New commodity supercomputing clusters deployed at NNSA labs declare TOP500 spots – HPC Information Evaluation

New commodity supercomputing clusters deployed at NNSA labs declare TOP500 spots – HPC Information Evaluation

Three of the Nationwide Nuclear Safety Administration’s (NNSA’s) state-of-the-art commodity computing clusters just lately deployed at Lawrence Livermore Nationwide Laboratory (LLNL) and Sandia Nationwide Laboratories (SNL) are among the many strongest supercomputing methods on the earth, Top500 organizers introduced Monday.

Funded below the Commodity Expertise Techniques II (CTS-2) contract by NNSA’s Superior Simulation and Computing (ASC) Program, machines at LLNL—named Dane and Bengal—started performing key modeling and simulation capabilities for NNSA’s stock administration mission in mid- -2023. Every system is constructed by Dell Applied sciences and is powered by 4y Scalable Gen Intel Xeon processors paired with high-speed Cornelis Networks Omni-Path networking.

The debut of the Dane cluster, a 7,041 petaflop (barely greater than 7 quadrillion calculations per second) cluster at quantity 108 and a 6,134 petaflop Bengal cluster at quantity 129, brings the overall of methods on LLNL’s spot within the Top500 listing to 11, essentially the most of any One other system. The world’s supercomputing heart. The third new CTS-2 system, featured on SNL and referred to as Stout, reached 8.987 petaflops on the LINPACK scale used to find out scores, touchdown it at No. 87 on the listing. Every system achieved an effectivity of 89 p.c or higher on LINPACK. Top500 organizers unveiled their semi-annual up to date listing of the world’s strongest computer systems on the 2023 Worldwide Convention on Excessive-Efficiency Computing, Networking, Storage and Evaluation (Supercomputing) in Denver.

Since their deployment in mission-critical work, the Dane, Bengal, and Stout system and the fourth CTS-2 system, named Amber, at SNL have demonstrated improved efficiency and effectivity over the earlier era of commodity methods (CTS-1), with NNSA researchers reporting preliminary outcomes of a 4-5X speedup throughout the vary Wide selection of excessive efficiency computing purposes.

“The deployment of those first CTS-2 platforms offers an vital addition to NNSA’s laboratory simulation environments in help of our nationwide safety mission,” stated Matt Leininger, CTS venture lead for NNSA’s tri-laboratories (LLNL, SNL, and Los Alamos). Nationwide Laboratory). “We’re proud to proceed the NNSA ASC’s custom of excellence in high-performance computing and America’s expertise partnerships at LLNL, SNL, and all through the NNSA advanced.”

Commodity technology-based methods function NNSA’s “workhorses” and help the every day simulation workload whereas decreasing prices by standardizing {hardware} and software program throughout NNSA laboratories, when mixed with the LLNL-led Tri-Laboratory Working System Stack (TOSS) and the Tri-Lab shared surroundings. CTS-2 platforms enable NNSA’s extra highly effective superior expertise system supercomputers, equivalent to the present Sierra system and the longer term El Capitan exascale system at LLNL, to concentrate on extra advanced issues essential to NNSA’s stock administration program.

Every CTS-2 system consists of constructing blocks referred to as “scalable items” (SUs) that account for about 1.5 petaflops of computing energy every. The modular design of SU permits NNSA laboratories to regulate system efficiency relying on mission wants or program budgets. Every SU is constructed with Dell PowerEdge C6620 and R760 servers that use 4y Scalable era Intel Xeon processors.

4th Technology Intel Xeon Scalable processors are outfitted with specifically designed accelerators that may improve HPC workload efficiency and energy effectivity by offloading duties to those accelerator options. Intel Superior Matrix Extensions (Intel AMX), one of many accelerator engines constructed into processors, can flip the massive mathematical matrix computations that type the core of deep studying workloads right into a single course of, making it important for delivering efficiency throughout workloads the place HPC and AI converge.

“These processors are perfect for constructing and deploying general-purpose AI workloads utilizing the most well-liked AI frameworks and libraries. These capabilities will allow engineers, researchers and scientists at Lawrence Livermore Nationwide Laboratory and Sandia Laboratory to “We’re proud to help laboratories with options that assist speed up the effort and time to investigate future and present information for his or her areas of focus.”

Every system can also be outfitted with Cornelis Networks’ high-speed 200 Gbps Omni-Path Specific community, offering CTS-2 with a extremely scalable cloth with open supply software program absolutely built-in with TOSS.

“With CTS-2, the Cornelis staff is happy to help the supply of commodity methods at NNSA as soon as once more, having beforehand delivered high-performance networks for the second era Tri-Lab Capability Clusters (TLCC2) and CTS,” stated Gunnar Okay. Gunnarsson, vp of resolution supply. and Help at Cornelis: “1.” “Working carefully with our ecosystem companions, Dell Applied sciences and Intel, Cornelis is happy to allow cluster-leading efficiency and effectivity at scale for Tri-Labs’ mission-critical workloads with Omni expertise -Path. We sit up for continued CTS-2 deployments, together with the introduction of the brand new 400 Gbps CN5000 Omni-Path product household as a part of a future structure modernization.”

A number of extra CTS-2 methods might be deployed at LLNL and SNL within the first half of 2024, supporting numerous NNSA applications that leverage the CTS and procurement structure developed by the NNSA ASC program.

You may also like...

Leave a Reply