Looking back at the past couple of years of extraordinary joint engineering projects SGI and ANSYS have undertaken, it is clear to me that when a synergetic hardware and software partnership is established you, our joint customers, are the clear beneficiary. To that end, I would like to walk you through four such examples.
The first example was outlined over a year ago in my ANSYS guest blog, “ Solving the Impossible Electromagnetic Simulation with HPC ” where with a “grand challenge” benchmark we jointly demonstrated that the SGI® UV platform and ANSYS HFSS software could solve very large, high frequency electromagnetics problems like cosite analysis and radar cross section (RCS) analysis, as well as allow multiple frequency sweeps to be run without running out of computer system memory.
Then last November, ANSYS approached SGI with a request to work together on another benchmark challenge, to demonstrate ANSYS Maxwell’s patent-pending Time Decomposition Method (TDM), which allows the solving of all time steps of a low frequency electromagnetic problem simultaneously. This breakthrough of performance and new capabilities was published in February of this year in “ANSYS Maxwell and SGI® UV™ 3000 Maximizes Electromagnetic Computation Throughput.”
Scaling Better than Ever
When ANSYS 17.0 was released early this year, SGI was excited that ANSYS Mechanical could now scale up to 1,000 cores. Always eager to test their latest software developments, our applications engineer who works closely with our structural mechanics customers took the V17 sp-5 Model and scaled it to 1,008 cores! The full results of this and other standard ANSYS Mechanical benchmarks on SGI systems can be found in the newly published “SGI® Technology Guide for ANSYS® Mechanical”.
And just last month ANSYS and SGI application engineers worked together to achieve a new world record for scaling ANSYS Fluent on a SGI® ICE™ XA , which is one of the world’s fastest commercial distributed memory supercomputer platforms. The SGI manufacturing team had built a new system for our customer NCAR in the United States called “Cheyenne” and the system was going through its testing phases on our factory floor. The SGI applications team wanted to run a large benchmark of a commercial code on the full system and ANSYS Fluent fit the bill.
New World Record Benchmark
Breaking last year’s 129,024 core record by more than 16,000 cores, SGI was able to run the ANSYS provided 830 million cell gas combustor model from 1,296 to 145,152 CPU cores.This reduces the total solver wall clock time to run a single simulation from 20 minutes for 1,296 cores to a mere 13 seconds using 145,152 cores and achieving an overall scaling efficiency of 83%. [Tweet "SGI and ANSYS Achieve New Commercial #CFD World Record in #HPC - 145,152 cores"] The ANSYS Fluent benchmark performance increase was achieved with the help of SGI® MPI PerfBoost. SGI's MPI PerfBoost allows technical applications written for other MPI implementations to take advantage of the performance, robustness and scalability of SGI Message Passage Toolkit (MPI) at runtime without recompiling.
HPE to Acquire SGI
Finally, some of you may have read the August 11th announcement that Hewlett Packard Enterprise plans to acquire SGI . Antonio Neri, Executive Vice President and General Manager, Enterprise Group stated, “Not only will the acquisition of SGI strengthen HPE’s position in the high-growth big data analytic segment, it will also extend our presence in HPC verticals, such as government, life sciences, higher education and research, and manufacturing.”
HPE also has a strong partnership with ANSYS as evidenced by a recent blog post by Bill Mannel, former SGI Executive and current Vice President and General Manager of High-Performance Computing (HPC), Big Data Solutions, and Internet of Things (IoT) for HPE Servers. Once the deal closes (which is scheduled sometime between November 2016 and January 2017), a combined HPE and SGI — together with ANSYS — will be able to offer, across the globe, one of the technology industry’s most comprehensive solution for ANSYS workloads.