Announcing new AMD EPYC™-based Azure Virtual Machines

Microsoft is committed to giving our customers industry-leading performance for all their workloads. After being the first global cloud provider to announce the deployment of AMD EPYC™ based Azure Virtual Machines in 2017, we’ve been working together to continue bringing the latest innovation to enterprises.

Today, we are announcing our second-generation HB-series Azure Virtual Machines, HBv2, which features the latest AMD EPYC 7002 processor. Customers will be able to increase HPC performance and scalability to run materially larger workloads on Azure. We’ll also be bringing the AMD 7002 processors and Radeon Instinct GPUs to our family of cloud-based virtual desktops. Finally, our new Dav3 and Eav3-series Azure Virtual Machines, in preview today, provide more customer choice to meet a broad range of requirements for general purpose workloads using the new AMD EPYC™ 7452 processor.

Our growing Azure HPC offerings

Customers are choosing our Azure HPC offerings (HB-series) incorporating first generation AMD EPYC Naples for their performance and scalability. We’ve seen a 33 percent memory bandwidth advantage with EPYC, and that’s a key factor for many of our customers’ HPC workloads. For example, fluid dynamics is one workload in which this advantage is valuable. Azure has an increasing number of customers for whom this is a core part of their R&D and even production activities. On ANSYS Fluent, a widely used fluid dynamics application, we have measured EPYC-powered HB instances delivering a 54x performance improvement by scaling across nearly 6,000 processor cores. And this is 24 percent faster than a leading bare-metal solution with an identical InfiniBand network. Additionally, earlier this year, Azure became the first cloud to scale a tightly coupled HPC application to 10,000 cores. This is 10x higher than what had been previously possible on any other cloud provider. Azure customers will be among the first to take advantage of this capability to tackle the toughest challenges and innovate with purpose.

New HPC, general purpose, and memory optimized Azure Virtual Machines

Azure is continuing to increase its HPC capabilities, thanks in part to our collaboration with AMD. In preliminary benchmarking, HBv2 VMs featuring 120 CPUs from the second generation EPYC processor are demonstrating performance gains of over 100 percent on HPC workloads like fluid dynamics and automotive crash test analysis. HBv2 scalability limits are also increasing with the cloud’s first deployment of 200 Gigabit InfiniBand, thanks to the second generation EPYC processor’s PCIe 4.0 capability. HBv2 virtual machines (VMs) will support up to 36,000 cores for MPI workloads in a single virtual machine scale set, and up to 80,000 cores for our largest customers.

We’ll also be bringing AMD EPYC 7002 processor to our family of cloud-based remote desktops, pairing with the Radeon MI25 GPU for customers running Windows-based environments. The new series offers unprecedented GPU resourcing flexibility, giving customers more choice than ever before to size virtual machines all the way from 1/8th of a single GPU up to a whole GPU.

Finally, we are also announcing new Azure Virtual Machines as part of the Dv3 and Ev3-series—optimized for general purpose and memory intensive workloads. These new VM sizes feature AMD’s EPYC™ 7452 processor. The new general purpose Da_v3 and Das_v3 Azure Virtual Machines provide up to 64 vCPUs, 256 GiBs of RAM, and 1,600 GiBs of SSD-based temporary storage. Additionally, the new memory optimized Ea_v3 and Eas_v3 Azure Virtual Machines provide up to 64 vCPUs, 432 GiBs of RAM, and 1,600 GiBs of SSD-based temporary storage. Both VM series support Premium SSD disk storage. The new VMs are currently in preview in the East US Azure region and with availability coming soon to other regions.

Da_v3 and Das_v3 virtual machines can be used for a broad range of general-purpose applications. Example use cases include most enterprise-grade applications, relational databases, in-memory caching, and analytics. Applications that demand faster CPUs, better local disk performance or higher memories can also benefit from these new VMs. Additionally, the Ea_v3 and Eas_v3 VM series are optimized for other large in-memory business critical workloads.

Taking advantage of these new offerings

Source: Azure Blog Feed

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.