Written by Rachel Pruitt, Product Marketing Manager, Azure Marketing, HPC+AI
The Azure High-Performance Computing (HPC) solution enables experts in silicon design by providing them the infrastructure and tools for chip design, IP design, silicon manufacturing, and silicon supply chain. Creating cutting-edge technologies requires extreme amounts of compute power, and as our customers face additional unique threats and challenges made worse by the pandemic, the Azure cloud can solve many of these by providing a secure, scalable platform that was built for HPC. Customers are increasingly using the Azure HPC platform to boost productivity, optimize resources, speed up time to market, and increase resourcing. AMD for example is using the Azure platform in combination with the HBv3 virtual machines to achieve better quality designs, speed up time to market, and increase capacity by bursting to the cloud.
We recently announced a preview for Azure HBv3 virtual machines enhanced by 3rd Gen AMD EPYCTM processors with 3D V-cache, codenamed “Milan-X.” These enhancements continue us on the path to becoming the best platform in the world for silicon design, by improving the performance, scaling efficiency, and cost-effectiveness of a variety of memory performance-bound workloads like RTL simulation workloads. Customers using Milan-X will experience up to 60 percent higher performance for RTL simulation, and the upgrade will be provided at zero cost beyond existing HBv3-series VMs, and with no changes required of customer workloads. In addition, all HBv3-series VMs globally will soon be upgraded with Milan-X processors. For more information, read the announcement and in-depth performance content.
In June, we announced the general availability of the FX-series virtual machines, powered by 2nd Generation Intel Xeon Scalable Processors. Designed for workloads demanding very high single-threaded and lightly threaded performance, these are ideal for Electronic Design Automation (EDA). The new FX-series virtual machines deliver up to a 19 percent better performance over FSv2-series virtual machines, with 1008 GB of RAM (42 GB per physical core), 2 TB of attached NVMe SSD, a single-core frequency of 4.1 GHz, a base frequency of 3.4 GHz, and an all-core-turbo frequency of 4.0 GHz. Additionally, customers needing additional per-CPU core performance can turn off hyper-threading. Read additional information about the FX-series virtual machines.
We will be at the Design Automation Conference (DAC) in person and can’t wait to see you there.
Read the full article
