{"id":64035,"date":"2021-11-15T20:00:45","date_gmt":"2021-11-15T19:00:45","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=64035"},"modified":"2021-11-15T20:00:45","modified_gmt":"2021-11-15T19:00:45","slug":"still-waiting-for-exascale-japans-fugaku-outperforms-all-competition-once-again","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/64035-still-waiting-for-exascale-japans-fugaku-outperforms-all-competition-once-again\/","title":{"rendered":"Still waiting for Exascale: Japan\u2019s Fugaku outperforms all competition once&nbsp;again"},"content":{"rendered":"<p><span class=\"caps\">FRANKFURT<\/span>, Ger\u00adma\u00adny; <span class=\"caps\">BERKELEY<\/span>, Calif.; and <span class=\"caps\">KNOXVILLE<\/span>, Tenn.\u2014 The 58th annu\u00adal edi\u00adti\u00adon of the <span class=\"caps\">TOP500<\/span> saw litt\u00adle chan\u00adge in the Top10. The Micro\u00adsoft Azu\u00adre sys\u00adtem cal\u00adled Voy\u00ada\u00adger-EUS2 was the only machi\u00adne to shake up the top spots, clai\u00adming No. 10. Based on an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> pro\u00adces\u00adsor with 48 cores and 2.45GHz working tog\u00ade\u00adther with an <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">A100<\/span> <span class=\"caps\">GPU<\/span> and 80 <span class=\"caps\">GB<\/span> of memo\u00adry, Voy\u00ada\u00adger-EUS2 also uti\u00adli\u00adzes a Mel\u00adlan\u00adox <span class=\"caps\">HDR<\/span> Infi\u00adni\u00adband for data transfer.&nbsp;<\/p>\n<p>While the\u00adre were no other chan\u00adges to the posi\u00adti\u00adons of the sys\u00adtems in the Top10, Perl\u00admut\u00adter at <span class=\"caps\">NERSC<\/span> impro\u00adved its per\u00adfor\u00admance to 70.9 Pflop\/s. Housed at the Law\u00adrence Ber\u00adke\u00adley Natio\u00adnal Labo\u00adra\u00adto\u00adry, Perlmutter\u2019s increased per\u00adfor\u00admance couldn\u2019t move it from its pre\u00advious\u00adly held No. 5&nbsp;spot.&nbsp;<\/p>\n<p>Fug\u00ada\u00adku con\u00adti\u00adnues to hold the No. 1 posi\u00adti\u00adon that it first ear\u00adned in June 2020. Its <span class=\"caps\">HPL<\/span> bench\u00admark score is 442 Pflop\/s, which excee\u00added the per\u00adfor\u00admance of Sum\u00admit at No. 2 by 3x. Instal\u00adled at the Riken Cen\u00adter for Com\u00adpu\u00adta\u00adtio\u00adnal Sci\u00adence (R\u2011<span class=\"caps\">CCS<\/span>) in Kobe, Japan, it was co-deve\u00adlo\u00adped by Riken and Fuji\u00adtsu and is based on Fujitsu\u2019s cus\u00adtom <span class=\"caps\">ARM<\/span> <span class=\"caps\">A64FX<\/span> pro\u00adces\u00adsor. Fug\u00ada\u00adku also uses Fujitsu\u2019s Tofu D inter\u00adcon\u00adnect to trans\u00adfer data bet\u00adween&nbsp;nodes.&nbsp;<\/p>\n<p>In sin\u00adgle or fur\u00adther-redu\u00adced pre\u00adcis\u00adi\u00adon, which are often used in machi\u00adne lear\u00adning and A.I. appli\u00adca\u00adti\u00adon, Fug\u00ada\u00adku has a peak per\u00adfor\u00admance abo\u00adve 1,000 PFlop\/s (1 Exaflop\/s). As a result, Fug\u00ada\u00adku is often intro\u00addu\u00adced as the first \u201cExas\u00adca\u00adle\u201d supercomputer.