With multi-instance tasks, you can run an Azure Batch task on multiple compute nodes simultaneously to enable high performance computing scenarios like Message Passing Interface (MPI) applications. Azure Batch enables you to run parallel compute workloads on both Linux and Windows virtual machines. Today, Microsoft announced the release of MPI support for Linux on Azure Batch.
By creating a pool of A8 or A9 compute nodes, Batch MPI tasks can fully leverage the high-speed, low-latency RDMA network for those Azure VMs. To run multi-instance (MPI) tasks, your Batch pool needs to be communication-enabled (“enableInterNodeCommunication = true”) with “maxTasksPerNode set to 1”. Additionally, all nodes in the pool should have MPI installed (OpenMPI, IntelMPI or any other MPI installer). You can use a Start Task to create a pool that installs MPI on the nodes.
Read more about it here.