AMD’s Milan-X already runs on Azure and is now GA

AMD announced its new data center CPUs with 3D die stacking, formerly code-named Milan-X, are now generally available.

 By using the new CPUs, customers should see a reduced need for data center servers and power reduction leading to a 50% lowered total cost of ownership over three years, AMD claimed.

The company said its 3d Gen AMD EPYC CPUs with 3D V-Cache technology offer the industry’s largest L3 cache, as well as innovation in packaging and security.

Available today, Microsoft Azure announced its HBv3 virtual machines are upgraded to and available with the new AMD processors in some regions, including East U.S. Other regions will follow.

 Micron said it has already seen a 40% performance improvement with some EDA workloads compared to non-stacked 3d Gen EPYC in Micron’s data centers. Micron’s Raj Hazra, general manager of compute and networking, said Micron and AMD have collaborated on delivering the full capability of DDR5 memory to high performance data center apps.

For EDA, the 16-core AMD EPYC 7373x CPU delivers up to 66% faster simulations via Synopsys VCS when compared to the EPYC 73F3, AMD said.

A 32-core AMD EPYC 7573X processor can solve an average of 88% more CFD (Contract for Difference) problems per day that an unnamed competitor’s comparable 32-core count processor when running Ansys CFX, AMD claimed.

AMD claimed the new performance capabilities with the latest processors mean customers can deploy fewer servers and reduce power consumption in data centers to reduce the total cost of ownership. In one example, AMD said a typical data center running 4600 jobs per day on the Ansys CFX test case cfx-50 using 2P 32-core AMD EYP 7573X CPU based servers can reduce the estimated number of servers from 20 to 10 and lower power consumption by 49 percent, offering a 51% lower TCO over three years.

RELATED: AMD touts VCK5000 performance improvements, claims it TOPS Nvidia for AI workloads