<\/p>\n<p>While the\u00adre were also reports about seve\u00adral Chi\u00adne\u00adse sys\u00adtems rea\u00adching Exa\u00adflop level per\u00adfor\u00admance, none of the\u00adse sys\u00adtems sub\u00admit\u00adted an <span class=\"caps\">HPL<\/span> result to the <span class=\"caps\">TOP500<\/span>.&nbsp;<\/p>\n<p>Here\u2019s a sum\u00adma\u00adry of the sys\u00adtems in the&nbsp;Top10:<\/p>\n<ul>\n<li>Fug\u00ada\u00adku remains the No. 1 sys\u00adtem. It has 7,630,848 cores which allo\u00adwed it to achie\u00adve an <span class=\"caps\">HPL<\/span> bench\u00admark score of 442 Pflop\/s. This puts it 3x ahead of the No. 2 sys\u00adtem in the&nbsp;list.&nbsp;<\/li>\n<li>Sum\u00admit, an IBM-built sys\u00adtem at the Oak Ridge Natio\u00adnal Labo\u00adra\u00adto\u00adry (<span class=\"caps\">ORNL<\/span>) in Ten\u00adnes\u00adsee, <span class=\"caps\">USA<\/span>, remains the fas\u00adtest sys\u00adtem in the U.S. and at the No. 2 spot world\u00adwi\u00adde. It has a per\u00adfor\u00admance of 148.8 Pflop\/s on the <span class=\"caps\">HPL<\/span> bench\u00admark, which is used to rank the <span class=\"caps\">TOP500<\/span> list. Sum\u00admit has 4,356 nodes, each housing two Power9 CPUs with 22 cores each and six <span class=\"caps\">NVIDIA<\/span> Tes\u00adla <span class=\"caps\">V100<\/span> GPUs, each with 80 strea\u00adming mul\u00adtipro\u00adces\u00adsors (S.M.). The nodes are lin\u00adked tog\u00ade\u00adther with a Mel\u00adlan\u00adox dual-rail <span class=\"caps\">EDR<\/span> Infi\u00adni\u00adBand network.<\/li>\n<li>Sier\u00adra, a sys\u00adtem at the Law\u00adrence Liver\u00admo\u00adre Natio\u00adnal Labo\u00adra\u00adto\u00adry, <span class=\"caps\">CA<\/span>, <span class=\"caps\">USA<\/span>, is at No. 3. Its archi\u00adtec\u00adtu\u00adre is very simi\u00adlar to the #2 sys\u00adtems Sum\u00admit. It is built with 4,320 nodes with two Power9 CPUs and four <span class=\"caps\">NVIDIA<\/span> Tes\u00adla <span class=\"caps\">V100<\/span> GPUs. Sier\u00adra achie\u00adved 94.6 Pflop\/s.<\/li>\n<li>Sun\u00adway Tai\u00adhu\u00adLight is a sys\u00adtem deve\u00adlo\u00adped by China\u2019s Natio\u00adnal Rese\u00adarch Cen\u00adter of Par\u00adal\u00adlel Com\u00adpu\u00adter Engi\u00adnee\u00adring <span class=\"amp\">&amp;<\/span> Tech\u00adno\u00adlo\u00adgy (<span class=\"caps\">NRCPC<\/span>) and instal\u00adled at the Natio\u00adnal Super\u00adcom\u00adpu\u00adting Cen\u00adter in Wuxi, China\u2019s Jiangsu pro\u00advin\u00adce is lis\u00adted at the No. 4 posi\u00adti\u00adon with 93 Pflop\/s.<\/li>\n<li>Perl\u00admut\u00adter at No. 5 was new\u00adly lis\u00adted in the <span class=\"caps\">TOP10<\/span> in last June. It is based on the <span class=\"caps\">HPE<\/span> Cray \u201cShas\u00adta\u201d plat\u00adform, and a hete\u00adro\u00adge\u00adneous sys\u00adtem with <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> based nodes and 1536 <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">A100<\/span> acce\u00adle\u00adra\u00adted nodes. Perl\u00admut\u00adter impro\u00adved its per\u00adfor\u00admance to 70.9 Pflop\/s<\/li>\n<li>Sele\u00adne, now at No. 6, is an <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">DGX<\/span> <span class=\"caps\">A100<\/span> Super\u00adPOD instal\u00adled in-house at <span class=\"caps\">NVIDIA<\/span> in the <span class=\"caps\">USA<\/span>. &nbsp;The sys\u00adtem is based on an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> pro\u00adces\u00adsor with <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">A100<\/span> for acce\u00adle\u00adra\u00adti\u00adon and a Mel\u00adlan\u00adox <span class=\"caps\">HDR<\/span> Infi\u00adni\u00adBand as a net\u00adwork. It achie\u00adved 63.4 Pflop\/s.