RSC Tornado systems, an Iintel® Select Solution for simulation and modeling
When it comes to compute-intensive research areas, like highenergy particle physics and weather modeling and prediction, or computer-aided engineering (CAE), simulation and modeling workloads need to run on high performance computing (HPC) clusters. To support those applications at scale, modern HPC systems require multi-core processors, high-bandwidth fabrics, and broad input/output (I/O) capabilities.
Because of the complexity and variety of technologies available on the market, assembling an HPC system can be time-consuming. And, integration and configuration of selected components an impact the performance of the solution. The level of expertise required to properly configure the combined solution can intimidate small to large organizations. RSC Group and its family of RSC Tornado and RSC Tornado Phi supercomputing solutions deliver ultrahigh-dense, energy-efficient, high-performance processing for organizations that need supercomputing resources quickly.
RSC – A World Renowned Leader in Energy-Efficient, High-Performance Solutions
RSC Group (www.rscgroup.ru/en) is a leading developer and integrator of innovative HPC clusters. Its RSC Tornado cluster solutions offer ultrahigh-dense, scalable, and energy-efficient multi-rack systems with 100 percent “hot water” liquid cooling, based on its patented direct liquid cooling technology.
RSC direct liquid cooling technology, with up to 63-degree Celsius inlet liquid temperature, does not require energy-demanding refrigeration and enables efficient heat reuse, record-breaking energy efficiency, and computing density.
RSC Tornado offers some of the world’s most advanced capabilities:
- Ultrahigh-dense configurations—306 nodes per rack
- Compute performance—1.41 petaFLOPS per rack
- Power density—200+ kW per rack
- Energy efficiency—PUE of 1.06 or lower at a customer site
RSC Tornado is based on the Intel® Scalable System Framework (Intel® SSF) with Intel® architecture and Intel® technologies:
- Intel® Xeon® Scalable processor family
- Intel® Xeon Phi™ Processor 7200 series
- Intel® Omni-Path Architecture (Intel® OPA) next-generation HPC fabric
- Intel® Optane™ SSD with Intel® Memory Drive Technology
- Intel® SSD Data Center Series for PCIe* storage with NVMe* over fabric technology
- Options for Intel® FPGA acceleration
RSC has delivered systems that have appeared on Top500, Green500, and HPCG lists and provided installations to the country’s leading scientific institutions, including the following:
- Joint Institute for Nuclear Research (JINR)
- Saint Petersburg Polytechnic University (SPbPU)
- Russian Academy of Sciences (JSCC RAS)
- Siberian Supercomputer Center (SSCC SB RAS)
- South Ural State University (SUSU)
- Institute of Oceanology of Russian Academy of Sciences (IO RAS)
- Russian Weather Forecast Agency (Roshydromet)
- Moscow Institute of Physics and Technology (MIPT)
- Aviation, Oil & Gas, Energy, Computer Graphics, and other vertical industries
Patented Liquid Cooling Drives High Performance and Low PUE
Power consumption is at the core of challenges for institutions hosting and running today’s massive supercomputers. RSC Tornado is a unified, energy-efficient, and ultrahigh-dense archi- tecture, enabling creation of advanced computing solutions that optimize power efficiency and performance. With its patented 100 percent hot water cooling technology, RSC systems can run with up to 63-degree Celsius inlet coolant temperature for efficient heat re-use, providing PUE of 1.06 or lower at 0.7 petaFLOPS in a standard 42U-high cabinet housing up to 153 nodes.
High-Performance Enabling Ingredients
The RSC Tornado blade is a high-performance dual-processor server based on Intel Xeon Scalable processors with direct liquid cooling of all electronic components. Intel Xeon Scalable processors, such as the Intel® Xeon® Gold 6154 processor, feature significant enhancements that benefit HPC applications, including improve- ments in I/O, memory, fabric integration, and Intel® Advanced Vector Instructions 512 (Intel® AVX-512). Optionally, Intel® Xeon® Platinum processors—with up to 28 cores—can be used to meet the most challenging compute needs, depending on the requirements of the workload and other customer needs.
RSC Tornado Phi is a high-density, high-performance single- processor server with direct liquid cooling of electronic components based on innovative Intel® Xeon Phi™ 7290 processor intended specifically for high-performance computing, digital signal processing, and machine learning.
