{"id":781,"date":"2019-11-19T19:05:43","date_gmt":"2019-11-20T01:05:43","guid":{"rendered":"http:\/\/graph500.org\/?page_id=781"},"modified":"2019-11-19T19:05:43","modified_gmt":"2019-11-20T01:05:43","slug":"november-2019-bfs","status":"publish","type":"page","link":"https:\/\/graph500.org\/?page_id=781","title":{"rendered":"November 2019 BFS"},"content":{"rendered":"\n\n\n\n\t\n\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t\n\t
Rank<\/th>previous rank<\/th>Machine<\/th>Vendor<\/th>Type<\/th>Network<\/th>Installation Site<\/th>Location<\/th>Country<\/th>Year<\/th>Application<\/th>Usage<\/th>Number of nodes<\/th>Number of cores<\/th>Memory<\/th>Implementation<\/th>Scale<\/th>GTEPS<\/th>C_time<\/th>Power<\/th>\n<\/tr>\n<\/thead>\n
1<\/td>2<\/td>Sunway TaihuLight<\/td>NRCPC<\/td>Sunway MPP<\/td>Sunway<\/td>National Supercomputing Center in Wuxi<\/td>Wuxi<\/td>China<\/td>2015<\/td>research<\/td>research<\/td>40768<\/td>10599680<\/td>1304580 gigabytes<\/td>Custom<\/td>40<\/td>23755.7<\/td>961.455 seconds<\/td><\/td>\n<\/tr>\n
2<\/td>3<\/td>DOE\/NNSA\/LLNL Sequoia<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Lawrence Livermore National Laboratory<\/td>Livermore CA<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Government<\/td>98304<\/td>1572864<\/td>1572860 gigabytes<\/td>Custom<\/td>41<\/td>23751<\/td>368.7 seconds<\/td>6389760 Watts<\/td>\n<\/tr>\n
3<\/td>4<\/td>DOE\/SC\/Argonne National Laboratory Mira<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>49152<\/td>786432<\/td>786432 gigabytes<\/td>Custom<\/td>40<\/td>14982<\/td>259.8 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
4<\/td>new<\/td>OLCF Summit (CPU-Only)<\/td>IBM<\/td>IBM POWER9<\/td><\/td>Oak Ridge National Laboratory<\/td>Oak Ridge TN<\/td>United States<\/td>2018<\/td>Government<\/td>Scientific Research<\/td>2048<\/td>86016<\/td>1048576<\/td>Custom<\/td>40<\/td>7665.7<\/td>370.647<\/td><\/td>\n<\/tr>\n
5<\/td>5<\/td>SuperMUC-NG<\/td>Lenovo<\/td>ThinkSystem SD530 Xeon Platinum 8174 24C 3.1GHz Intel Omni-Path<\/td><\/td>Leibniz Rechenzentrum<\/td>Garching<\/td>Germany<\/td>2018<\/td>Academic<\/td>Research<\/td>4096<\/td>196608<\/td>393216<\/td>custom-aml-heavy-diropt<\/td>39<\/td>6279.47<\/td>1547.10<\/td><\/td>\n<\/tr>\n
6<\/td>7<\/td>Fermi<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>CINECA<\/td>Casalecchio Di Reno<\/td>Italy<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>8192<\/td>131072<\/td>131072 gigabytes<\/td>Custom<\/td>37<\/td>2567<\/td>400 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
7<\/td>8<\/td>NERSC Cori - 1024 haswell partition<\/td>Cray<\/td>XC40<\/td>Aries<\/td>NERSC\/LBNL<\/td>DOE\/SC\/LBNL\/NERSC<\/td>United States<\/td>2017<\/td>Government<\/td>basic science and simulation<\/td>1024<\/td>32768<\/td>133376<\/td>custom-aml-heavy-diropt<\/td>37<\/td>2562.16<\/td>1369.29<\/td><\/td>\n<\/tr>\n
8<\/td>9<\/td>Tianhe-2 (MilkyWay-2)<\/td>National University of Defense Technology<\/td>MPP<\/td>TH Express-2<\/td>Changsha China<\/td>Changsha China<\/td>China<\/td>2013<\/td>Science research Big data<\/td>Research<\/td>8192<\/td>196608<\/td>524288 gigabytes<\/td>custom<\/td>36<\/td>2061.48<\/td>456.071 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
9<\/td>10<\/td>Nurion<\/td>Cray<\/td>CS500 Intel Xeon Phi 7250 68C 1.4GHz Intel Omni-Path<\/td>Intel Omni-Path<\/td>Korea Institute of Science and Technology Information<\/td>Daejeon<\/td>Korea Republic Of<\/td>2018<\/td>Research<\/td>Various scientific and instudrial fields<\/td>1024<\/td>65536<\/td>98304<\/td>custom-aml-heavy-diropt<\/td>37<\/td>1456.46<\/td>3519<\/td><\/td>\n<\/tr>\n
10<\/td>11<\/td>Turing<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60GHz<\/td>custom<\/td>CNRS\/IDRIS-GENCI<\/td>Orsay<\/td>France<\/td>2012<\/td>Environment Fluid Dynamics Biomedical Astrophysics Physics Plasmas Computer Science Mathematics Chemistry<\/td>Academic<\/td>4096<\/td>65536<\/td>65536 gigabytes<\/td>custom<\/td>36<\/td>1427<\/td>1348 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
10<\/td>11<\/td>Blue Joule<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Science and Technology Facilities Council - Daresbury Laboratory<\/td><\/td>UK<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>4096<\/td>65536<\/td>65536 gigabytes<\/td>Custom<\/td>36<\/td>1427<\/td>1348 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
10<\/td>11<\/td>DIRAC<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>University of Edinburgh<\/td><\/td>UK<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>4096<\/td>65536<\/td>65536 gigabytes<\/td>Custom<\/td>36<\/td>1427<\/td>1348 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
10<\/td>11<\/td>Zumbrota<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>EDF R&D<\/td><\/td>France<\/td>2012<\/td>Energy Research<\/td>Industry<\/td>4096<\/td>65536<\/td>65536 gigabytes<\/td>Custom<\/td>36<\/td>1427<\/td>1348 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
10<\/td>11<\/td>Avoca<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Victorian Life Sciences Computation Initiative<\/td><\/td>Australia<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>4096<\/td>65536<\/td>65536 gigabytes<\/td>Custom<\/td>36<\/td>1427<\/td>1348 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
15<\/td>16<\/td>ALCF Cetus - 2048<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td><\/td>DOE\/ALCF<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>34<\/td>1384<\/td>421 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
16<\/td>17<\/td>TSUBAME 2.5<\/td>HP<\/td>Cluster Platform SL390s G7<\/td>QDR InfiniBand x 2 (80Gbps)<\/td>Global Scientific Information and Computing Center Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2013<\/td>Research<\/td>University<\/td>1024<\/td>12288<\/td>51200 gigabytes<\/td>Custom (CPU only)<\/td>36<\/td>1344.95<\/td>172.487 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
17<\/td>18<\/td>DARPA Trial Subset<\/td>IBM<\/td>Power 775 POWER7 8C 3.836GHz<\/td>Custom<\/td>Endicott<\/td>Endicott<\/td>USA<\/td>2012<\/td>Research<\/td>Research<\/td>1960<\/td>62720<\/td>250880 gigabytes<\/td>Custom 2D partitioned<\/td>36<\/td>1171.91<\/td>326 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
18<\/td>19<\/td>AI Bridging Cloud Infrastructure (ABCI)<\/td>Fujitsu<\/td>PRIMERGY CX2570 M4<\/td>Infiniband EDR<\/td>National Institute of Advanced Industrial Science and Technology (AIST)<\/td>Kashiwa Chiba<\/td>Japan<\/td>2018<\/td>Research<\/td>Artificial Intelligence Research<\/td>256<\/td>10240<\/td>98304<\/td>Custom<\/td>32<\/td>1063.6<\/td>34.236<\/td><\/td>\n<\/tr>\n
19<\/td>20<\/td>Oakleaf-FX<\/td>Fujitsu<\/td>Fujitsu - Fujitsu PRIMEHPC FX 10<\/td>Tofu<\/td>Information Technology Center The University of Tokyo<\/td>Tokyo<\/td>Japan<\/td>2011<\/td>Varioss scientific applications<\/td>University<\/td>4800<\/td>76800<\/td><\/td>Cutom<\/td>36<\/td>1002.72<\/td>1302.84 seconds<\/td><\/td>\n<\/tr>\n
20<\/td>21<\/td>NERSC Edison<\/td>Cray<\/td>XC30<\/td>Cray Aries<\/td>NERSC\/LBNL<\/td>DOE\/SC\/LBNL\/NERSC<\/td>USA<\/td>2013<\/td>basic science and simulation<\/td>Government<\/td>4817<\/td>115600<\/td>308288 gigabytes<\/td>Custom<\/td>36<\/td>865.3<\/td>1884.3 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
21<\/td>22<\/td>SAKURA<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>High Energy Accelerator Research Organization \/KEK<\/td><\/td>Japan<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>35<\/td>769<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
21<\/td>22<\/td>DOE\/NNSA\/LLNL Vulcan<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Lawrence Livermore National Laboratory<\/td>Livermore CA<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Government<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>35<\/td>769<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
21<\/td>22<\/td>IBM Rochester 1<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>IBM Rochester<\/td>Rochester MN<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Industry<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>35<\/td>769<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
21<\/td>22<\/td>IBM Rochester 2<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>IBM Rochester<\/td>Rochester MN<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Industry<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>35<\/td>769<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
21<\/td>22<\/td>HIMAWARI<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>High Energy Accelerator Research Organization \/KEK<\/td><\/td>Japan<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>35<\/td>769<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
21<\/td>22<\/td>BGQ<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Southern Ontario Smart Comuputing Innovation Consortium\/Univeristy of Toronto<\/td><\/td>Canada<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>2048<\/td>32768<\/td>32768 gigabytes<\/td>Custom<\/td>35<\/td>769<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
27<\/td>28<\/td>tatara<\/td>Fujitsu<\/td>PRIMERGY CX400<\/td>InfiniBand FDR\u00d71<\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2012<\/td>Scientific Research<\/td>University<\/td>1024<\/td>16384<\/td>131072 gigabytes<\/td>Custom<\/td>37<\/td>736.54<\/td>311.453 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
28<\/td>29<\/td>NERSC Cori - 512 KNL partition<\/td>Cray<\/td>XC40<\/td>Aries<\/td>NERSC\/LBNL<\/td>DOE\/SC\/LBNL\/NERSC<\/td>United States<\/td>2017<\/td>Government<\/td>basic science and simulation<\/td>512<\/td>32768<\/td>57344<\/td>custom-aml-heavy-diropt<\/td>35<\/td>678.933<\/td>1188<\/td><\/td>\n<\/tr>\n
29<\/td>30<\/td>Zeus<\/td>IBM<\/td>Power 775 POWER7 8C 3.836 GHz<\/td>Custom<\/td>DARPA Trial Subset IBM Development Engineering<\/td><\/td>USA<\/td>2012<\/td>Research<\/td>Research<\/td>1024<\/td>62720<\/td>131072 gigabytes<\/td>Custom-IBMRes-Austin<\/td>35<\/td>508.05<\/td>353.247 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
30<\/td>31<\/td>TSUBAME 2.0<\/td>HP<\/td>Cluster Platform SL390s G7 w\/ three Tesla cards per node<\/td>QDR InfiniBand x 2 (80Gbps)<\/td>GSIC Center Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2010<\/td>Research<\/td>University<\/td>1366<\/td>16392<\/td>71032 gigabytes<\/td>Custom (GPU)<\/td>35<\/td>462.25<\/td>107.08 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
31<\/td>32<\/td>BGQ<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Ecole Polytechnique Federale de Lausanne<\/td><\/td>Switzerland<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>754 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
31<\/td>32<\/td>BGQ<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Interdisciplinary Centre for Mathematical and Computational Modelling University of Warsaw<\/td><\/td>Poland<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>754 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
31<\/td>32<\/td>Vesta<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>DOE\/SC\/Argonne National Laboratory<\/td>Chicago IL<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Government<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>754 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
