dcsimg
 

AMD Details Data Center Server Silicon Innovation at Next Horizon

Wednesday Nov 7th 2018 by Sean Michael Kerner

AMD announces new 7nm Radeon Instinct GPU and provides details on its next generation EPYC 2 Rome data center server processors.

AMD announced a series of new products and services at its Next Horizon event on Nov. 6, as the silicon vendor continued its competitive march to take share from Intel in the data center.

At the Next Horizon event, AMD announced it has launched its 7nm-based Radeon Instinct MI60 and MI50 GPUs, revealed new details about the Zen2 CPU core and the next generation EPYC Rome processors, and reported that Amazon Web Services (AWS) was launching EPYC-powered cloud instances.

"We are well positioned to accelerate our momentum as we introduce the industry’s broadest, most powerful portfolio of datacenter CPUs and GPUs featuring industry-leading 7nm process technology over the coming quarters," Lisa Su, president and CEO, AMD, stated.

Datacenter GPUsRadeon Instinct GPU Accelerator

The new AMD Radeon Instinct™ MI60 and MI50 accelerators are 7 nanometer (nm) process-based datacenter GPUs and are positioned to be competitive against NVIDIA's datacenter GPU lineup.

The Radeon Instinct GPUs make use of the HBM2 (High-Bandwidth Memory) interface that enables up to 1 TB/s memory bandwidth speeds. The new GPUs also work with the next generation PCIe 4.0 interconnect, which according to AMD is six times faster than existing PCIe Gen3 interconnect speeds.

In support of the new GPUs, AMD also announced a new version of its open source ROCm software, which provides libraries and framework support for accelerated GPU-powered workloads on Linux.

"Combining world-class performance and a flexible architecture with a robust software platform and the industry's leading-edge ROCm open software ecosystem, the new AMD Radeon Instinct accelerators provide the critical components needed to solve the most difficult cloud computing challenges today and into the future," David Wang, senior vice president of engineering, Radeon Technologies Group at AMD, wrote in a statement.

EPYC RomeEPYC rome

AMD also provided new details on its next generation EPYC 2 processors that are codenamed "Rome."

The EPYC 2 processors use a 7nm process and are based on AMD's next generation Zen 2 cores. Each EPYC chip will have up to 64 Zen 2 cores, providing more compute power and memory bandwidth than the first generation of EPYC.

AMD claims the EPYC Rome chips will double the compute performance per socket vs. the first generation of EPYC chips.

"AMD is showing yet again its commitment to a very aggressive product improvement roadmap," Analyst Patrick Moorhead wrote in a research note. "With 'Rome,' AMD is changing everything. It is changing its SoC architecture to 7nm 'chiplets' with an improved Infinity Fabric, doubling cores per socket, doubling bandwidth per socket, adding PCIe 4.0 and improving core and FPU capabilities," Moorhead continued.

AMD first announced its EPYC Processor for data center servers back in June 2017. The EPYC platform has been scoring wins with server hardware vendors as well cloud providers. Oracle announced on Oct. 23 it would be offering EPYC instances on its cloud.

At the Next Horizon event, AWS announced that it too is now supporting EPYC with three new AMD EPYC processor-based cloud instances. The new instances include the general purpose M5 and T3 as well as the memory-optimized R5 instance types.

AWS is pricing the new EPYC instances to be 10 percent less expensive than the current Intel-based  M5, T3, and R5 instances.

Sean Michael Kerner is a senior editor at ServerWatch and InternetNews.com. Follow him on Twitter @TechJournalist.

Home
Mobile Site | Full Site