Now Elastic Fabric Adapter Supported By AWS Batch

AWS Batch allows you to execute batch computing workloads on the AWS Cloud. AWS Batch is a cloud-native scheduler which handles instance provisioning and job scheduling. AWS Batch automatically provisions instances as per the job identification, with the proper placement group, networking configurations, and with any user-specified file system. Now, AWS Batch provides the potential to uncover host devices to your AWS Batch jobs, containing the Elastic Fabric Adapter (EFA), that allows you to execute highly performant distributed HPC and machine-learning workloads by using AWS Batch’s managed instance provisioning and scheduling. Elastic Fabric Adapter EFA is a network interface for Amazon EC2 instances which allows users to execute applications needing high levels of inter-node communications at scale on AWS. High Performance Computing (HPC) applications using the Message Passing Interface (MPI) and Machine Learning (ML) applications using NVIDIA Collective Communications Library (NCCL) can scale to thousands of CPUs or GPUs with the Elastic Fabric Adapter. As a result, you get the application performance of on-premises HPC clusters with the on-demand elasticity and flexibility of the AWS cloud. AWS Batch automatically sets up the EFA interconnect to the instances it starts, which the customer describes via one API parameter. To read further about Elastic Fabric Adapter and uncovering host devices to AWS Batch, refer documentation.