AWS ParallelCluster is a fully supported and maintained open source cluster management tool that makes it easy for scientists, researchers, and IT administrators to deploy and manage High Performance Computing (HPC) clusters in the AWS cloud. HPC clusters are collections of tightly coupled compute, storage, and networking resources that enable customers to run large scale scientific and engineering workloads. 

Significant feature enhancements to this latest release of AWS ParallelCluster include:

• Support for P4d instances: Customers can now select P4d instances for use in their clusters. These instances include support for NVIDIA GPUDirect Remote Direct Memory Access (RDMA) enabled through Elastic Fabric Adapter, which can accelerate tightly coupled applications using the NVIDIA Collective Communications Library (NCCL) for GPU-to-GPU communication. This option can be enabled using the new enable_efa_gdr configuration setting.

• Support for the CentOS 8 Operating System: Customers can now choose CentOS 8 as their base operating system of choice to run their clusters for both x86 and Arm architectures. As with other operating systems supported by AWS ParallelCluster, you can choose your operating system using the base_os configuration option, and can also choose to create and use your own custom AMI built on top of CentOS 8. CentOS 8 support also includes compatibility with all of AWS ParallelCluster’s supported schedulers and NICE DCV for remote visualization.

• Amazon CloudWatch Cluster Metrics Dashboard: Customers can track and visualize operational metrics for their clusters in CloudWatch. This includes metrics such as CPU and network utilization, file system read and write data operations, and read and write operations for Amazon Elastic Block Store volumes. Customers can use this dashboard to visualize cluster usage or identify performance bottlenecks to diagnose how best to improve cluster performance.

AWS ParallelCluster is available at no additional cost, and you pay only for the AWS resources needed to run your applications. Learn how to launch an HPC cluster using AWS ParallelCluster here. 

For more detail you can find the complete release notes for the latest version of AWS ParallelCluster here.