in the HPC market “by solving problems that they [other CSPs] haven’t solved before.” He also added that the philosophy driving this shift is that customers “want the best of on prem — meaning performance and specialization — as well as the best of what supposedly the cloud should offer, which is pay-per-use, scalability, etc.”
Another major announcement by the company comes in the form of offering GPUs via the cloud. The Californian CSP will now offer workload instances based on NVIDIA A100 GPUs. Oracle promises to deliver a bump of 20 percent in performance as compared to the older Volta GPU. These instances will be available from 30 September in the US, Europe, Middle East, Africa, and APAC regions.
Other than the performance bump, Oracle is also targeting cost efficiency with the A100 GPUs. The instances will be available on an on-demand price of $3.05 per GPU hour, and Oracle claims this price is lower than what the competitors currently offer. The A100 GPU instances are targeted toward HPC as well as AI/ML workloads.
Oracle also made several other announcements, including the launch of E4 compute instances for general purpose workloads powered by AMD’s next-gen CPUs and the introduction of ARM-based compute instances in partnership with Ampere. Finally, Oracle also announced full integration of Rescale products and services with Oracle Cloud for easier onboarding of HPC workloads.