Google Cloud announced the preview of new virtual machines based on ARM processors for Compute Engine, its cloud computing service that allows the creation and execution of tasks using the company’s high-performance servers. This family of machines will be called the Tau T2A.
Big tech uses solutions from the “Altra” line of processors from Ampere, a North American semiconductor company specializing in creating ARM platforms. Google will provide pre-built machines with up to 48 vCPUs (Virtual Processing Units) and 4 GB of memory for each vCPU, i.e. a total of up to 192 GB of RAM.
The virtual machines will have a network bandwidth of up to 32Gbps and multiple network storage options that promise to make Tau T2A VMs “ideal for scale-out workloads including web servers, containerized services, data logging, media codec, and Java applications”.
According to Ampere, T2A virtual machines can deliver up to 31% more performance than services based on current generation x86 processors, in addition to maintaining the low consumption that increasingly popularizes the ARM architecture.
Harvard University carried out tests with the new virtual machines and, citing the heavy workload in their research in the medical field, highlighted the advantages of T2A.
“We ported our workload to Google’s new T2A family and run with minimal effort,” says Christoph Gorgulla, a research associate at the university. “The price-performance ratio of T2A will help us study more compounds and therefore discover more promising drug candidates.”
Google took advantage of the launch of its new ARM-based virtual machines to unveil Batch, its new job scheduling system designed to intelligently and automatically process heavier applications, something the company says will improve workload execution. of AI and machine learning.
Tau T2A is now available in preview in several regions, including the US, Europe, and Asia. This Google Cloud service is expected to officially arrive to all eligible people in the coming months.