HPCG benchmark
{{Short description|Benchmark in high-performance computing}}
The High Performance Conjugate Gradients Benchmark (HPCG benchmark) is a supercomputing benchmark test proposed by Michael Heroux from Sandia National Laboratories, and Jack Dongarra and Piotr Luszczek from the University of Tennessee.{{cite news|url=http://www.hpcwire.com/2014/06/26/development-pushes-ahead-new-hpc-benchmark/|title=New HPC Benchmark Delivers Promising Results|last=Hemsoth|first=Nicole|date=June 26, 2014|publisher=HPCWire|access-date=2014-09-08|archive-date=2014-09-08|archive-url=https://web.archive.org/web/20140908203018/http://www.hpcwire.com/2014/06/26/development-pushes-ahead-new-hpc-benchmark/|url-status=live}}{{Cite web|url=http://www.sandia.gov/~maherou/docs/HPCG-Benchmark.pdf|title=Toward a New Metric for Ranking High Performance Computing Systems|last1=Dongarra|first1=Jack|last2=Heroux|first2=Michael|date=June 2013|website=|publisher=Sandia National Laboratory|access-date=2016-07-04 |archive-date=August 20, 2013 |archive-url=https://web.archive.org/web/20130820214451/https://www.sandia.gov/~maherou/docs/HPCG-Benchmark.pdf }}
Benchmark
It is intended to model the data access patterns of real-world applications such as sparse matrix calculations, thus testing the effect of limitations of the memory subsystem and internal interconnect of the supercomputer on its computing performance.{{Cite web|url=https://www.hpcwire.com/2015/07/15/linpacks-companion-metric-gains-traction/|title=LINPACK's 'Companion Metric' Gains Traction|last=Trader|first=Tiffany|date=2015-07-16|website=HPCwire|access-date=2016-07-04|archive-date=2016-05-30|archive-url=https://web.archive.org/web/20160530020957/http://www.hpcwire.com/2015/07/15/linpacks-companion-metric-gains-traction/|url-status=live}} Because it is internally I/O bound (the data for the benchmark resides in main memory as it is too large for processor caches), HPCG testing generally achieves only a tiny fraction of the peak FLOPS the computer could theoretically deliver.{{Cite web|url=https://www.epcc.ed.ac.uk/blog/2015/07/30/hpcg|title=HPCG: benchmarking supercomputers|last=Jackson|first=Adrian|date=30 July 2015|website=www.epcc.ed.ac.uk|publisher=EPCC at the University of Edinburgh|access-date=2016-07-04|archive-date=2016-08-20|archive-url=https://web.archive.org/web/20160820104711/https://www.epcc.ed.ac.uk/blog/2015/07/30/hpcg|url-status=live}}
HPCG is intended to complement benchmarks such as the LINPACK benchmarks that put relatively little stress on the internal interconnect.{{Cite web|url=http://insidehpc.com/2015/07/latest-hpcg-performance-list-complements-top500/|title=Latest HPCG Performance List Complements TOP500|last=Brueckner|first=Rich|date=2015-07-13|website=Inside HPC|language=en-US|access-date=2016-07-04|archive-date=2016-05-28|archive-url=https://web.archive.org/web/20160528172138/http://insidehpc.com/2015/07/latest-hpcg-performance-list-complements-top500/|url-status=live}} The source of the HPCG benchmark is available on GitHub.{{cite web |title=HPC-G source code |url=https://github.com/hpcg-benchmark/hpcg |website=Github |access-date=6 February 2019 |archive-date=11 June 2018 |archive-url=https://web.archive.org/web/20180611005543/https://github.com/hpcg-benchmark/hpcg |url-status=live }}
As of November 2024, the Fugaku supercomputer held the top spot in the HPCG performance rankings, followed by the Frontier, Aurora Grabs Number Three Spot.
In June of 2020, Summit was superseded by Fugaku with a speed of 16.0 HPCG-petaflops (an increase of 540%). Summit is currently 4th,{{Cite web |title=HPCG - November 2022 {{!}} TOP500 |url=https://www.top500.org/lists/hpcg/2022/11/ |access-date=2023-02-08 |website=www.top500.org |archive-date=2023-01-20 |archive-url=https://web.archive.org/web/20230120074623/https://www.top500.org/lists/hpcg/2022/11/ |url-status=live }} LUMI 3rd and Frontier 2nd.
See also
References
{{reflist}}
External links
- {{Official website|https://www.hpcg-benchmark.org/}}
- {{Github|hpcg-benchmark/hpcg|HPCG Benchmark}}
{{authority control}}