31<\/td>32<\/td>Grotius<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>IBM Thomas J. Watson Research Center<\/td>Yorktown Heights NY<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Industry<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>754 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
31<\/td>32<\/td>BGQ<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Rensselaer Polytechnic Institute<\/td>Troy NY<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
31<\/td>32<\/td>BGQ<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>University of Rochester<\/td><\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>784 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
37<\/td>38<\/td>Cluster Platform SL390s G7 w\/ three Tesla cards per node<\/td>NEC\/HP<\/td>Cluster Platform SL390s G7 w\/ three Tesla cards per node<\/td><\/td>GSIC Center Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2010<\/td>Research<\/td>University<\/td>1366<\/td>16392<\/td>73764 gigabytes<\/td>unknown<\/td>35<\/td>317.09<\/td>536.357 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
38<\/td>39<\/td>snq<\/td>IBM<\/td>BlueGene\/Q<\/td><\/td>Brookhaven National Laboratory<\/td>Upton NY 11973<\/td>USA<\/td>2012<\/td>physics geophysics materials biology chemistry comutational fluid dynamics and others<\/td>government industryuniversity<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>custom<\/td>34<\/td>294.293<\/td>405.953 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
39<\/td>40<\/td>Pleiades<\/td>HPE\/SGI<\/td>ICE-X dual plane hypercube FDR infiniband E5-2670 sandybridge<\/td>Infiniband dual rail FDR<\/td>NASA-Ames \/ Parallel Computing Lab Intel Labs<\/td>Mountain View CA<\/td>USA<\/td>2012<\/td>Geophysics automotive CFS Climate modeling Astrophysics<\/td>Government university industry and Government research<\/td>1024<\/td>16384<\/td>32768 gigabytes<\/td>custom<\/td>34<\/td>270.33<\/td>1550.12 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
40<\/td>41<\/td>Microway GPU Test Drive<\/td>Microway<\/td>14-core Xeon E5-2690v4 CPUs Two NVIDIA Tesla P100<\/td><\/td>University of Massachusetts Lowell<\/td>Microway<\/td>United States<\/td>2016<\/td>Graph Traversal on new GPUs<\/td>Graph Traversal<\/td>1<\/td>14<\/td>256 gigabytes<\/td>Several Developed<\/td>20<\/td>257.225<\/td>49.8944 seconds<\/td>Watts<\/td>\n<\/tr>\n
41<\/td>42<\/td>DOE\/SC Hopper<\/td>Cray<\/td>XE6<\/td><\/td>NERSC\/Lawrence Berkeley National Laboratory<\/td>DOE\/SC\/LBNL\/NERSC<\/td>USA<\/td>2010<\/td>Basic science and simulation<\/td>governmen<\/td>4817<\/td>115600<\/td>154144 gigabytes<\/td>Custom<\/td>35<\/td>254.074<\/td>4639.4 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
42<\/td>new<\/td>High Throughput Computer(+GPU)<\/td>Institute of Computing Technology Chinese Academy of Science<\/td>High Throughput Computer<\/td><\/td>SMARCO<\/td>Beijing<\/td>China<\/td>2019<\/td>High Throughput big data processing<\/td>Graph Computing<\/td>1<\/td>58<\/td>32<\/td>Custom GPU<\/td>26<\/td>237.94<\/td>155.55<\/td>130<\/td>\n<\/tr>\n
43<\/td>43<\/td>HA-PACS<\/td>Appro International \/ Cray Inc.<\/td>Xtream-X GreenBlade 8204 Xeon E5-2670 8C 2.600GHz Infiniband QDR NVIDIA 2090<\/td>Infiniband QDR<\/td>Center for Computational Sciences University of Tsukuba<\/td>Tsukuba<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>128<\/td>2048<\/td>16384 gigabytes<\/td>Custom (GPU)<\/td>32<\/td>223.634<\/td>191.805 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
44<\/td>44<\/td>SGI UV 300<\/td>HPE\/SGI<\/td>UV 300 (32-socket Intel(R) Xeon(R) CPU E7-8890 v3 16 TB RAM)<\/td>NUMAlink7<\/td>SGI<\/td>Chippewa Falls WI<\/td>USA<\/td>2015<\/td>Vendor<\/td>Vendor<\/td>1<\/td>576<\/td>16384 gigabytes<\/td>Custom (NUMA-aware OpenMP with ULIBC SGI & Kyushu University)<\/td>34<\/td>219.428<\/td>3422.83 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
45<\/td>45<\/td>IBM POWER8+ Tesla P100<\/td>IBM && Nvidia<\/td>IBM POWER8+ Tesla P100<\/td><\/td>IBM Moscow && MSU Alex Kolganov<\/td>Moscow<\/td>Russian Federation<\/td>2016<\/td>research<\/td>research<\/td>1<\/td>66<\/td>256 gigabytes<\/td>GPU Custom<\/td>26<\/td>204<\/td>150 sec seconds<\/td>175 watts Watts<\/td>\n<\/tr>\n
46<\/td>46<\/td>SHOUBU<\/td>PEZY Computing \/ Exascaler Inc.<\/td>ZettaScaler-1.4 80Brick<\/td>FDR Infiniband<\/td>Advanced Center for Computing and Communication RIKEN<\/td>Wako Saitama<\/td>Japan<\/td>2015<\/td>Research<\/td>Government<\/td>256<\/td>2048<\/td>32768 gigabytes<\/td>Custom<\/td>31<\/td>184.725<\/td>11.5564 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
47<\/td>47<\/td>ismuv2k<\/td>HPE\/SGI<\/td>SGI UV 2000<\/td>NUMAlink6<\/td>The Institute of Statistical Mathematics<\/td>Tokyo<\/td>Japan<\/td>2014<\/td>Data assimilation<\/td>National research institute<\/td>1<\/td>1280<\/td>32768 gigabytes<\/td>Custom (OpenMP + NUMA-aware)<\/td>33<\/td>174.704<\/td>20240 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
48<\/td>48<\/td>BGQdev<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Southern Ontario Smart Computing Innovation Consortium\/Univeristy of Toronto<\/td><\/td>Canada<\/td>2012<\/td>Scientific Research<\/td>Academic<\/td>512<\/td>8192<\/td>8192 gigabytes<\/td>Custom<\/td>33<\/td>172<\/td>717 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
48<\/td>48<\/td>BGQdev<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>IBM Thomas J. Watson Research Center<\/td>Yorktown Heights NY<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Industry<\/td>512<\/td>8192<\/td>8192 gigabytes<\/td>Custom<\/td>33<\/td>172<\/td>717 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
48<\/td>48<\/td>DOE\/NNSA rzuseq<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Lawrence Livermore National Laboratory<\/td>Livermore CA<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Government<\/td>512<\/td>8192<\/td>8192 gigabytes<\/td>Custom<\/td>33<\/td>172<\/td>717 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
51<\/td>51<\/td>Bem<\/td>SuperMicro\/WASKO<\/td>Cluster<\/td>infiniband FDR<\/td>Wroc\u0142aw University of Technology<\/td>Plac Grunwaldzki 11 50-377 Wroc\u0142aw<\/td>Poland<\/td>2015<\/td>general purpose<\/td>University<\/td>720<\/td>17280<\/td>47000 gigabytes<\/td>Tuned MPI implementation with 2-D data distribution<\/td>33<\/td>168.288<\/td>544 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
52<\/td>52<\/td>NEDO-AIRC-NAIROBI-UV<\/td>HPE\/SGI<\/td>UV300 (Intel Xeon E7-8867v3 x16)<\/td><\/td>National Institute of Advanced Industrial Science and Technology (AIST)<\/td>Tokyo<\/td>Japan<\/td>2016<\/td>Artificial Intelligence<\/td>Government<\/td>1<\/td>256<\/td>11775.7 gigabytes<\/td>Custom (graph500_hpgp2016_2016-05-27)<\/td>33<\/td>161.215<\/td>2759.46 seconds<\/td>4663.56 Watts<\/td>\n<\/tr>\n
53<\/td>53<\/td>ismuv2k2<\/td>HPE\/SGI<\/td>SGI UV 2000<\/td>NUMAlink6<\/td>The Institute of Statistical Mathematics<\/td>Tokyo<\/td>Japan<\/td>2014<\/td>Data assimilation<\/td>National research institute<\/td>1<\/td>640<\/td>16384 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>33<\/td>149.8<\/td>7654.7 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
54<\/td>54<\/td>Hawk028os1<\/td>Hewlett Packard Enterprise<\/td>HPE Integrity Superdome X Gen9<\/td>N\/A<\/td>HPE Ft. Collins<\/td>Ft. Collins CO<\/td>USA<\/td>2015<\/td>Multi-discipline<\/td>Vendor Benchmarking<\/td>1<\/td>288<\/td>12288 gigabytes<\/td>Custom (Kyushu University Japan)<\/td>33<\/td>141.9<\/td>5242 seconds<\/td>9508 Watts<\/td>\n<\/tr>\n
55<\/td>55<\/td>TitanX_forsite<\/td>Intel(R) Server<\/td>CPU Intel(R) Core i5 4th Gen (4 cores) && NVidia GTX Titan X<\/td>none<\/td>forsite-company.ru && NVidia LTD<\/td>Moscow State University && NVidia LTD<\/td>Russia<\/td>2015<\/td>Scientific Research<\/td>University<\/td>1<\/td>28<\/td>32 gigabytes<\/td>Custom GPU<\/td>26<\/td>132.14<\/td>305 seconds<\/td>162 Watts<\/td>\n<\/tr>\n
56<\/td>56<\/td>Xeon(R) Gold 5118+Tesla P100<\/td>Intel<\/td>Xeon(R) Gold 5118+Tesla P100<\/td><\/td>SMARCO<\/td>Beijing<\/td>China<\/td>2018<\/td>High Performance computing<\/td>High Performance computing<\/td>1<\/td>24<\/td>192<\/td>Custom GPU<\/td>26<\/td>131.53<\/td>155.5<\/td>220<\/td>\n<\/tr>\n
57<\/td>57<\/td>hawk040os1<\/td>Hewlett-Packard<\/td>HP Integrity Superdome X<\/td>N\/A<\/td>HP Ft. Collins<\/td>Ft Collins CO<\/td>USA<\/td>2014<\/td>Benchmarking<\/td>Vendor Benchmarking<\/td>1<\/td>240<\/td>12288 gigabytes<\/td>Custom (Kyushu's implementation)<\/td>33<\/td>128.336<\/td>6382 seconds<\/td>6264 Watts<\/td>\n<\/tr>\n
58<\/td>58<\/td>Colonial<\/td>Supermicro<\/td>Intel(R) Xeon(R) CPU E5-2620 x 2 NVIDIA K40c x 2<\/td><\/td>George Washington University<\/td>Washington DC<\/td>USA<\/td>2014<\/td>Computer Engineering<\/td>University<\/td>1<\/td>12<\/td>20 gigabytes<\/td>Custom<\/td>20<\/td>122.182<\/td>72.8705 seconds<\/td>274 Watts<\/td>\n<\/tr>\n
59<\/td>59<\/td>Endeavor<\/td>Intel<\/td>Dual-socket 2.6GHz SNB-EP<\/td>Mellanox Infiniband Febric<\/td>Intel Dupont \/ Parallel Computing Lab Intel Labs<\/td>Dupont Washington<\/td>USA<\/td>2012<\/td>Geophysics automotive and commercial benchmarking for all fields<\/td>Government university industry and commercial benchmarking<\/td>320<\/td>5120<\/td>20480 gigabytes<\/td>custom<\/td>32<\/td>115.935<\/td>44.6482 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
60<\/td>60<\/td>Titan_Keldysh<\/td>Intel(R) Server<\/td>CPU Intel(R) Xeon(R) E5-1660v2 (6 cores 3.7GHz) && GPU NVidia GTX Titan<\/td>none<\/td>Keldysh Institute of Applied Mathematics (Russian Academy of Sciences)<\/td>Moscow State University<\/td>Russia<\/td>2012<\/td>Scientific Research<\/td>University<\/td>1<\/td>20<\/td>40 gigabytes<\/td>Custom GPU<\/td>25<\/td>114.68<\/td>58.6 seconds<\/td>212 Watts<\/td>\n<\/tr>\n
61<\/td>61<\/td>TSUBAME-KFC<\/td>NEC<\/td>self-made<\/td>Infiniband FDR<\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2013<\/td>Research<\/td>University<\/td>32<\/td>384<\/td>2048 gigabytes<\/td>custom<\/td>32<\/td>104.31<\/td>133.436 seconds<\/td>7419.85 Watts<\/td>\n<\/tr>\n
62<\/td>62<\/td>Lomonosov<\/td>T-Platforms<\/td>MPP<\/td><\/td>Moscow State University<\/td><\/td>Russia<\/td><\/td><\/td><\/td>4096<\/td>32768<\/td>0 gigabytes<\/td>Custom<\/td>37<\/td>103.079<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
63<\/td>63<\/td>Jugene<\/td>IBM<\/td>BlueGene\/P<\/td><\/td>Forschungszentrum Juelich<\/td><\/td>Germany<\/td><\/td><\/td><\/td>64000<\/td>256000<\/td>0 gigabytes<\/td>Custom<\/td>37<\/td>92.3418<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
64<\/td>64<\/td>Big Red 2<\/td>Cray Inc.<\/td>XE6\/XK7<\/td>Cray Gemini<\/td>Indiana University<\/td>Bloomington Indiana<\/td>USA<\/td>2013<\/td>Academic research<\/td>University<\/td>1020<\/td>21824<\/td>43648 gigabytes<\/td>Custom<\/td>32<\/td>88.8<\/td>8.61388 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
65<\/td>65<\/td>HYPATIA<\/td>Data Vortex Technologies<\/td>DV206<\/td>Data Vortex Proprietary<\/td>Plexus Engineering<\/td>Neenah Wi<\/td>USA<\/td>2015<\/td>Scientific Computing Graph Analytics<\/td>Development<\/td>32<\/td>256<\/td>8192 gigabytes<\/td>Custom<\/td>33<\/td>88.112<\/td>4344.97 seconds<\/td>10498.2 Watts<\/td>\n<\/tr>\n