<\/li>\n<li>Tian\u00adhe-2A (Mil\u00adky Way-2A), a sys\u00adtem deve\u00adlo\u00adped by China\u2019s Natio\u00adnal Uni\u00adver\u00adsi\u00adty of Defen\u00adse Tech\u00adno\u00adlo\u00adgy (<span class=\"caps\">NUDT<\/span>) and deploy\u00aded at the Natio\u00adnal Super\u00adcom\u00adpu\u00adter Cen\u00adter in Guang\u00adzhou, Chi\u00adna, is now lis\u00adted as the No. 7 sys\u00adtem with 61.4 Pflop\/s.<\/li>\n<li>A sys\u00adtem cal\u00adled \u201c<span class=\"caps\">JUWELS<\/span> Boos\u00adter Modu\u00adle\u201d is No. 8. The Bull\u00adSe\u00adqua\u00adna sys\u00adtem build by Atos is instal\u00adled at the For\u00adschungs\u00adzen\u00adtrum Jue\u00adlich (<span class=\"caps\">FZJ<\/span>) in Ger\u00adma\u00adny. The sys\u00adtem uses an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> pro\u00adces\u00adsor with <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">A100<\/span> for acce\u00adle\u00adra\u00adti\u00adon and a Mel\u00adlan\u00adox <span class=\"caps\">HDR<\/span> Infi\u00adni\u00adBand as a net\u00adwork simi\u00adlar to the Sele\u00adne Sys\u00adtem. This sys\u00adtem is the most powerful sys\u00adtem in Euro\u00adpe, with 44.1 Pflop\/s.&nbsp;<\/li>\n<li><span class=\"caps\">HPC5<\/span> at No. 9 is a PowerEdge sys\u00adtem built by Dell and instal\u00adled by the Ita\u00adli\u00adan com\u00adpa\u00adny Eni S.p.A. It achie\u00adves a per\u00adfor\u00admance of 35.5 Pflop\/s due to using <span class=\"caps\">NVIDIA<\/span> Tes\u00adla <span class=\"caps\">V100<\/span> as acce\u00adle\u00adra\u00adtors and a Mel\u00adlan\u00adox <span class=\"caps\">HDR<\/span> Infi\u00adni\u00adBand as the network.&nbsp;<\/li>\n<li>Voy\u00ada\u00adger-EUS2, a Micro\u00adsoft Azu\u00adre sys\u00adtem instal\u00adled at Micro\u00adsoft in the U.S., is the only new sys\u00adtem in the <span class=\"caps\">TOP10<\/span>. It achie\u00adved 30.05 Pflop\/s and is lis\u00adted at No. 10. This archi\u00adtec\u00adtu\u00adre is based on an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> pro\u00adces\u00adsor with 48 cores and 2.45GHz working tog\u00ade\u00adther with an <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">A100<\/span> <span class=\"caps\">GPU<\/span> with 80 G.B. memo\u00adry and uti\u00adli\u00adzing a Mel\u00adlan\u00adox <span class=\"caps\">HDR<\/span> Infi\u00adni\u00adband for data transfer.&nbsp;<\/li>\n<\/ul>\n<h2>Other <span class=\"caps\">TOP500<\/span> highlights&nbsp;<\/h2>\n<p>While the\u00adre were not many chan\u00adges to the Top10, we did see a smat\u00adte\u00adring of shifts within the Top15. The new Voy\u00ada\u00adger-EUS sys\u00adtem from Micro\u00adsoft fol\u00adlo\u00adwed its sibling into the No. 11 spot, while the <span class=\"caps\">SSC-21<\/span> sys\u00adtem from Sam\u00adsung intro\u00addu\u00adced its\u00adelf to the list at No. 12. Pola\u00adris, also a new sys\u00adtem, came in at No. 13 while the new <span class=\"caps\">CEA-HF<\/span> took No.