Supercomputing Infrastructure Solutions for Critical Scientific Problems

Scientists utilize supercomputers to address some of the most challenging compute- and data-intensive issues. Supercomputers include systems such as high-performance computing (HPC) clusters that leverage powerful CPUs and often GPUs, similar to our newly introduced A4 and A4X VMs. These resources provide the significant performance necessary for large-scale simulations, data analysis, and AI model training. Accessible supercomputing-class resources, HPC cluster software, and AI tools have become essential for scientific discovery and AI advancement.

Today, we are launching H4D VMs, our most powerful CPU-based virtual machines, designed to unlock new levels of performance for scientific applications. H4D VMs are equipped with the latest AMD CPUs and feature advanced Titanium network acceleration. Together, these innovations empower scientists to deploy supercomputing-class HPC clusters and scale applications across thousands of processors, allowing them to resolve complex problems more efficiently and accurately. H4D VMs with Titanium network acceleration are presently available in preview.

“This leap in computational capability will dramatically accelerate our pursuit of breakthrough therapeutics,” states Petros Koumoutsakos at Harvard University, “bringing us closer to effective precision therapies for blood vessel damage in heart disease.”

To assist in overcoming the challenges associated with designing, deploying, and managing complex clusters, we are introducing the Cluster Toolkit, which enables simple, reliable, and repeatable cluster deployments. Combined with the Cluster Director (formerly Hypercompute Cluster), users can deploy and manage a large cluster as a single unit, enhancing performance, efficiency, and resilience.

Scientific and AI applications frequently have extreme storage requirements, and for this reason, we are also announcing Google Cloud Managed Lustre, a new high-performance, fully managed parallel file system developed in collaboration with DDN and based on EXAScaler Lustre.