66<\/td>66<\/td>Intrepid<\/td>IBM<\/td>BlueGene\/P<\/td><\/td>DOE\/SC\/Argonne National Laboratory<\/td>Argonne National Laboratory<\/td>USA<\/td>2009<\/td>Science<\/td>Government<\/td>32768<\/td>131071<\/td>65536 gigabytes<\/td>Custom<\/td>35<\/td>80.64<\/td>26.73 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
67<\/td>67<\/td>Graph-WICIL-Node0<\/td>MICRON<\/td>PicoComputing AC520<\/td>NA<\/td>University of Wisconsin-Madison<\/td>Madison WI<\/td>United States<\/td>2018<\/td>Research<\/td>Research<\/td>1<\/td>18<\/td>4<\/td>Custom<\/td>23<\/td>79.531<\/td>29.6908<\/td>56.8<\/td>\n<\/tr>\n
68<\/td>68<\/td>SALOMON<\/td>HPE\/SGI<\/td>SGI ICE X (2016 Intel Xeon E5-2680 v3 12-core processors @ 2.5 GHz (24192 cores) + 864 Intel Xeon Phi 7120P 61-core accelerators @ 1.238 GHz (52704 cores)<\/td>FDR Infiniband<\/td>IT4Innovations national supercomputing center VSB-Technical University of Ostrava<\/td>Ostrava<\/td>Czech Republic<\/td>2015<\/td>Research<\/td>University Industry<\/td>722<\/td>17328<\/td>0 gigabytes<\/td>Tuned MPI Implementation with 2-D data distribution from http: \/\/www.graph500.org\/referencecode<\/td>33<\/td>70.7969<\/td>23.5858 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
69<\/td>69<\/td>Sierra<\/td>IBM \/ NVIDIA \/ Mellanox<\/td>IBM Power System S922LC IBM POWER9 22C 3.1GHz NVIDIA Volta GV100 Dual-rail Mellanox EDR Infiniband<\/td>Dual-rail Mellanox EDR Infiniband<\/td>DOE\/NNSA\/LLNL<\/td>Livermore CA<\/td>United States<\/td>2018<\/td>Government<\/td>Scientific Research<\/td>2048<\/td>81920<\/td>524288<\/td>Custom NVRAM out-of-core<\/td>42<\/td>67.258<\/td>11572.795<\/td><\/td>\n<\/tr>\n
70<\/td>70<\/td>GraphCREST-Huawei<\/td>Huawei<\/td>Huawei RH5885H V3 (4-way Intel Xeon E7-4890 v2 and 2TB RAM)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2014<\/td>Graph analysis<\/td>University<\/td>1<\/td>60<\/td>2048 gigabytes<\/td>Custom (OpenMP + NUMA-aware + Degree-aware)<\/td>32<\/td>59.9<\/td>3209.9 seconds<\/td>1153.1 Watts<\/td>\n<\/tr>\n
71<\/td>71<\/td>JunGPU<\/td>Nvidia<\/td>E5-2683 v3 Titan X<\/td><\/td>Tsinghua University<\/td>Beijing<\/td>China<\/td>2016<\/td>High performance computing<\/td>University<\/td>1<\/td>3072<\/td>12 gigabytes<\/td>Custom<\/td>26<\/td>48.295<\/td>22.3719 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
72<\/td>72<\/td>altix<\/td>HPE\/SGI<\/td>SGI Altix UV 1000<\/td><\/td>Research Center for Advanced Computing Infrastructure Japan Advanced Institute of Science and Technology<\/td>Ishikawa<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>16<\/td>256<\/td>2048 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>31<\/td>47.071<\/td>9906.29 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
73<\/td>73<\/td>GraphCREST-Sandybridge-EP-2.7GHz<\/td>HPCTECH Corporation<\/td>Intel(R) Xeon(R) CPU E5-4650 @ 2.70GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Graph analysis<\/td>University<\/td>1<\/td>32<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>27<\/td>45.71<\/td>234.545 seconds<\/td>665 Watts<\/td>\n<\/tr>\n
74<\/td>74<\/td>GraphCREST-SandybridgeEP-2.4GHz<\/td>HPCTECH Corporation<\/td>Intel Xeon CPU E5-4640 2.40GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Graph analysis<\/td>University<\/td>1<\/td>32<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>27<\/td>42.45<\/td>260.017 seconds<\/td>520.9 Watts<\/td>\n<\/tr>\n
75<\/td>75<\/td>GraphCREST-Westmere-EX-2.4GHz<\/td>HPCTECH Corporation<\/td>Intel(R) Xeon(R) CPU E7 4870 2.40GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2012<\/td>Graph analysis<\/td>University<\/td>1<\/td>40<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>27<\/td>42.114<\/td>238.947 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
76<\/td>76<\/td>I950-G<\/td>Sugon<\/td>8x Intel Xeon X7550 CPUs (8 core 2.0GHz)<\/td><\/td>Institute of Computing Technology Chinese Academy of Sciences<\/td>Beijing<\/td>China<\/td>2011<\/td>Research<\/td>University<\/td>16<\/td>1024<\/td>4096 gigabytes<\/td>Optimized<\/td>32<\/td>41.9924<\/td>423.075 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
77<\/td>77<\/td>IBM Power8+ Tesla P100<\/td>IBM && Nvidia<\/td>IBM Power8+ Tesla P100<\/td><\/td>IBM Moscow && MSU Alex Kolganov<\/td>Moscow<\/td>Russian Federation<\/td>2016<\/td>research<\/td>research<\/td>1<\/td>66<\/td>256 gigabytes<\/td>GPU Custom<\/td>30<\/td>41.7<\/td>2713 sec seconds<\/td>235 Watts<\/td>\n<\/tr>\n
78<\/td>78<\/td>Monty Pi-thon<\/td>Newark Element14<\/td>Raspberry Pi 2 Model B<\/td>10\/100 Fast Ethernet driven through onboard USB 2.0 bus connected to Dell PowerConnect 6248 Switch via flat topology<\/td>CODE@TACC (Texas Advanced Computing Center Summer High School STEM Program)<\/td>Austin Texas<\/td>USA<\/td>2015<\/td>Education Outreach<\/td>University<\/td>32<\/td>128<\/td>32 gigabytes<\/td>graph500-2.1.4-energy<\/td>26<\/td>35.8254<\/td>403.705 seconds<\/td>147.175 Watts<\/td>\n<\/tr>\n
79<\/td>79<\/td>Super2PiR<\/td>Newark Element14<\/td>Raspberry Pi 2 Model B<\/td>10\/100 Fast Ethernet driven through onboard USB 2.0 bus connected to Dell PowerConnect 6248 Switch via flat topology<\/td>CODE@TACC (Texas Advanced Computing Center Summer High School STEM Program)<\/td>Austin Texas<\/td>USA<\/td>2015<\/td>Education Outreach<\/td>University<\/td>32<\/td>128<\/td>32 gigabytes<\/td>Reference 2.1.4 with *validation disabled*<\/td>26<\/td>35.5917<\/td>552.526 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
80<\/td>80<\/td>kitty6<\/td>SYSTEM WORKS<\/td>Xeon E5-2650 v2 GeForce GTX TITAN Black<\/td><\/td>University of Tsukuba<\/td>Tsukuba<\/td>Japan<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>8<\/td>32 gigabytes<\/td>Custom (GPU)<\/td>25<\/td>35.18<\/td>335.485 seconds<\/td>206.164 Watts<\/td>\n<\/tr>\n
81<\/td>new<\/td>High Throughput Computer<\/td>Institute of Computing Technology Chinese Academy of Science<\/td>Single-Node High Throughput Computer<\/td><\/td>SMARCO<\/td>Beijing<\/td>China<\/td>2019<\/td>IOT3.0 Big Data<\/td>High Throughput Big Data Processing<\/td>1<\/td>40<\/td>384<\/td>Custom BFS v2<\/td>30<\/td>34.49<\/td>478.52<\/td>122 watts<\/td>\n<\/tr>\n
82<\/td>81<\/td>GraphCREST-Sandybridge-EP-2.7GHz<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>64<\/td><\/td><\/td>27<\/td>31.648<\/td><\/td>771.71 Watts<\/td>\n<\/tr>\n
83<\/td>82<\/td>RSC KNL<\/td>RSC && Intel<\/td>RSC Intel Xeon KNL 7250<\/td><\/td>RSC Moscow && MSU Alex Kolganov<\/td>Moscow<\/td>Russian Federation<\/td>2017<\/td>research<\/td>research<\/td>1<\/td>68<\/td>98 gigabytes<\/td>Custom<\/td>29<\/td>31.3<\/td>1742 sec seconds<\/td>215 Watts<\/td>\n<\/tr>\n
84<\/td>83<\/td>GraphCREST-4Way-SandybridgeEP-2.4GHz<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>64<\/td><\/td><\/td>27<\/td>29.034<\/td><\/td>639.1 Watts<\/td>\n<\/tr>\n
85<\/td>84<\/td>NTHU Quanta Cluster<\/td>Quanta Computer<\/td>Quanta - Intel(R) Xeon(R) E5-2670 2.60GHz<\/td>InfiniBand<\/td>National Tsing Hua University<\/td>Hsinchu<\/td>Taiwan<\/td>2012<\/td>HPC applications<\/td>university<\/td>8<\/td>128<\/td>0 gigabytes<\/td>reference<\/td>8<\/td>28.3746<\/td>67.4514 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
86<\/td>85<\/td>GraphCREST-Magny-Cours<\/td>Dell<\/td>AMD Opteron(tm) Processor 6174 2.2 GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2011<\/td>Graph analysis<\/td>University<\/td>1<\/td>48<\/td>256 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>27<\/td>25.243<\/td>359.16 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
87<\/td>86<\/td>rc1.umb.edu<\/td>Microway<\/td>Microway NumberSmasher with two Intel Xeon 6150<\/td><\/td>UMass Boston<\/td>Boston<\/td>United States<\/td>1<\/td>university<\/td>research<\/td>1<\/td>36<\/td>192<\/td>custom<\/td>27<\/td>23.77<\/td>398<\/td><\/td>\n<\/tr>\n
88<\/td>88<\/td>UV 2000<\/td>HPE\/SGI<\/td>SGI UV 2000<\/td>NUMAlink6<\/td>SGI<\/td>Chippewa Falls WI<\/td>USA<\/td>2012<\/td>Vendor<\/td>Vendor<\/td>128<\/td>2048<\/td>32768 gigabytes<\/td>Reference<\/td>34<\/td>19.62<\/td>1607.04 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
89<\/td>89<\/td>Alkindi27-hybrid<\/td>Dell<\/td>Single node commodity server<\/td><\/td>The University of British Columbia<\/td>Vancouver<\/td>Canada<\/td>2015<\/td>University<\/td>Research<\/td>1<\/td>28<\/td>256 gigabytes<\/td>custom<\/td>29<\/td>19.4685<\/td>634.11 seconds<\/td>579.6 Watts<\/td>\n<\/tr>\n
90<\/td>90<\/td>Franklin<\/td>Cray<\/td>XT4<\/td><\/td>Lawrence Berkeley National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>4000<\/td>16000<\/td>0 gigabytes<\/td>Custom<\/td>36<\/td>19.3274<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
91<\/td>91<\/td>Tryton<\/td>Action\/WASKO<\/td>Actina SOLAR 820 S6<\/td>InfiniBand<\/td>Academic Computer Centre in Gdansk (CI TASK)<\/td>Gda\u0144sk<\/td>Poland<\/td>2015<\/td>BiochemistryOceanagraphyCFD<\/td>University<\/td>147<\/td>3528<\/td>0 gigabytes<\/td>mpi-tuned-2d<\/td>29<\/td>18.9<\/td>95.4345 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
92<\/td>92<\/td>PowerfullServer<\/td>Intel<\/td>Intel Xeon E5 2699 v3<\/td><\/td>Alex Kolganov<\/td>Moscow<\/td>Russian Federation<\/td>2016<\/td>research<\/td>research<\/td>1<\/td>18<\/td>384 gigabytes<\/td>Custom<\/td>30<\/td>16.3<\/td>2700 sec seconds<\/td>145 Watts Watts<\/td>\n<\/tr>\n
93<\/td>new<\/td>Alkindi28<\/td>Dell<\/td>Commodity Machine \/ Intel Xeon E7-4870 v2 (4 Sockets 60 cores 120 hardware threads)<\/td><\/td>The University of British Columbia<\/td>Vancouver BC<\/td>Canada<\/td>2015<\/td>University<\/td>Research<\/td>1<\/td>60<\/td>1536<\/td>custom NUMA-aware - Joint work of UBC and RIKEN-AIP<\/td>31<\/td>15.98<\/td>3874<\/td>0<\/td>\n<\/tr>\n
94<\/td>93<\/td>Todi<\/td>Cray Inc.<\/td>Xk7<\/td>Gemini<\/td>Swiss National Supercomputing Center<\/td>Lugano<\/td>Switzerland<\/td>2011<\/td>General purpose scientific computing<\/td>University<\/td>272<\/td>8160<\/td>10336 gigabytes<\/td>Custom developed by M. Fatica group at Nvidia<\/td>29<\/td>15.5983<\/td>382.819 seconds<\/td>79012 Watts<\/td>\n<\/tr>\n
95<\/td>94<\/td>Nano-Graphybrid++<\/td>AMD<\/td>4 AMD Radeon R9 Nano GPUs<\/td><\/td>AMD Research<\/td>Austin TX<\/td>USA<\/td>2015<\/td>Architecture Testbed<\/td>Industry<\/td>1<\/td>16<\/td>128 gigabytes<\/td>Custom (Hybrid++)<\/td>25<\/td>15.58<\/td>22.9 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
96<\/td>95<\/td>EBD-RH2288V3<\/td>Huawei<\/td>RH2288\/V3 (Intel\u00aeSSD 750 series 1.2TB crucial\u00aem4 mSATA 256GB x8)<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2015<\/td>Research<\/td>University<\/td>1<\/td>36<\/td>256 gigabytes<\/td>Custom (netalx-v7.7fz-sg OpenMP NUMA-optimized NVM)<\/td>32<\/td>14.8566<\/td>23556.3 seconds<\/td>334.985 Watts<\/td>\n<\/tr>\n
97<\/td>96<\/td>fox6<\/td>Convey<\/td>MX-100 host-210<\/td>n\/a<\/td>Convey Computer Corporation<\/td>Richardson TX<\/td>USA<\/td>2012<\/td>big data<\/td>vendor<\/td>1<\/td>16<\/td>384 gigabytes<\/td>custom<\/td>29<\/td>14.56<\/td>7682 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
98<\/td>97<\/td>Altix ICE 8400EX<\/td>HPE\/SGI<\/td>Altix ICE 8400EX<\/td><\/td>SGI<\/td><\/td>USA<\/td><\/td><\/td><\/td>256<\/td>1024<\/td>0 gigabytes<\/td>Reference<\/td>31<\/td>13.9586<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