&nbsp;15.&nbsp;<\/p>\n<p>Like the last list, <span class=\"caps\">AMD<\/span> pro\u00adces\u00adsors are see\u00ading a lot of suc\u00adcess. Fron\u00adte\u00adra, which has a Xeon Pla\u00adti\u00adnum 8280 pro\u00adces\u00adsor, got bum\u00adped by Voy\u00ada\u00adger-EUS2, which has an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> pro\u00adces\u00adsor. What\u2019s more, all of the new Top15 machi\u00adnes descri\u00adbed abo\u00adve have <span class=\"caps\">AMD<\/span> processors&nbsp;<\/p>\n<p>Unsur\u00adpri\u00adsin\u00adgly, sys\u00adtems from Chi\u00adna and the <span class=\"caps\">USA<\/span> domi\u00adna\u00adted the list. Alt\u00adhough Chi\u00adna drop\u00adped from 186 sys\u00adtems to 173, the <span class=\"caps\">USA<\/span> increased from 123 machi\u00adnes to 150. All told, the\u00adse two count\u00adries account for near\u00adly two-thirds of the super\u00adcom\u00adpu\u00adters on the <span class=\"caps\">TOP500<\/span>.&nbsp;<\/p>\n<p>The new edi\u00adti\u00adon of the list didn\u2019t show\u00adca\u00adse much chan\u00adge in terms of sys\u00adtem inter\u00adcon\u00adnects. Ether\u00adnet still domi\u00adna\u00adted at 240 machi\u00adnes, while Infi\u00adni\u00adband accoun\u00adted for 180. &nbsp;Omin\u00adpath inter\u00adcon\u00adnects saw 40 spots on the list, the\u00adre were 34 cus\u00adtom inter\u00adcon\u00adnects, and only 6 sys\u00adtems with pro\u00adprie\u00adta\u00adry networks.&nbsp;<\/p>\n<h2>Green500 results&nbsp;<\/h2>\n<p>The sys\u00adtem to cla\u00adim the No. 1 spot for the Green500 was <span class=\"caps\">MN<\/span>\u20113 from Pre\u00adfer\u00adred Net\u00adworks in Japan. Rely\u00ading on the MN-Core chip and an acce\u00adle\u00adra\u00adtor opti\u00admi\u00adzed for matrix arith\u00adme\u00adtic, this machi\u00adne was able to achie\u00adve an incre\u00addi\u00adble 39.38 gigaflops\/watt power-effi\u00adci\u00aden\u00adcy. This machi\u00adne pro\u00advi\u00added a per\u00adfor\u00admance 29.7- gigaflops\/watt on the last list, cle\u00adar\u00adly show\u00adca\u00adsing some impres\u00adsi\u00adve impro\u00adve\u00adment. It also enhan\u00adced its stan\u00adding on the <span class=\"caps\">TOP500<\/span> list, moving from No. 337 to No.&nbsp;302.&nbsp;<\/p>\n<p>The new <span class=\"caps\">SSC-21<\/span> Sca\u00adlable Modu\u00adle an <span class=\"caps\">HPE<\/span> Apol\u00adlo 6500 sys\u00adtem instal\u00adled at Sam\u00adsung Elec\u00adtro\u00adnics in South Korea achie\u00adved an impres\u00adsi\u00adve 33.98 gigaflops\/watt. They did so by sub\u00admit\u00adting an power opti\u00admi\u00adzed run of the <span class=\"caps\">HPL<\/span> bench\u00admark. It is lis\u00adted at &nbsp;posi\u00adti\u00adon 292&nbsp;in the <span class=\"caps\">TOP500<\/span>.<\/p>\n<p><span class=\"caps\">NVIDIA<\/span> instal\u00adled a new liquid coo\u00adled <span class=\"caps\">DGX<\/span> <span class=\"caps\">A100<\/span> pro\u00adto\u00adty\u00adpe sys\u00adtem cal\u00adled Tethys. With a power opti\u00admi\u00adzed <span class=\"caps\">HPL<\/span> run Tethys achie\u00adved 31.5 gigaflops\/watt and gar\u00adne red the No. 3 spot on the Green500. It is lis\u00adted at posi\u00adti\u00adon 296&nbsp;in the <span class=\"caps\">TOP500<\/span>.