Intel OPA provides 100 gigabits per second (Gbps) bandwidth and a low-latency, next-generation fabric for HPC clusters. Its 48-port switch chip delivers a 33 percent increase in density over the traditional 36-port switch ASIC historically used for InfiniBand* networking. The denser radix chip reduces the number of required switches, cutting costs and enhancing reliability. Intel OPA also reduces cabling-related costs, power consumption, space require- ments, and ongoing system maintenance requirements. These advancements can lower fabric costs by up to 61 percent.2
Intel® Optane™ SSD
Intel Optane SSD is a system acceleration solution for the latest generations of Intel® Xeon® processors. Intel® Optane™ technology, based on Intel® 3D XPoint™ Memory Media and using the Intel® Rapid Storage Technology (Intel® RST) driver is located between the processor and slower spinning and solid state storage devices (HDD, SSHD, or SATA SSD). It can store commonly used data and programs closer to the processor, so the system can access informa- tion more quickly with improved overall system responsiveness.
Select Solution Configuration
Table 1 lists RSC Simulation and Modeling Solution typical configu- rations. Hardware and software specifications are determined by the overall workload requirements and operating system.
In addition to the Intel Xeon processor-based hardware foundation and Intel OPA, other technologies for Intel, and RSC provide further performance gains:
Intel AVX-512: Boosts performance for the most demanding computational workloads, with up to double the number of floating point operations per second (FLOPS) per clock cycle, compared to previous-generation Intel® processors.
Intel® Cluster Checker: Inspects more than 100 characteristics related to cluster health. Intel Cluster Checker examines the system at both the node and cluster level, making sure all compo- nents work together to deliver optimal performance. It assesses firmware, kernel, storage, and network settings and conducts high-level tests of node and network performance using the Intel® MPI Library benchmarks, STREAM*, the High-Performance LINPACK* (HPL*) benchmark, the High Performance Conjugate Gradients* (HPCG*) benchmark, and other benchmarks. Intel Cluster Checker can be extended with custom tests, and its functionality can be embedded into other software.
Intel® Cluster Runtimes: Supplies key software runtime elements that are required on each cluster to ensure optimal performance paths for applications. Intel runtime performance libraries, including Intel® Math Kernel Library (Intel® MKL) and Intel MPI Library, deliver excellent performance optimized for clusters based on Intel architecture.
RSC BasIS HPC and Cloud Solution Stacks
RSC BasIS is a data center software stack specifically targeted to HPC and cloud centers. It is based on optimized and validated builds of leading open-source components for data center management (such as xCAT*, OpenStack*, Ceph*, Puppet*, ELK Stack*, Zabbix*, and others) combined in a single distribution that is enhanced with installation, support, and additional maintenance tools.
RSC’s innovative supercomputing solutions deliver ultrahigh-dense, energy-efficient, high-performance systems for demanding MPI-based simulation and modeling applications. For more information, visit RSC Group at rscgroup.ru
For more information on Intel Select Solutions, visit intel.com/selectsolutions
1 Intel. “Performance Benchmarks and Configuration Details for Intel® Xeon® Scalable Processors.” intel.com/content/www/us/en/benchmarks/xeon-scalable-benchmark.html.
2 Configuration assumes a 750-node cluster, and the number of switch chips required is based on a full bisection bandwidth (FBB) fat-tree configuration. Intel® Omni-Path Architecture uses one fully populated 768-port director switch; the Mellanox EDR* solution uses a combination of 648-port director switches and 36-port edge switches. Intel and Mellanox component pricing is from kernelsoftware.com, with prices as of May 2016. Compute node pricing is based on the Dell PowerEdge R730* server from dell.com, with prices as of November 2015. Intel Omni-Path Architec¬ture pricing is based on estimated reseller pricing, which is based on projected Intel manufacturer’s suggested retail price (MSRP) pricing at time of launch.
Tests document performance of components on a particular test, in specific systems. Differences in hardware, software, or configuration will affect actual performance. Consult other sources of information to evaluate performance as you consider your purchase. For more complete information about performance and benchmark results, visit intel.com/benchmarks.
Cost reduction scenarios described are intended as examples of how a given Intel- based product, in the specified circumstances and configurations, may affect future costs and provide cost savings. Circumstances will vary. Intel does not guarantee any costs or cost reduction.
Intel technologies’ features and benefits depend on system configuration and may require enabled hardware, software or service activation. Performance varies depending on system configuration. No computer system can be absolutely secure. Check with your system manufacturer or retailer or learn more at intel.com.
For more complete information about performance and benchmark results, visit intel.com/benchmarks.
Intel disclaims all express and implied warranties, including without limitation, the implied warranties of merchantability, fitness for a particular purpose, and non-infringement, as well as any warranty arising from course of performance, course of dealing, or usage in trade.
Intel, the Intel logo, Intel Xeon Phi, Intel Optane, Intel 3D XPoint, and Intel Xeon are trademarks of Intel Corporation in the U.S. and/or other countries.
*Other names and brands may be claimed as the property of others.
RSC, RSC BasIS, RSC Tornado Phi and the RSC logo are trademarks of RSC Group in Russia, USA, Japan and the most of Europe countries.