99<\/td>98<\/td>GraphCREST-Custom #1<\/td>Custom<\/td>Intel\u00ae Xeon\u2122 E5-2690 crucial@m4 mSATA 256GB x16<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>32<\/td>256 gigabytes<\/td>Custom (netalx-v7.7fz-sg OpenMP NUMA-optimized NVM)<\/td>31<\/td>13.7963<\/td>90234.5 seconds<\/td>391.825 Watts<\/td>\n<\/tr>\n
100<\/td>99<\/td>GraphCREST-Bulldozer<\/td>HPCTECH Corporation<\/td>AMD Opteron(tm) Processor 6386 SE (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Graph analysis<\/td>University<\/td>1<\/td>64<\/td>1024 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>31<\/td>13.633<\/td>3218.79 seconds<\/td>790.8 Watts<\/td>\n<\/tr>\n
101<\/td>100<\/td>IBM POWER8+<\/td>IBM<\/td>IBM POWER8+<\/td><\/td>IBM Moscow & MSU Alex Kolganov<\/td>Moscow<\/td>Russian Federation<\/td>2016<\/td>research<\/td>research<\/td>1<\/td>10<\/td>256 gigabytes<\/td>Custom<\/td>30<\/td>13.2<\/td>2713 sec seconds<\/td>200 watts Watts<\/td>\n<\/tr>\n
102<\/td>101<\/td>Colonial0<\/td>Microway<\/td>Intel(R) Xeon(R) CPU E5-2683 v3 @ 2.00GHz<\/td><\/td>George Washington University<\/td>Washington DC<\/td>USA<\/td>2015<\/td>High-Performance Computing Big Data Computing<\/td>Computer Engineering<\/td>1<\/td>28<\/td>320 gigabytes<\/td>Custom (Self-develop code)<\/td>30<\/td>12.9<\/td>532.52 seconds<\/td>350 Watts<\/td>\n<\/tr>\n
103<\/td>102<\/td>MIC<\/td>Intel<\/td>Xeon Phi 7120P<\/td><\/td>Tsinghua University<\/td>Tsinghua University<\/td>China<\/td>2014<\/td>High performance computing<\/td>University<\/td>1<\/td>61<\/td>16 gigabytes<\/td>custom<\/td>25<\/td>12.87<\/td>14.7 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
104<\/td>103<\/td>Nebulae<\/td>Sugon<\/td>TC3600 Blade System Xeon X5650 6C 2.66GHz<\/td>InfiniBand QDR<\/td>National Supercomputing Centre in Shenzhen<\/td>Shenzhen<\/td>China<\/td>2010<\/td>Multiphysics Fluid mechanics<\/td>Industry<\/td>512<\/td>6144<\/td>12288 gigabytes<\/td>Custom<\/td>33<\/td>12.119<\/td>94.3242 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
105<\/td>104<\/td>ROMEO HPC Center<\/td>Bull Atos<\/td>BullX R421<\/td>Infiniband QDR<\/td>University of Reims Champagne-Ardenne<\/td>Reims<\/td>France<\/td>2013<\/td>All scientific domains<\/td>Academic and Industry<\/td>130<\/td>2080<\/td>4160 gigabytes<\/td>Custom<\/td>27<\/td>11.7806<\/td>203 seconds<\/td><\/td>\n<\/tr>\n
106<\/td>105<\/td>thunder4<\/td>Convey<\/td>HC-2ex host-110<\/td>n\/a<\/td>Convey Computer Corporation<\/td>Richardson TX<\/td>USA<\/td>2012<\/td>big data<\/td>vendor<\/td>1<\/td>12<\/td>256 gigabytes<\/td>custom<\/td>27<\/td>11.614<\/td>1432 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
107<\/td>106<\/td>Celero<\/td>Convey<\/td>HC-2ex host-210<\/td>n\/a<\/td>Argonne National Laboratory<\/td>Illinois<\/td>USA<\/td>2012<\/td>Bioinformatics<\/td>DOE<\/td>1<\/td>8<\/td>192 gigabytes<\/td>custom<\/td>27<\/td>11.448<\/td>1368 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
108<\/td>107<\/td>Convey HC-2ex<\/td>Convey<\/td>HC-2ex host-210<\/td><\/td>UC Riverside<\/td>Riverside CA<\/td>USA<\/td>2012<\/td>HPC<\/td>University<\/td>1<\/td>8<\/td>200 gigabytes<\/td>custom<\/td>27<\/td>11.447<\/td>1414 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
109<\/td>108<\/td>coconino<\/td>Convey<\/td>HC-2ex host-210<\/td>n\/a<\/td>Convey Computer Corporation<\/td>Richardson TX<\/td>USA<\/td>2012<\/td>big data<\/td>vendor<\/td>1<\/td>8<\/td>448 gigabytes<\/td>custom<\/td>27<\/td>11.444<\/td>1470 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
110<\/td>109<\/td>Intel-8180<\/td>Intel<\/td>Intel<\/td><\/td>Shenzhen Institutes of Advanced Technology(SIAT) Chinese Academy of Science<\/td>Shenzhen<\/td>China<\/td>2019<\/td>Scientific Research<\/td>Deep Learning<\/td>1<\/td>56<\/td>96<\/td>Custom<\/td>23<\/td>11.3<\/td>117<\/td>0<\/td>\n<\/tr>\n
111<\/td>110<\/td>Alkindi28-CPU<\/td>Dell<\/td>Commodity Machine \/ Intel Xeon E7-4870 v2 (4 Sockets 60 cores 120 hardware threads)<\/td><\/td>The University of British Columbia<\/td>Vancouver BC<\/td>Canada<\/td>2015<\/td>University<\/td>Research<\/td>1<\/td>60<\/td>1536 gigabytes<\/td>Custom<\/td>32<\/td>11.1678<\/td>31889.4 seconds<\/td>591.024 Watts<\/td>\n<\/tr>\n
112<\/td>111<\/td>GraphCREST-4waySBep2.4<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>32<\/td><\/td><\/td>26<\/td>11.115<\/td><\/td>639 Watts<\/td>\n<\/tr>\n
113<\/td>112<\/td>GraphCREST-Wex40<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>40<\/td><\/td><\/td>26<\/td>11.061<\/td><\/td>1010.13 Watts<\/td>\n<\/tr>\n
114<\/td>113<\/td>Grace<\/td>Cray<\/td>XMT2<\/td>Cray SeaStar<\/td>Mayo Clinic Rochester<\/td>Rochester MN<\/td>USA<\/td>2011<\/td>Semantic Data Analysis and Bioinformatics Number of Processors<\/td>Research<\/td>64<\/td>64<\/td>2048 gigabytes<\/td>custom<\/td>31<\/td>10.9415<\/td>4035.8 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
115<\/td>114<\/td>RAIDEN MMS Subsystem<\/td>Fujitsu<\/td>single node commodity server (Intel(R) Xeon(R) Platinum 8160M CPU @ 2.10GHz)<\/td><\/td>RIKEN AIP<\/td>Tokyo<\/td>Japan<\/td>2018<\/td>Research<\/td>scientific research<\/td>1<\/td>96<\/td>6144<\/td>custom (NUMA-aware Totem from the University of British Columbia)<\/td>34<\/td>10.63<\/td>55582<\/td><\/td>\n<\/tr>\n
116<\/td>115<\/td>grace<\/td>Cray<\/td>XMT-2<\/td><\/td>Mayo Clinic<\/td>Rochester MN<\/td>USA<\/td>2011<\/td>Medicine<\/td>Research<\/td>64<\/td>64<\/td>2048 gigabytes<\/td>Custom<\/td>31<\/td>10.319<\/td>4306 seconds<\/td>14100 Watts<\/td>\n<\/tr>\n
117<\/td>116<\/td>Alkindi26-hybrid<\/td>Dell<\/td>Commodity Machine \/ Intel Xeon E5-2670 (2 Sockets 20 cores 40 hardware threads) Nvidia Kepler K40 (2 GPUs)<\/td><\/td>The University of British Columbia<\/td>Vancouver BC<\/td>Canada<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>20<\/td>512 gigabytes<\/td>Custom<\/td>30<\/td>10.3079<\/td>7040 seconds<\/td>461 Watts<\/td>\n<\/tr>\n
118<\/td>117<\/td>Altix UV 1000<\/td>HPE\/SGI<\/td>Altix UV 1000<\/td><\/td>SGI<\/td><\/td>USA<\/td><\/td><\/td><\/td>128<\/td>2048<\/td>0 gigabytes<\/td>Reference<\/td>32<\/td>10.0932<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
119<\/td>118<\/td>TH-IVB-FEP\/C<\/td>National University of Defense Technology<\/td>SMP<\/td><\/td>Changsha China<\/td>Changsha China<\/td>China<\/td>2013<\/td>Science research Big data<\/td>Research<\/td>1<\/td>24<\/td>64 gigabytes<\/td>custom<\/td>26<\/td>9.74402<\/td>180.843 seconds<\/td>248 Watts<\/td>\n<\/tr>\n
120<\/td>119<\/td>Xeon Server<\/td>Dell<\/td>4 x Intel(R) Xeon(R) Gold 5115 CPU @ 2.40GHz<\/td><\/td>Industry<\/td>Boise ID<\/td>United States<\/td>2017<\/td>Research<\/td>Research<\/td>1<\/td>32<\/td><\/td>heavy-newreference<\/td>21<\/td>9.68<\/td>3.78<\/td><\/td>\n<\/tr>\n
121<\/td>120<\/td>Red Sky<\/td>Sun Microsystems<\/td>Sun Blade x6275 Xeon X55xx 2.93 Ghz<\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>512<\/td>4096<\/td>0 gigabytes<\/td>Custom\/ETI SWARM<\/td>33<\/td>9.44893<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
122<\/td>121<\/td>Lonestar<\/td>Dell<\/td>PowerEdge 1955 2.66 GHz Infiniband<\/td><\/td>TACC<\/td><\/td>USA<\/td><\/td><\/td><\/td>1024<\/td>12288<\/td>0 gigabytes<\/td>Custom\/ETI SWARM<\/td>35<\/td>9.23418<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
123<\/td>123<\/td>PowerEdge M610<\/td>Dell<\/td>PowerEdge M610<\/td><\/td>TACC<\/td><\/td>USA<\/td><\/td><\/td><\/td>512<\/td>6144<\/td>0 gigabytes<\/td><\/td>34<\/td>8.05306<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
124<\/td>124<\/td>MEM-CREST Node #2<\/td>Supermicro<\/td>2027GR-TRF w\/ FusionIO ioDrive2 1.2TB x2<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>1<\/td>16<\/td>128 gigabytes<\/td>Custom (netalx-v7.7fz-sg OpenMP NUMA-optimized NVRAM)<\/td>30<\/td>7.98417<\/td>52984.9 seconds<\/td>276.507 Watts<\/td>\n<\/tr>\n
125<\/td>125<\/td>DAS-4\/VU<\/td>SuperMicro<\/td><\/td><\/td>VU University<\/td><\/td>Netherlands<\/td><\/td><\/td><\/td>128<\/td>1024<\/td>0 gigabytes<\/td>Reference MPI\/OpenMP<\/td>30<\/td>7.0867<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
126<\/td>126<\/td>IBM BlueGene\/P<\/td>IBM<\/td>BlueGene\/P<\/td><\/td>Moscow State University<\/td><\/td>Russia<\/td><\/td><\/td><\/td>2048<\/td>8192<\/td>0 gigabytes<\/td>Custom<\/td>32<\/td>6.87195<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
127<\/td>127<\/td>GraphCREST-SBep2.9<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>16<\/td><\/td><\/td>25<\/td>6.777<\/td><\/td>362.79 Watts<\/td>\n<\/tr>\n
128<\/td>128<\/td>Intel Westmere X5670<\/td>Intel<\/td>Westmere X5670<\/td><\/td>Intel<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>3072<\/td>0 gigabytes<\/td><\/td>33<\/td>6.76457<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
129<\/td>129<\/td>Vortex<\/td>Convey<\/td>HC-1ex<\/td><\/td>Convey Computer Corporation<\/td>Richardson TX<\/td>USA<\/td>2012<\/td>Geophysics automotive etc.<\/td>Government university industry etc.<\/td>1<\/td>4<\/td>192 gigabytes<\/td>custom<\/td>27<\/td>6.63583<\/td>3747.82 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
130<\/td>130<\/td>URAN<\/td>HP<\/td>Cluster\/16 nodes 32 CPU Intel Xeon X5675<\/td>IB DDR<\/td>Institute of Mathematics and Mechanics UrB RAS<\/td>Yekaterinburg<\/td>Russia<\/td>2012<\/td>Mathematical and physical modelling<\/td>Science education industry<\/td>16<\/td>192<\/td>1904 gigabytes<\/td>graph500_custom_imm<\/td>26<\/td>6.25902<\/td>18.6411 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
131<\/td>131<\/td>JaguarPF<\/td>Cray<\/td>XT5<\/td><\/td>Oak Ridge National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>512<\/td>1024<\/td>0 gigabytes<\/td>Custom\/ETI SWARM<\/td>33<\/td>6.2277<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
132<\/td>132<\/td>GraphCREST-SBep2.0<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>16<\/td><\/td><\/td>25<\/td>6.142<\/td><\/td>298.44 Watts<\/td>\n<\/tr>\n
133<\/td>133<\/td>Convey HC-1ex<\/td>Convey<\/td>Computer Corporation HC-1ex<\/td>Infiniband<\/td>Virginia Bioinformatics Institute<\/td>Blacksburg VA<\/td>USA<\/td>2011<\/td>Bioinformatics<\/td>University<\/td>1<\/td>4<\/td>160 gigabytes<\/td>Custom (Convey)<\/td>27<\/td>5.9962<\/td>3588.6 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
134<\/td>134<\/td>Cray XT4<\/td>Cray<\/td>XT4<\/td><\/td>NERSC<\/td><\/td>USA<\/td><\/td><\/td><\/td>9544<\/td>38176<\/td>0 gigabytes<\/td>Optimized<\/td>32<\/td>5.60493<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
135<\/td>135<\/td>UND Cluster<\/td>Dell<\/td>Cluster<\/td><\/td>University of Notre Dame<\/td>Notre Dame IN<\/td>USA<\/td>2016<\/td>Research<\/td>University<\/td>8<\/td>128<\/td>2048 gigabytes<\/td>New Reference<\/td>27<\/td>5.55522<\/td>60.0327<\/td>0 watts<\/td>\n<\/tr>\n
135<\/td>135<\/td>University of Notre Dame cluster<\/td>Dell<\/td>cluster<\/td>Ethernet<\/td>University of Notre Dame<\/td>University of Notre Dame<\/td>United States<\/td>2016<\/td>University<\/td>Research<\/td>8<\/td>128<\/td>2048 gigabytes<\/td>newreference+heavy<\/td>27<\/td>5.55522<\/td>60.0327 seconds<\/td>Watts<\/td>\n<\/tr>\n
