Supermicro Shipping new Scale-Up In-Memory Computing Platforms with Intel Xeon Scalable processors

Today Supermicro announced that its new 4-socket servers featuring 2nd Gen Intel Xeon Scalable processors and Intel Optane DC persistent memory are now shipping in volume.

Our 4-socket servers are proven, reliable machines that have been powering mission-critical workloads for large enterprises and top cloud service providers around the world” said Raju Penumatcha, Senior Vice President and Chief Product Officer of Supermicro. “Using 2nd Generation Intel Xeon Scalable processors and Intel Optane DC persistent memory, these servers deliver tremendous performance boost for in-memory computing all at a dramatically lower cost compared to the previous generation architectures.”

Supermicro’s SuperServer 2049U-TR4 is powering the most demanding and intensive enterprise workloads for the top cloud vendors in the industry. This system is optimized to fully leverage the new 2nd Gen Intel Xeon Scalable processors as well as paradigm-changing Intel Optane DC persistent memory, and this 2U server provides up to 112 processor cores at a very compelling price point. In a recent study by Intel and SAP, Intel’s persistent memory enabled SAP S/4 HANA to restart up to 12.5 times faster than DDR4 and SSDs alone.

SUSE and Supermicro have long partnered on solutions that serve our customers in the digital economy,” said Phillip Cockrell, SUSE vice president of Worldwide Alliance Sales. “This partnership once again leads to SUSE support for Supermicro solutions, now including the newest in-memory 4-socket and 8-socket systems from Supermicro. SUSE, which provides the leading Linux platform for SAP workloads, and Supermicro continue to work together to deliver outstanding value to our customers by providing a foundation for SAP HANA on Kubernetes installations throughout the world.”

In addition to powering the cloud, the stability and scalability of Supermicro’s SYS-2049-TR4 makes it an ideal building block for enterprise customers who are looking to migrate from their current mainframe infrastructure. In a recent internal test conducted by Neeve Research, Supermicro Multi-Processor (MP) servers were able to demonstrate superior performance against typical mainframe workloads.

Using Supermicro MP servers, we were able to achieve real-time fraud detection against 10TB of data to process an astonishing 10 thousand transactions per second (TPS) with less than 10 microsecond response time.” said Grish Mutraja, CEO of Neeve Research, a pioneer in Next Generation in-memory computing technology. “The baseline today with current mainframes is around 800 TPS with a latency of 50 milliseconds. Supermicro’s servers provide amazing performance and strong value proposition for any enterprises that are seriously considering mainframe migration strategies.”

OmniTier, a global leader in advanced genome sequencing software, strongly agrees with this assessment. “Supermicro’s 4-socket server family has been proven to be a major asset to OmniTier’s CompStor advanced tiered-memory computing platform for genome sequencing,” said Dr. Jonathan Coker, CTO of OmniTier, Inc. “Our compute-heavy applications demand high core count per node, and our tiered-memory implementation demands access to state-of-the-art, very high performance local storage. Supermicro’s 4-socket server solution has exceeded both requirements.”

Supermicro’s Multi-Processor servers are designed for enterprise mission-critical workloads such as real-time OLTP + OLAP, in-memory analytics, virtualization and scale-up high performance computing. These servers are certified for a wide range of popular Operating Systems including RedHat, SUSE, Ubuntu, VMWare, and Windows.

Supermicro is sampling a new 4U server with four 2nd Gen Intel Xeon Scalable processors for up to 112 compute cores and up to 18 terabytes of memory. The SuperServer 8049U-E1CR4T supports 16 expansion slots, eight NVMe SSDs and up to six GPU cards. The best applications for this system include in-memory computing, real-time data analytics, AI inference and virtualization.

Sign up for our insideHPC Newsletter