137<\/td>137<\/td>ULP2012 ultra-green oil-cooled server : GraphCREST<\/td>self-made<\/td>Xeon E5-2620 2.0GHz x 2 (12 cores 24 threads) (oil cooling) \/ 1 node<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2012<\/td>high-performance computing<\/td>university<\/td>1<\/td>12<\/td>32 gigabytes<\/td>custom (OpenMP)<\/td>24<\/td>5.12372<\/td>25.8703 seconds<\/td>168.5 Watts<\/td>\n<\/tr>\n
138<\/td>138<\/td>mirasol<\/td>Intel<\/td>Westmere-EX<\/td><\/td>Georgia Tech<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>10<\/td>0 gigabytes<\/td><\/td>28<\/td>5.04659<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
139<\/td>139<\/td>SuperMicro<\/td>SuperMicro<\/td><\/td><\/td>VU University<\/td><\/td>Netherlands<\/td><\/td><\/td><\/td>128<\/td>0<\/td>0 gigabytes<\/td><\/td>31<\/td>4.61709<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
140<\/td>140<\/td>MIC<\/td>Intel<\/td>Xeon Phi 7120P<\/td><\/td>PACMAN Group<\/td>Tsinghua University Beijing<\/td>China<\/td>2014<\/td>High performance computing<\/td>University<\/td>1<\/td>61<\/td>16 gigabytes<\/td>custom<\/td>25<\/td>4.53221<\/td>19.3895 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
141<\/td>141<\/td>Blacklight<\/td>HPE\/SGI<\/td>Altix UV1000<\/td><\/td>PSC<\/td><\/td>USA<\/td><\/td><\/td><\/td>32<\/td>512<\/td>0 gigabytes<\/td>Custom<\/td>32<\/td>4.40234<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
142<\/td>142<\/td>MIC<\/td>Intel<\/td>Xeon Phi 7120P<\/td><\/td>DISLab.org NICEVT \/ svetcorp.net<\/td>Moscow<\/td>Russia<\/td>2013<\/td>Computer engineering<\/td>Industry<\/td>1<\/td>61<\/td>16 gigabytes<\/td>Custom<\/td>25<\/td>4.36578<\/td>491.4 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
143<\/td>143<\/td>Alkindi26-CPU<\/td>Dell<\/td>Commodity Machine \/ Intel Xeon E5-2670 (2 Sockets 20 cores 40 hardware threads)<\/td><\/td>The University of British Columbia<\/td>Vancouver BC<\/td>Canada<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>20<\/td>512 gigabytes<\/td>Custom<\/td>30<\/td>4.24712<\/td>3247 seconds<\/td>391 Watts<\/td>\n<\/tr>\n
144<\/td>144<\/td>Catalyst<\/td>Cray \/ Intel<\/td>Cray CS300 Supercomputer dual Intel E5-2695v2 processors 128GB memory Intel 910 800GB SSD<\/td>Intel QDR-80 dual rail Infiniband<\/td>Lawrence Livermore National Laboratory<\/td>Livermore CA<\/td>USA<\/td>2014<\/td>Research<\/td>Government<\/td>300<\/td>7200<\/td>38400 gigabytes<\/td>Custom<\/td>40<\/td>4.175<\/td>172800 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
145<\/td>145<\/td>GPU-based cluster<\/td>Self-assembled<\/td>GPU-based cluster<\/td><\/td>Seoul National University Korea<\/td>Seoul<\/td>South Korea<\/td>2010<\/td>Computer Science<\/td>University<\/td>8<\/td>192<\/td>384 gigabytes<\/td>custom<\/td>26<\/td>4.06453<\/td>512.861 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
146<\/td>146<\/td>Mustang<\/td>Appro<\/td>Xtreme-X cluster with dual-socket 12-core AMD MagnyCours processors<\/td>Mellanox InfiniScale IV 4x-QDR InfiniBand<\/td>Los Alamos National Laboratory<\/td>Los Alamos New Mexico<\/td>USA<\/td>2012<\/td>open science (any field)<\/td>government<\/td>86<\/td>2064<\/td>5504 gigabytes<\/td>graph500_mpi_replicated_csc from v2.1.4<\/td>31<\/td>3.79502<\/td>455.079 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
147<\/td>147<\/td>I840r-GP<\/td>Sugon<\/td>4x Intel Xeon E7-4830 CPU (8 core 2.13GHz)<\/td><\/td>Institute of Computing Technology Chinese Academy of Sciences<\/td>Beijing<\/td>China<\/td>2012<\/td>Research<\/td>University<\/td>1<\/td>32<\/td>1024 gigabytes<\/td>Custom<\/td>30<\/td>3.69<\/td>15400.8 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
148<\/td>148<\/td>EBD-RH5885v2<\/td>Huawei<\/td>Tecal RH5885 V2 Server w\/ Tecal ES3000 PCIe SSD 800GBx2 1.2Tx2<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2013<\/td>Research<\/td>University<\/td>1<\/td>48<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA-optimized Hybrid Double Graph NVRAM)<\/td>30<\/td>3.67475<\/td>2822.84 seconds<\/td>844.337 Watts<\/td>\n<\/tr>\n
149<\/td>149<\/td>MEM-CREST Node #1<\/td>Supermicro<\/td>Intel\u00ae Xeon\u2122 E5-2660 x 2socets FusionIO ioDrive2 1.2TB<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>1<\/td>16<\/td>64 gigabytes<\/td>Custom (OpenMP NUMA-optimized Hybrid Single Graph NVRAM)<\/td>29<\/td>3.11171<\/td>6686.73 seconds<\/td>303.6 Watts<\/td>\n<\/tr>\n
150<\/td>150<\/td>Todi<\/td>Cray<\/td>XK7<\/td><\/td>CSCS<\/td><\/td>Switzerland<\/td><\/td><\/td><\/td>0<\/td>0<\/td>0 gigabytes<\/td>Custom GPU Result<\/td>28<\/td>3.00648<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
151<\/td>151<\/td>GraphCREST-M48<\/td>Dell<\/td>AMD Opteron 6174 2.2GHz<\/td><\/td>Chuo University<\/td>Tokyo<\/td>Japan<\/td>2011<\/td>Mathematical optimization<\/td>University<\/td>1<\/td>48<\/td>256 gigabytes<\/td>custom<\/td>29<\/td>2.42145<\/td>276.558 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
152<\/td>152<\/td>APU<\/td>AMD<\/td>Kaveri(AMD A10-7850K Radeon R7)<\/td><\/td>Tsinghua University<\/td>Beijing<\/td>China<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>4<\/td>32 gigabytes<\/td>Custom<\/td>26<\/td>2.07<\/td>14.6525 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
153<\/td>153<\/td>GraphCREST-Retina15<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>4<\/td><\/td><\/td>24<\/td>1.987<\/td><\/td>38.5 Watts<\/td>\n<\/tr>\n
154<\/td>154<\/td>GraphCREST-Mac-mini<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>4<\/td><\/td><\/td>24<\/td>1.941<\/td><\/td>36.3 Watts<\/td>\n<\/tr>\n
155<\/td>155<\/td>B2-mic0<\/td>Intel<\/td>Xeon Phi 5110P<\/td><\/td>DISLab.org NICEVT<\/td>Moscow<\/td>Russia<\/td>2013<\/td>Computer engineering<\/td>Industry<\/td>1<\/td>60<\/td>8 gigabytes<\/td>Custom<\/td>23<\/td>1.80143<\/td>13.78 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
156<\/td>156<\/td>GraphCREST-7048GR-TR<\/td>Supermicro<\/td>7048GR-TR (Huawei ES3000 2.4TB)<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2015<\/td>Research<\/td>University<\/td>1<\/td>36<\/td>128 gigabytes<\/td>Custom (netalx-v7.7fz-sg OpenMP NUMA-optimized NVM)<\/td>31<\/td>1.77788<\/td>11561.8 seconds<\/td>336.018 Watts<\/td>\n<\/tr>\n
157<\/td>157<\/td>GLADOS<\/td>Convey<\/td>Computer Corporation HC-1<\/td><\/td>Imperial College London<\/td>South Kensington Campus London<\/td>UK<\/td>2010<\/td>Research on acceleration of data-intensive applications<\/td>University<\/td>1<\/td>6<\/td>128 gigabytes<\/td>unknown<\/td>24<\/td>1.72364<\/td>393 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
158<\/td>158<\/td>Dingus<\/td>Convey<\/td>HC-1<\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>4<\/td>0 gigabytes<\/td>Convey Custom<\/td>28<\/td>1.71799<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
158<\/td>158<\/td>Wingus<\/td>Convey<\/td>HC-1<\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>4<\/td>0 gigabytes<\/td>Convey Custom<\/td>27<\/td>1.71799<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
160<\/td>160<\/td>Graphybrid++<\/td>AMD<\/td>AMD A10-7850K APU + AMD W9100 Firepro GPU<\/td><\/td>AMD Research<\/td>Austin TX<\/td>USA<\/td>2014<\/td>Architecture Testbed<\/td>Industry<\/td>1<\/td>4<\/td>24 gigabytes<\/td>Custom (Hybrid++)<\/td>25<\/td>1.68555<\/td>205.9 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
161<\/td>161<\/td>EBD-GoldenBox-Prototype<\/td>NVIDIA<\/td>Jetson TK1<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>5<\/td>2 gigabytes<\/td>Custom (OpenMP Degree-aware BFS)<\/td>21<\/td>1.63503<\/td>40.153 seconds<\/td>8 Watts<\/td>\n<\/tr>\n
162<\/td>162<\/td>Convey01<\/td>Convey<\/td>HC-1<\/td><\/td>Bielefeld University CeBiTec<\/td><\/td>Germany<\/td><\/td><\/td><\/td>1<\/td>4<\/td>0 gigabytes<\/td>Convey Custom<\/td>28<\/td>1.61061<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
162<\/td>162<\/td>HC-1d<\/td>Convey<\/td>Computer Corporation HC-1<\/td><\/td>Convey Computer Corporation<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>4<\/td>0 gigabytes<\/td>Convey Custom<\/td>28<\/td>1.61061<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
164<\/td>164<\/td>Convey2<\/td>Convey<\/td>HC-1<\/td><\/td>Lawrence Berkeley National Laboratory\/NERSC<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>4<\/td>0 gigabytes<\/td>Convey Custom<\/td>28<\/td>1.50324<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
164<\/td>164<\/td>IBM with ScaleMP\/vSMP Foundation<\/td>IBM<\/td><\/td>ScaleMP\/vSMP Fundation<\/td>LArgonne National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>128<\/td>0 gigabytes<\/td>Reference<\/td>29<\/td>1.50324<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
166<\/td>166<\/td>Anton's Macbook<\/td>Apple<\/td>Macbook Pro 2017 LPDDR3<\/td><\/td>personal<\/td>Boise ID<\/td>United States<\/td>2017<\/td>laptop<\/td>laptop<\/td>1<\/td>2<\/td>16<\/td>custom heavy+dir opt<\/td>20<\/td>1.4153<\/td>9.06<\/td>39<\/td>\n<\/tr>\n
167<\/td>167<\/td>SuperDragon-1<\/td>unknown<\/td><\/td><\/td>Institute of Computing Technology<\/td>Beijing<\/td>China<\/td><\/td><\/td><\/td>32<\/td>384<\/td>0 gigabytes<\/td>Custom<\/td>30<\/td>1.39586<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
168<\/td>168<\/td>GraphNode<\/td>Intel Sever<\/td>Intel(R) Xeon(R) CPU E5-2683 v3<\/td><\/td>ICT<\/td>China<\/td>United States<\/td>2015<\/td>University<\/td>Scientific research<\/td>1<\/td>28<\/td>384<\/td>custom<\/td>25<\/td>1.3358<\/td>162<\/td><\/td>\n<\/tr>\n
169<\/td>169<\/td>Cray XMT<\/td>Cray<\/td>XMT<\/td><\/td>Pacific Northwest National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>128<\/td>128<\/td>0 gigabytes<\/td>Optimized<\/td>29<\/td>1.30997<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
170<\/td>170<\/td>PLX<\/td>IBM<\/td>PLX<\/td><\/td>CINECA<\/td>Bologna<\/td>Italy<\/td><\/td><\/td><\/td>32<\/td>64<\/td>0 gigabytes<\/td>GPU Customized<\/td>26<\/td>1.28849<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
171<\/td>171<\/td>EBD-prototype-m<\/td>Handmade<\/td>Custom (Intel\u00ae Core\u2122 i7 @ 3.50GHz crucial@m4 msata 256GB * 16)<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2013<\/td>Research<\/td>University<\/td>1<\/td>4<\/td>16 gigabytes<\/td>Custom (OpenMP NUMA-optimized Hybrid Single Graph NVRAM)<\/td>28<\/td>1.28359<\/td>10674 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
172<\/td>172<\/td>Cray XMT<\/td>Cray<\/td>XMT<\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>128<\/td>128<\/td>0 gigabytes<\/td>Optimized<\/td>29<\/td>1.25628<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
173<\/td>173<\/td>Anton's Macbook Pro<\/td>Apple<\/td>laptop<\/td><\/td>personal<\/td>Boise ID<\/td>United States<\/td>2017<\/td>personal<\/td>research<\/td>1<\/td>2<\/td><\/td>custom_aml_heavy_diropt<\/td>17<\/td>1.22818<\/td>0.602644<\/td><\/td>\n<\/tr>\n
174<\/td>174<\/td>GraphCREST-MBA13<\/td>Apple<\/td>Macbook Air 13inch (Intel(R) Core(TM) i5-2557M CPU @ 1.70GHz) \/ 1node<\/td><\/td>Chuo University<\/td>Tokyo<\/td>Japan<\/td>2012<\/td>mathematical optimization<\/td>university<\/td>1<\/td>2<\/td>8 gigabytes<\/td>custom (OpenMP)<\/td>23<\/td>1.22761<\/td>69.7941 seconds<\/td>23.6 Watts<\/td>\n<\/tr>\n
175<\/td>175<\/td>Alkindi<\/td>Supermicro workstation<\/td>Commodity Machine \/ Intel Xeon E5-2650 (2 Sockets 32 cores) Nvidia Kepler K20c (2 GPUs)<\/td><\/td>The University of British Columbia<\/td>Vancouver BC<\/td>Canada<\/td>2012<\/td>Generic<\/td>University<\/td>1<\/td>0<\/td>256 gigabytes<\/td>custom<\/td>28<\/td>1.20576<\/td>2860 seconds<\/td>496 Watts<\/td>\n<\/tr>\n
176<\/td>176<\/td>cougarxmt<\/td>Cray<\/td>XMT<\/td><\/td>Pacific Northwest National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>128<\/td>128<\/td>0 gigabytes<\/td>Reference<\/td>29<\/td>1.18112<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
176<\/td>176<\/td>GraphCREST-W12<\/td>AMD<\/td>Opteron 6174 2.2GHz<\/td><\/td>Chuo University<\/td>Tokyo<\/td>Japan<\/td><\/td><\/td><\/td>1<\/td>12<\/td>0 gigabytes<\/td><\/td>0<\/td>1.18112<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
178<\/td>178<\/td>PowerEdge R815 Opteron 6174<\/td>Dell<\/td>PowerEdge R815 Opteron 6174<\/td>Infiniband<\/td>STE Lab Nagoya University<\/td>Nagoya<\/td>Japan<\/td>2010<\/td>Geophysics and astrophysics<\/td>University<\/td>4<\/td>192<\/td>384 gigabytes<\/td>unknown<\/td>22<\/td>1.1623<\/td>2898.1 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
179<\/td>179<\/td>GraphCREST-Intel-NUC<\/td>Intel<\/td>Intel\u00ae NUC Kit DC3217IYE \/ 1 node<\/td><\/td>Chuo University<\/td>Tokyo<\/td>Japan<\/td>2013<\/td>mathematical optimization<\/td>university<\/td>1<\/td>2<\/td>8 gigabytes<\/td>custom (OpenMP)<\/td>23<\/td>1.08175<\/td>71.1322 seconds<\/td>20.1 Watts<\/td>\n<\/tr>\n
180<\/td>180<\/td>graphstorm<\/td>Cray<\/td>XMT<\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>128<\/td>128<\/td>0 gigabytes<\/td>Reference<\/td>29<\/td>1.07374<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
181<\/td>181<\/td>GraphCREST-Xperia-Z1-SO-01F<\/td>Sony<\/td>Android Smartphone (Qualcomm Snapdragon 800 MSM8974 2.2GHz)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Smartphone<\/td>Smartphone<\/td>1<\/td>4<\/td>2 gigabytes<\/td>Custom (OpenMP Android-NDK Degree-aware BFS)<\/td>20<\/td>1.033<\/td>19.149 seconds<\/td>4.393 Watts<\/td>\n<\/tr>\n
182<\/td>182<\/td>ScaleMP\/vSMP Foundation<\/td>IBM<\/td><\/td>ScaleMP\/vSMP Fundation<\/td>LArgonne National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>128<\/td>0 gigabytes<\/td>Reference<\/td>30<\/td>1.03079<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
183<\/td>183<\/td>Hyperion<\/td>Dell Intel Supermicro QLogic Cisco Mellanox DDN Sun LSI and RedHat<\/td><\/td><\/td>Lawrence Livermore National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>64<\/td>512<\/td>0 gigabytes<\/td>Custom<\/td>31<\/td>0.99858<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
184<\/td>184<\/td>Cray XT5-HE<\/td>Cray<\/td>XT5-HE<\/td><\/td>Oak Ridge National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>18688<\/td>224256<\/td>0 gigabytes<\/td><\/td>30<\/td>0.966368<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
185<\/td>185<\/td>tara<\/td>IBM<\/td>Server x iDataPlex<\/td>InfiniBand (QDR)<\/td>University of Maryland Baltimore County<\/td>Baltimore MD<\/td>USA<\/td>2009<\/td>Mathematics and statistics computer science atmospheric physics computational chemistry mechanical engineering and more<\/td>University<\/td>86<\/td>688<\/td>2088 gigabytes<\/td>Modified reference implementation<\/td>31<\/td>0.946<\/td>135.525 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
186<\/td>186<\/td>Matterhorn<\/td>Cray Inc.<\/td>XMT2<\/td>Seastar<\/td>Swiss National Supercomputing Center<\/td>Lugano<\/td>Switzerland<\/td>2011<\/td>General purpose scientific computing<\/td>University<\/td>64<\/td>64<\/td>2048 gigabytes<\/td>Green500-modified Cray version<\/td>31<\/td>0.897<\/td>2203 seconds<\/td>19557 Watts<\/td>\n<\/tr>\n
187<\/td>187<\/td>Matterhorn<\/td>Cray<\/td>XMT<\/td><\/td>CSCS<\/td><\/td>Switzerland<\/td><\/td><\/td><\/td>64<\/td>64<\/td>0 gigabytes<\/td>Reference<\/td>31<\/td>0.880468<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
188<\/td>188<\/td>Cray XMT2<\/td>Cray<\/td>XMT2<\/td><\/td>CSCS<\/td><\/td>Switzerland<\/td><\/td><\/td><\/td>64<\/td>64<\/td>0 gigabytes<\/td><\/td>31<\/td>0.858993<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
189<\/td>189<\/td>Alkindi-Jetson<\/td>Nvidia<\/td>Jetson<\/td><\/td>The University of British columbia<\/td>Vancouver<\/td>Canada<\/td>2015<\/td>Research<\/td>University<\/td>0<\/td>0<\/td>0 gigabytes<\/td>Custom<\/td>22<\/td>0.850828<\/td>37.54 seconds<\/td>6.346 Watts<\/td>\n<\/tr>\n
190<\/td>190<\/td>Minerva<\/td>Intel<\/td>Westmere 2.66 GHz<\/td><\/td>University of Warwick<\/td><\/td>UK<\/td><\/td><\/td><\/td>1<\/td>12<\/td>0 gigabytes<\/td>Reference<\/td>26<\/td>0.837519<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
191<\/td>191<\/td>AMD-APU<\/td>AMD<\/td>AMD A10-7850K APU<\/td><\/td>AMD Research<\/td>Austin TX<\/td>USA<\/td>2014<\/td>Architecture Testbed<\/td>Industry<\/td>1<\/td>0<\/td>16 gigabytes<\/td>Custom (Hybrid++)<\/td>24<\/td>0.827<\/td>89.6 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
192<\/td>192<\/td>GraphCREST-Xperia-A-SO-04E<\/td>Sony<\/td>Android Smartphone (Qualcomm Snapdragon S4 Pro APQ8064 1.5GHz)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Smartphone<\/td>Smartphone<\/td>1<\/td>4<\/td>2 gigabytes<\/td>Custom (OpenMP Android-NDK Degree-aware BFS)<\/td>20<\/td>0.739<\/td>26.664 seconds<\/td>3.2074 Watts<\/td>\n<\/tr>\n
193<\/td>193<\/td>Brokoli Cluster<\/td>Inspur<\/td>Intel Xeon E5-2670v3 2.3GHz<\/td>FDR Infiniband<\/td>Telkom University<\/td>Bandung<\/td>Indonesia<\/td>2017<\/td>Academic<\/td>Scientific computing<\/td>1<\/td>48<\/td>128 gigabytes<\/td>Graph500-2.1.4 omp-csr<\/td>20<\/td>0.7128<\/td>0.467 seconds<\/td>Watts<\/td>\n<\/tr>\n
194<\/td>194<\/td>Erdos<\/td>unknown<\/td><\/td><\/td>Oak Ridge National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>0<\/td>0 gigabytes<\/td>Reference<\/td>29<\/td>0.697932<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
194<\/td>194<\/td>Intel Westmere E7-4870 2.4GHz<\/td>Intel<\/td>E7-4870 2.4GHz<\/td><\/td>Intel Research<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>40<\/td>0 gigabytes<\/td><\/td>27<\/td>0.697932<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
196<\/td>196<\/td>Intel Westmere X5670 2.93<\/td>Intel<\/td>Westmere X5670 2.93 IB network<\/td><\/td>Intel Corporation<\/td><\/td>USA<\/td><\/td><\/td><\/td>128<\/td>256<\/td>0 gigabytes<\/td>Reference<\/td>26<\/td>0.660351<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
197<\/td>197<\/td>Hyperion + FusionIO<\/td>unknown<\/td><\/td><\/td>Lawrence Livermore National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>64<\/td>512<\/td>0 gigabytes<\/td>Custom\/NAND FLASH<\/td>36<\/td>0.601295<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
198<\/td>198<\/td>GraphCREST-NEXUS7-2013<\/td>ASUS<\/td>Android Tablet (Qualcomm Snapdragon S4 Pro 1.5GHz)<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2013<\/td>Research<\/td>University<\/td>1<\/td>4<\/td>2 gigabytes<\/td>Custom (OpenMP Android NDK)<\/td>20<\/td>0.534713<\/td>33.3676 seconds<\/td>4.1248 Watts<\/td>\n<\/tr>\n
199<\/td>199<\/td>Nehalem X5570<\/td>Intel<\/td>Nehalem X5570<\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>512<\/td>0 gigabytes<\/td>Reference<\/td>28<\/td>0.512175<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
200<\/td>200<\/td>Aziz<\/td>Fujitsu<\/td>Cluster<\/td>IB (QDR) Intel True Scale<\/td>king abdulaziz university<\/td>Jeddah<\/td>Saudi Arabia<\/td>2015<\/td>Genomics Weather<\/td>University<\/td>0<\/td>0<\/td>0 gigabytes<\/td>MPI-Replicated<\/td>16<\/td>0.478866<\/td>0.170506 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
201<\/td>201<\/td>GraphCREST-Xperia-A-SO-04E<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>4<\/td><\/td><\/td>20<\/td>0.478<\/td><\/td>3.12071 Watts<\/td>\n<\/tr>\n
202<\/td>202<\/td>Vedur<\/td>HP<\/td>AMD Opteron 6276 CPU (32 cores) @ 2.3GHz<\/td>4x QDR Infiniband<\/td>University of Tartu<\/td>Tartu<\/td>Estonia<\/td>2011<\/td>University<\/td>General<\/td>8<\/td>512<\/td><\/td>Reference<\/td>21<\/td>0.455214<\/td>10.284522<\/td><\/td>\n<\/tr>\n
203<\/td>203<\/td>ultraviolet<\/td>unknown<\/td><\/td><\/td>Sandia National Laboratories<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>32<\/td>0 gigabytes<\/td>Custom<\/td>29<\/td>0.418759<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
204<\/td>204<\/td>Viridis HPC@Uni.lu<\/td>Boston<\/td>Boston Viridis - ARM Microservers 2U<\/td>Ethernet 10G<\/td>University of Luxembourg<\/td>Campus Belval Esch-sur-Alzette<\/td>Luxembourg<\/td>2013<\/td>Computer Science research<\/td>University<\/td>32<\/td>128<\/td>128 gigabytes<\/td>Reference implementation graph500_mpi_replicated<\/td>21<\/td>0.389643<\/td>8.71395 seconds<\/td>206.38 Watts<\/td>\n<\/tr>\n
205<\/td>205<\/td>Owens<\/td>unknown<\/td><\/td><\/td>The Ohio State University<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>0<\/td>0 gigabytes<\/td>Reference<\/td>26<\/td>0.257698<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
206<\/td>206<\/td>Trestles<\/td>Appro<\/td>1U-1143H Cluster Opteron 8C 2.4Ghz<\/td>Infiniband QDR<\/td>UC San Diego<\/td>San Diego CA<\/td>USA<\/td><\/td><\/td><\/td>144<\/td>4608<\/td>0 gigabytes<\/td>Custom\/NAND FLASH<\/td>36<\/td>0.236223<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
207<\/td>207<\/td>Okorok<\/td>Dell<\/td>PowerEdge C1100<\/td><\/td>Home<\/td>Yekaterinburg<\/td>Russia<\/td>2013<\/td>Natural language processing<\/td>Personal<\/td>1<\/td>8<\/td>48 gigabytes<\/td>Reference OpenMP<\/td>17<\/td>0.228<\/td>0.208 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
208<\/td>208<\/td>LANL-LARP-hbchen-001<\/td>HP<\/td>Z840<\/td>Local<\/td>Los Alamos National Lab<\/td>Los Alamos New Mexico<\/td>United States<\/td>2016<\/td>Government<\/td>Metadata indexing<\/td>1<\/td>24<\/td>64 gigabytes<\/td>Reference<\/td>27<\/td>0.179422<\/td>47.2368 seconds<\/td>Watts<\/td>\n<\/tr>\n
209<\/td>209<\/td>Knot<\/td>unknown<\/td><\/td><\/td>UC Santa Barbara<\/td>Santa Barabara CA<\/td>USA<\/td><\/td><\/td><\/td>0<\/td>64<\/td>0 gigabytes<\/td>Custom\/NAND FLASH<\/td>30<\/td>0.171799<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
210<\/td>210<\/td>GraphCREST-Tegra3<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>4<\/td><\/td><\/td>20<\/td>0.154<\/td><\/td>2.402 Watts<\/td>\n<\/tr>\n
211<\/td>211<\/td>larp-hbchen-002<\/td>HP<\/td>Z840<\/td>NO<\/td>UltraScale Systems research Center Los Alamos National lab<\/td>Los Alamos New Mexcio<\/td>United States<\/td>2016<\/td>Government Lab<\/td>Storage system metadata indexing research<\/td>1<\/td>24<\/td>gigabytes<\/td>graph500_reference_bfs_sssp<\/td>29<\/td>0.13869<\/td>56.219773 seconds<\/td>Watts<\/td>\n<\/tr>\n
212<\/td>212<\/td>GraphCREST-NEXUS10<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>2<\/td><\/td><\/td>20<\/td>0.119<\/td><\/td>3.69 Watts<\/td>\n<\/tr>\n
213<\/td>213<\/td>Twizzle<\/td>Dell<\/td>Optiplex 9010 Intel Ivy Bridge Core i7 3770<\/td><\/td>Sandia National Laboratories<\/td>New Mexico USA<\/td>USA<\/td>2012<\/td>Architectural Testbed<\/td>Government<\/td>1<\/td>4<\/td>16 gigabytes<\/td>Reference OpenMP-CSR<\/td>23<\/td>0.0992<\/td>39.18 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
214<\/td>214<\/td>Kraken<\/td>Cray<\/td><\/td><\/td>Lawrence Livermore National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>6128<\/td>0 gigabytes<\/td>Custom\/NAND FLASH<\/td>31<\/td>0.0966368<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
215<\/td>215<\/td>Caribou<\/td>unknown<\/td><\/td><\/td>Link Analytics<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>0<\/td>0 gigabytes<\/td>Reference<\/td>27<\/td>0.0751619<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
216<\/td>216<\/td>Bertha<\/td>Supermicro \/ Saratoga Speed<\/td>Quad-socket Intel E7-4870v2 server w\/ Saratoga Speed Altamont XP flash arrays<\/td><\/td>Lawrence Livermore National Laboratory<\/td>Livermore CA<\/td>USA<\/td>2014<\/td>Research<\/td>Government<\/td>1<\/td>60<\/td>2048 gigabytes<\/td>Custom<\/td>37<\/td>0.0627<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
217<\/td>217<\/td>Leviathan<\/td>Dell<\/td><\/td>FusionIO<\/td>Lawrence Livermore National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>40<\/td>0 gigabytes<\/td>Custom\/NAND FLASH<\/td>36<\/td>0.0536871<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
217<\/td>217<\/td>Appro<\/td>Dell<\/td><\/td><\/td>Lawrence Livermore National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>1<\/td>32<\/td>0 gigabytes<\/td><\/td>34<\/td>0.0536871<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
219<\/td>219<\/td>Simons PC<\/td>Intel<\/td>Desktop Computer<\/td><\/td>Home<\/td>Aachen<\/td>Germany<\/td>2012<\/td>Research<\/td>Private<\/td>1<\/td>4<\/td>8 gigabytes<\/td>Reference omp-csr<\/td>23<\/td>0.05199<\/td>19.12 seconds<\/td><\/td>\n<\/tr>\n
220<\/td>220<\/td>Neumann<\/td>AMD<\/td><\/td><\/td>UC Santa Barbara<\/td>Santa Barabara CA<\/td>USA<\/td><\/td><\/td><\/td>0<\/td>32<\/td>0 gigabytes<\/td>Custom<\/td>26<\/td>0.0322123<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
221<\/td>221<\/td>Scott Beamer's iPad<\/td>Apple<\/td>iPad 3<\/td>none<\/td>UC Berkeley<\/td>Berkeley CA<\/td>USA<\/td>2012<\/td>research<\/td>university<\/td>1<\/td>2<\/td>1 gigabytes<\/td>direction-optimizing BFS<\/td>14<\/td>0.0304<\/td>13.211 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
222<\/td>222<\/td>Chaos (LABOS T-Rex)<\/td>DALCO AG<\/td>DALCO r2164i5 (2x Intel Xeon E5-2630v3 2x 6TB SATAIII)<\/td>40G Mellanox<\/td>Ecole Polytechnique F\u00e9d\u00e9rale de Lausanne<\/td>Lausanne<\/td>Switzerland<\/td>2015<\/td>University<\/td>Computer Science<\/td>8<\/td>128<\/td>128<\/td>Chaos (http:\/\/labos.epfl.ch\/chaos)<\/td>33<\/td>0.0264<\/td>46 minutes 12 seconds<\/td>1824<\/td>\n<\/tr>\n
223<\/td>223<\/td>Gordon<\/td>Appro<\/td>Xtreme-X GreenBlade GB512X Xeon E5-2670 8C 2.600GHz<\/td>Infiniband QDR<\/td>San Diego Supercomputing Center<\/td>San Diego CA<\/td>USA<\/td><\/td><\/td><\/td>7<\/td>84<\/td>0 gigabytes<\/td>Reference<\/td>29<\/td>0.0214748<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>1<\/td>K computer<\/td>Fujitsu<\/td>Custom<\/td>Tofu<\/td>RIKEN Advanced Institute for Computational Science (AICS)<\/td>Kobe Hyogo<\/td>Japan<\/td>2011<\/td>Various scientific and instudrial fields<\/td>Academic and industry<\/td>82944<\/td>663552<\/td>1327100 gigabytes<\/td>Custom<\/td>40<\/td>31302.4<\/td>609.395 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 32768 partition<\/td>IBM<\/td>BGQ<\/td>IBM BGQ proprietary<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>US DOE National Lab<\/td>Scientific Computing<\/td>32768<\/td>524288<\/td>524288<\/td>custom<\/td>38<\/td>14716<\/td>429.7<\/td>0<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>DOE\/SC\/Argonne National Laboratory Mira<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>49152<\/td>786432<\/td>786432 gigabytes<\/td>Custom<\/td>40<\/td>14328<\/td>836 seconds<\/td>3244030 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 16384 partition<\/td>IBM<\/td>BGQ<\/td>IBM BGQ proprietary<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>US DOE National Lab<\/td>Scientific Computing<\/td>16384<\/td>262144<\/td>262144<\/td>custom<\/td>37<\/td>8661<\/td>314.8<\/td><\/td>\n<\/tr>\n
-1<\/td>6<\/td>JUQUEEN<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>Forschungszentrum Juelich (FZJ)<\/td>Juelich<\/td>Germany<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>16384<\/td>262144<\/td>262144 gigabytes<\/td>Custom<\/td>38<\/td>5848<\/td>565 seconds<\/td>1081340 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 8192 partition<\/td>IBM<\/td>Argonne National Laboratory<\/td>IBM BGQ proprietary<\/td>Chicago IL<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>US DOE National Lab<\/td>Scientific Computing<\/td>8192<\/td>131072<\/td>131072<\/td>custom<\/td>36<\/td>5422<\/td>247<\/td><\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 8192 partition<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td><\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>8192<\/td>131072<\/td>131072 gigabytes<\/td>Custom<\/td>36<\/td>4212<\/td>467 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 4096 partition<\/td>IBM<\/td>BGQ<\/td>IBM BGQ proprietary<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>US DOE National Lab<\/td>Scientific Simulation<\/td>4096<\/td>65536<\/td>65536<\/td>custom<\/td>35<\/td>3069.6<\/td>241<\/td>0<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 4096 partition<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>5D Torus<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>Scientific Research<\/td>Research<\/td>4096<\/td>65536<\/td>65536 gigabytes<\/td>Custom<\/td>35<\/td>2348<\/td>462 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 4096 partition<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>5D torus<\/td>DOE\/ALCF<\/td>Argonne National Laboratory<\/td>USA<\/td>2012<\/td>Scientific Computing<\/td>Government<\/td>4096<\/td>65536<\/td>16 gigabytes<\/td>Custom<\/td>35<\/td>2050<\/td>993.063 seconds<\/td><\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 2048 partition<\/td>IBM<\/td>BGQ<\/td>IBM BGQ proprietary<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>US DOE National Lab<\/td>Scientific Simulation<\/td>2048<\/td>32768<\/td>32768<\/td>custom<\/td>34<\/td>1954<\/td>207.5<\/td>0<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Cetus - 2048<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>5D torus<\/td>DOE\/ALCF<\/td>Argonne National Laboratory<\/td>USA<\/td>2012<\/td>Scientific Computing<\/td>Government<\/td>2048<\/td>32768<\/td>16 gigabytes<\/td>Custom<\/td>34<\/td>1249<\/td>953.185 seconds<\/td><\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Nurion<\/td>Cray<\/td>CS500 Intel Xeon Phi 7250 68C 1.4GHz Intel Omni-Path<\/td>Intel Omni-Path<\/td>Korea Institute of Science and Technology Information<\/td>Daejeon<\/td>Korea Republic Of<\/td>2018<\/td>Research<\/td>Various scientific and instudrial fields<\/td>3844<\/td>246016<\/td>369024<\/td>ComBLAS<\/td>37<\/td>1048.86<\/td>3949.43<\/td><\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Mira - 1024 partition<\/td>IBM<\/td>BGQ<\/td>IBM BGQ proprietary<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>US DOE National Lab<\/td>Scientific Computing<\/td>1024<\/td>16384<\/td>16384<\/td>custom<\/td>33<\/td>1037<\/td>192<\/td><\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Oakleaf-FX<\/td>Fujitsu<\/td>Fujitsu PRIMEHPC FX 10<\/td>Tofu interconnect<\/td>Information Technology Center The University of Tokyo<\/td>Japan<\/td>Japan<\/td>2012<\/td>Academic<\/td>University<\/td>4800<\/td>76800<\/td>153600 gigabytes<\/td>Custom<\/td>38<\/td>992.6<\/td>1028.74 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Cetus<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>5D Torus<\/td>Argonne National Laboratory<\/td>Chicago IL<\/td>United States<\/td>2012<\/td>Scientific Research<\/td>Government<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>custom<\/td>33<\/td>740<\/td>401 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ALCF Cetus<\/td>IBM<\/td>IBM - BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>5D torus<\/td>DOE\/ALCF<\/td>Argonne National Laboratory<\/td>USA<\/td>2012<\/td>Scientific Computing<\/td>Government<\/td>1024<\/td>16384<\/td>16 gigabytes<\/td>Custom<\/td>33<\/td>705.1<\/td>930.961 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Oakleaf-FX<\/td>Fujitsu<\/td>PRIMEHPC FX10<\/td>Fujitsu Tofu interconnect<\/td>Information Technology Center The University of Tokyo<\/td>Tokyo<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>4800<\/td>76800<\/td>153600 gigabytes<\/td>Custom<\/td>38<\/td>609.37<\/td>1656.1 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Cetus<\/td>IBM<\/td>BlueGene\/Q Power BQC 16C 1.60 GHz<\/td>Custom<\/td>DOE\/SC\/Argonne National Laboratory<\/td>Chicago IL<\/td>USA<\/td>2012<\/td>Scientific Research<\/td>Government<\/td>1024<\/td>16384<\/td>16384 gigabytes<\/td>Custom<\/td>34<\/td>382<\/td>754 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>TSUBAME 2.0<\/td>HP<\/td>Cluster Platform SL390s G7<\/td>Infiniband QDR<\/td>GSIC Center Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2010<\/td>Research<\/td>University<\/td>1366<\/td>16392<\/td>73764 gigabytes<\/td>Custom<\/td>36<\/td>202.68<\/td>1213.19 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>ismuv2k2<\/td>HPE\/SGI<\/td>SGI UV 2000<\/td>NUMAlink6<\/td>The Institute of Statistical Mathematics<\/td>Tokyo<\/td>Japan<\/td>2014<\/td>Data assimilation<\/td>National research institute<\/td>1<\/td>640<\/td>16384 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>32<\/td>131.427<\/td>5646.93 seconds<\/td>10530 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>HA-PACS<\/td>Appro International \/ Cray Inc.<\/td>Xtream-X GreenBlade 8204 Xeon E5-2670 8C 2.600GHz Infiniband QDR NVIDIA 2090<\/td>Infiniband QDR<\/td>Center for Computational Sciences University of Tsukuba<\/td>Tsukuba<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>256<\/td>4096<\/td>32768 gigabytes<\/td>Custom (GPU)<\/td>31<\/td>117.546<\/td>21.1513 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>SGI UV 300<\/td>HPE\/SGI<\/td>UV 300 (16-socket Intel(R) Xeon(R) CPU E7-8867 v3 8 TB RAM)<\/td>NUMAlink7<\/td>SGI<\/td>Chippewa Falls WI<\/td>USA<\/td>2015<\/td>Vendor<\/td>Vendor<\/td>1<\/td>256<\/td>8192 gigabytes<\/td>Custom (NUMA-aware OpenMP with ULIBC SGI & Kyushu University)<\/td>33<\/td>99.794<\/td>3971.38 seconds<\/td>4214 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Huawei<\/td>Huawei<\/td>Huawei RH5885H V3 (4-way Intel Xeon E7-4890 v2 and 2TB RAM)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2014<\/td>Graph analysis<\/td>University<\/td>1<\/td>60<\/td>2048 gigabytes<\/td>Custom (OpenMP + NUMA-aware)<\/td>32<\/td>55.741<\/td>4844.38 seconds<\/td>1255 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Lomonosov<\/td>T-Platforms<\/td>MPP<\/td><\/td>Moscow State University<\/td><\/td>Russia<\/td><\/td><\/td><\/td>4096<\/td>8192<\/td>0 gigabytes<\/td><\/td>37<\/td>43.3792<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-SandybridgeEP-2.4GHz<\/td>HPCTECH Corporation<\/td>Intel Xeon CPU E5-4640 2.40GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2012<\/td>Graph analysis<\/td>University<\/td>1<\/td>32<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>27<\/td>41.802<\/td>280.082 seconds<\/td>479.8 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>altix<\/td>HPE\/SGI<\/td>SGI Altix UV1000<\/td>Numalink5<\/td>Research Center for Advanced Computing Infrastructure Japan Advanced Institute of Science and Technology<\/td>Ishikawa<\/td>Japan<\/td>2012<\/td>Research<\/td>University<\/td>32<\/td>512<\/td>4096 gigabytes<\/td>Custom (OpenMP NUMA-optimized)<\/td>30<\/td>37.6593<\/td>12167.8 seconds<\/td>19873 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Sandybridge-EP-2.7GHz<\/td>HPCTECH Corporation<\/td>Intel(R) Xeon(R) CPU E5-4650 @ 2.70GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Graph analysis<\/td>University<\/td>1<\/td>32<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>30<\/td>31.95<\/td>2256.67 seconds<\/td>661.7 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-SandybridgeEP-2.4GHz<\/td>HPCTECH Corporation<\/td>Intel Xeon CPU E5-4640 2.40GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Graph analysis<\/td>University<\/td>1<\/td>32<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA-aware ULIBC)<\/td>30<\/td>31.325<\/td>1798.21 seconds<\/td>497.8 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Westmere-EX-2.4GHz<\/td>HPCTECH Corporation<\/td>Intel(R) Xeon(R) CPU E7 4870 2.40GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2012<\/td>Graph analysis<\/td>University<\/td>1<\/td>40<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>30<\/td>30.212<\/td>2726.76 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Huawei<\/td>Huawei<\/td>RH5885 V3(Huawei ES3000 2.4TB x2 Intel\u00aeSSD DC S3500 series 800GB x8)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>60<\/td>2048 gigabytes<\/td>Custom (netalx-v7.7fz-sg OpenMP NUMA-optimized NVM)<\/td>33<\/td>29.2924<\/td>79276.8 seconds<\/td>1407.31 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-SandybridgeEP-2.4GHz<\/td>HPCTECH Corporation<\/td>Intel Xeon CPU E5-4640 2.40GHz (4 sockets)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Graph analysis<\/td>University<\/td>1<\/td>32<\/td>512 gigabytes<\/td>Custom (OpenMP NUMA Degree-aware BFS)<\/td>30<\/td>28.614<\/td>2376.16 seconds<\/td>465.4 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Sandybridge-EP-2.7GHz<\/td><\/td><\/td><\/td>Chuo University<\/td><\/td><\/td><\/td><\/td><\/td><\/td>64<\/td><\/td><\/td>29<\/td>22.983<\/td><\/td>779.88 Watts<\/td>\n<\/tr>\n
-1<\/td>87<\/td>High Throughput Computer<\/td>Institute of Computing Technology Chinese Academy of Science<\/td>Single Node ARM Server<\/td><\/td>SMARCO<\/td>Beijing<\/td>China<\/td>2018<\/td>Big Data<\/td>High Throughput big data processingroug<\/td>1<\/td>40<\/td>384<\/td>Custom<\/td>30<\/td>22.24<\/td>478.02<\/td>134<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>High Throughput Computer<\/td>Institute of Computing Technology Chinese Academy of Science<\/td>Single Node ARM Server<\/td><\/td>SMARCO<\/td>Beijing<\/td>China<\/td>2018<\/td>Big Data<\/td>High throughput big data processing<\/td>1<\/td>40<\/td>384<\/td>Custom<\/td>30<\/td>21.97<\/td>482.4<\/td>134<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Alkindi27-hybrid<\/td>Dell<\/td>Single node commodity server<\/td><\/td>The University of British Columbia<\/td>Vancouver<\/td>Canada<\/td>2015<\/td>Research<\/td>University<\/td>0<\/td>0<\/td>0 gigabytes<\/td>Custom<\/td>29<\/td>17.7959<\/td>476.18 seconds<\/td>575 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>kitty6<\/td>SYSTEM WORKS<\/td>Xeon E5-2650 v2 GeForce GTX TITAN<\/td><\/td>University of Tsukuba<\/td>Tsukuba<\/td>Japan<\/td>2013<\/td>Research<\/td>University<\/td>1<\/td>8<\/td>32 gigabytes<\/td>Custom (GPU)<\/td>25<\/td>17.207<\/td>51.558 seconds<\/td>233.894 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Alkindi28-CPU<\/td>Dell<\/td>Commodity Machine \/ Intel Xeon E7-4870 v2 (4 Sockets 60 cores)<\/td><\/td>The University of British Columbia<\/td>Vancouver<\/td>Canada<\/td>2015<\/td>University<\/td>Research<\/td>1<\/td>60<\/td>1536<\/td>Custom<\/td>31<\/td>10.73<\/td>5802<\/td><\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Titan_Keldysh<\/td>Intel(R) Server<\/td>CPU Intel(R) Xeon(R) E5-1660v2 (6 cores 3.7GHz)<\/td>none<\/td>Keldysh Institute of Applied Mathematics (Russian Academy of Sciences)<\/td>Moscow State University<\/td>Russia<\/td>2012<\/td>Scientific Research<\/td>University<\/td>1<\/td>6<\/td>40 gigabytes<\/td>Custom Openmp<\/td>27<\/td>10.6<\/td>419 seconds<\/td>130 Watts<\/td>\n<\/tr>\n
-1<\/td>122<\/td>Alkindi28<\/td>Dell<\/td>Single node commodity server \/ Intel Xeon E7-4870 v2 (4 Sockets 60 cores 120 hardware threads)<\/td><\/td>The University of British Columbia<\/td>Vancouver<\/td>Canada<\/td>2015<\/td>University<\/td>Research<\/td>1<\/td>60<\/td>1500 gigabytes<\/td>custom<\/td>31<\/td>8.37<\/td>2234.67 seconds<\/td>602.06 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>NTHU Quanta Cluster<\/td>Quanta Computer<\/td>Quanta - Intel(R) Xeon(R) X5670 C12 2.93GHz<\/td>InfiniBand<\/td>National Tsing Hua University<\/td>Hsinchu<\/td>Taiwan<\/td>2012<\/td>general application<\/td>university<\/td>32<\/td>384<\/td>0 gigabytes<\/td>reference<\/td>29<\/td>5.61732<\/td>96.7725 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>NTHU Quanta Cluster<\/td>Quanta Computer<\/td>Quanta - Intel(R) Xeon(R) X5670 C12 2.93GHz<\/td>InfiniBand<\/td>National Tsing Hua University<\/td>Hsinchu<\/td>Taiwan<\/td>2012<\/td>general application<\/td>university<\/td>16<\/td>192<\/td>0 gigabytes<\/td>reference<\/td>26<\/td>4.32486<\/td>15.6816 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>NTHU Quanta Cluster<\/td>Quanta Computer<\/td>Quanta - Intel(R) Xeon(R) X5670 C12 2.93GHz<\/td>InfiniBand<\/td>National Tsing Hua University<\/td>Hsinchu<\/td>Taiwan<\/td>2012<\/td>general application<\/td>university<\/td>16<\/td>192<\/td>0 gigabytes<\/td>reference<\/td>24<\/td>4.0492<\/td>3.55049 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>NTHU Quanta Cluster<\/td>Quanta Computer<\/td>Quanta - Intel(R) Xeon(R) X5670 C12 2.93GHz<\/td>InfiniBand<\/td>National Tsing Hua University<\/td>Hsinchu<\/td>Taiwan<\/td>2012<\/td>general application<\/td>university<\/td>16<\/td>192<\/td>0 gigabytes<\/td>reference<\/td>26<\/td>3.26098<\/td>16.6105 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>EBD-RH5885v2<\/td>Huawei<\/td>Tecal RH5885 V2 Server w\/ Tecal ES3000 PCIe SSD 800GBx21.2TBx2 crucial@M500 mSATA 480GB x32<\/td><\/td>Tokyo Institute of Technology<\/td>Tokyo<\/td>Japan<\/td>2014<\/td>Research<\/td>University<\/td>1<\/td>48<\/td>1024 gigabytes<\/td>Custom (netalx-v5-sg OpenMP NUMA-optimized NVM)<\/td>33<\/td>3.10722<\/td>201717 seconds<\/td>908.22 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>URAN<\/td>HP<\/td>Cluster \/ 16 nodes HP SL390sG7 (2 CPU Intel Xeon X5675 + 8 GPU Tesla M2050 in each node)<\/td>IB DDR<\/td>Institute of Mathematics and Mechanics UrB RAS<\/td>Yekaterinburg<\/td>Russia<\/td>2013<\/td>Mathematical and physical modelling<\/td>Science education industry<\/td>16<\/td>1984<\/td>768 gigabytes<\/td>graph500_custom_imm_gpu<\/td>30<\/td>2.79635<\/td>776.35 seconds<\/td>57600 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Wex40<\/td>HPCTECH<\/td>Corporation Intel(R) Xeon(R) CPU E7-4870 2.40GHz<\/td><\/td>Chuo University<\/td>Tokyo<\/td>Japan<\/td>2012<\/td>Mathematical optimization<\/td>University<\/td>1<\/td>40<\/td>512 gigabytes<\/td>custom<\/td>30<\/td>2.16103<\/td>476.184 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>NTHU Quanta Cluster<\/td>Quanta Computer<\/td>Intel(R) Xeon(R) E5-2670 C16 2.60GHz<\/td>InfiniBand<\/td>National Tsing Hua University<\/td>Hsinchu<\/td>Taiwan<\/td>2012<\/td>general application<\/td>university<\/td>4<\/td>64<\/td>0 gigabytes<\/td>reference<\/td>18<\/td>1.3357<\/td>1.00949 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Xperia-Z1-SO-01F<\/td>Sony<\/td>Android Smartphone (Qualcomm Snapdragon 800 MSM8974 2.2GHz)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Smartphone<\/td>Smartphone<\/td>1<\/td>4<\/td>2 gigabytes<\/td>Custom (OpenMP Android-NDK Degree-aware BFS)<\/td>21<\/td>0.909<\/td>42.21 seconds<\/td>5.292 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Cray XT5-HE<\/td>Cray<\/td>XT5-HE<\/td><\/td>Oak Ridge National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>512<\/td>0<\/td>0 gigabytes<\/td>Reference<\/td>27<\/td>0.858993<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-M48<\/td>AMD<\/td>Opteron 6174 2.2GHz<\/td><\/td>Chuo University<\/td>Tokyo<\/td>Japan<\/td><\/td><\/td><\/td>1<\/td>48<\/td>0 gigabytes<\/td>Custom<\/td>27<\/td>0.826781<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>GraphCREST-Xperia-A-SO-04E<\/td>Sony<\/td>Android Smartphone (Qualcomm Snapdragon S4 Pro APQ8064 1.5GHz)<\/td><\/td>Kyushu University<\/td>Fukuoka<\/td>Japan<\/td>2013<\/td>Smartphone<\/td>Smartphone<\/td>1<\/td>4<\/td>2 gigabytes<\/td>Custom (OpenMP Android-NDK Degree-aware BFS)<\/td>21<\/td>0.588<\/td>283.645 seconds<\/td>3.253 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Intel Westmere X5670 2.93<\/td>Intel<\/td>Westmere X5670 2.93 IB network<\/td><\/td>Intel Corporation<\/td><\/td>USA<\/td><\/td><\/td><\/td>256<\/td>512<\/td>0 gigabytes<\/td>Reference<\/td>29<\/td>0.572304<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Kraken<\/td>Dell<\/td><\/td>FusionIO<\/td>Lawrence Livermore National Laboratory<\/td><\/td>USA<\/td><\/td><\/td><\/td>0<\/td>6128<\/td>0 gigabytes<\/td>Custom\/NAND FLASH<\/td>34<\/td>0.0536871<\/td>0 seconds<\/td>0 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Chaos (LABOS T-Rex)<\/td>DALCO AG<\/td>DALCO r2164i5 (2x Intel Xeon E5-2630v3 2x 6TB SATAIII)<\/td>40G Mellanox<\/td>Ecole Polytechnique F\u00e9d\u00e9rale de Lausanne<\/td>Lausanne<\/td>Switzerland<\/td>2015<\/td>Big Data<\/td>University<\/td>20<\/td>320<\/td>128 gigabytes<\/td>Chaos (http:\/\/labos.epfl.ch\/chaos)<\/td>40<\/td>0.0257<\/td>38346 seconds<\/td>4560 Watts<\/td>\n<\/tr>\n
-1<\/td>-1<\/td>Brokoli Cluster<\/td>Inspur<\/td>Intel Xeon E5-2670v3 2.3GHz<\/td>FDR Infiniband<\/td>Telkom University<\/td>Bandung<\/td>Indonesia<\/td>2017<\/td>Academic<\/td>Scientific computing<\/td>1<\/td>48<\/td>128 gigabytes<\/td>Reference<\/td>22<\/td>0.0184467<\/td>31.0229 seconds<\/td>0 Watts<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n\n","protected":false},"excerpt":{"rendered":"","protected":false},"author":53,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":[],"_links":{"self":[{"href":"https:\/\/graph500.org\/index.php?rest_route=\/wp\/v2\/pages\/781"}],"collection":[{"href":"https:\/\/graph500.org\/index.php?rest_route=\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/graph500.org\/index.php?rest_route=\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/graph500.org\/index.php?rest_route=\/wp\/v2\/users\/53"}],"replies":[{"embeddable":true,"href":"https:\/\/graph500.org\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=781"}],"version-history":[{"count":1,"href":"https:\/\/graph500.org\/index.php?rest_route=\/wp\/v2\/pages\/781\/revisions"}],"predecessor-version":[{"id":782,"href":"https:\/\/graph500.org\/index.php?rest_route=\/wp\/v2\/pages\/781\/revisions\/782"}],"wp:attachment":[{"href":"https:\/\/graph500.org\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=781"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}