<\/p>\n<p>The Wilkes\u20113 sys\u00adtem impro\u00adved its results but was still pushed down to the No.4 spot on the Green500. &nbsp;Wilkes\u20113, which is housed at the Uni\u00adver\u00adsi\u00adty of Cam\u00adbridge in the U.K., had a power-effi\u00adci\u00aden\u00adcy of 30.8 gigaflops\/watt. Howe\u00adver, it was pushed from No. 100 to No. 281 on the <span class=\"caps\">TOP500<\/span> list.<\/p>\n<p>The Uni\u00adver\u00adsi\u00adty of Flo\u00adri\u00adda in the <span class=\"caps\">USA<\/span> with its HiP\u00ader\u00adGa\u00adtor <span class=\"caps\">AI<\/span> sys\u00adtem was pushed from the No. 2 spot to the No. 5 spot. This machi\u00adne held ste\u00adady at 29.52 gigaflops\/watt. This <span class=\"caps\">NVIDIA<\/span> sys\u00adtem has 138,880 cores and reli\u00ades on an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> 7742 pro\u00adces\u00adsor. Despi\u00adte this impres\u00adsi\u00adve per\u00adfor\u00admance, HiP\u00ader\u00adGa\u00adtor <span class=\"caps\">AI<\/span> was pushed from No. 22 to No. 31 on the <span class=\"caps\">TOP500<\/span><\/p>\n<h2><span class=\"caps\">HPCG<\/span> Results&nbsp;<\/h2>\n<p>The <span class=\"caps\">TOP500<\/span> list has incor\u00adpo\u00adra\u00adted the High-Per\u00adfor\u00admance Con\u00adju\u00adga\u00adte Gra\u00addi\u00adent (<span class=\"caps\">HPCG<\/span>) Bench\u00admark results, which pro\u00advi\u00adde an alter\u00adna\u00adti\u00adve metric for asses\u00adsing super\u00adcom\u00adpu\u00adter per\u00adfor\u00admance and is meant to com\u00adple\u00adment the <span class=\"caps\">HPL<\/span> measurement.<\/p>\n<p>The <span class=\"caps\">HPCG<\/span> results here are very simi\u00adlar to the last list. Fug\u00ada\u00adku was the clear win\u00adner with 16.0 HPCG-peta\u00adflops, while Sum\u00admit retai\u00adned its No. 2 spot with 2.93 HPCG-peta\u00adflops. Perl\u00admut\u00adter, a <span class=\"caps\">USA<\/span> machi\u00adne housed at Law\u00adrence Ber\u00adke\u00adley Natio\u00adnal Labo\u00adra\u00adto\u00adry, took the No. 3 spot with 1.91 HPCG-petaflops.&nbsp;<\/p>\n<h2><span class=\"caps\">HPL-AI<\/span> Results&nbsp;<\/h2>\n<p>The <span class=\"caps\">HPL-AI<\/span> bench\u00admark seeks to high\u00adlight the con\u00adver\u00adgence of <span class=\"caps\">HPC<\/span> and arti\u00adfi\u00adci\u00adal intel\u00adli\u00adgence (<span class=\"caps\">AI<\/span>) workloads based on machi\u00adne lear\u00adning and deep lear\u00adning by sol\u00adving a sys\u00adtem of line\u00adar equa\u00adtions using novel, mixed-pre\u00adcis\u00adi\u00adon algo\u00adrith\u00adms that exploit modern hardware.<\/p>\n<p>Achie\u00adving an <span class=\"caps\">HPL-AI<\/span> bench\u00admark of 2 Exa\u00adflops, Fug\u00ada\u00adku is lea\u00adding the pack in this regard. With such excel\u00adlent metrics year-after-year, com\u00adbi\u00adned with a con\u00adside\u00adra\u00adti\u00adon by many as the first \u201cExas\u00adca\u00adle\u201d super\u00adcom\u00adpu\u00adter, Fug\u00ada\u00adku is cle\u00adar\u00adly an exci\u00adting system.&nbsp;<\/p>\n","protected":false},"excerpt":{"rendered":"<p><span class=\"caps\">FRANKFURT<\/span>, Ger\u00adma\u00adny; <span class=\"caps\">BERKELEY<\/span>, Calif.; and <span class=\"caps\">KNOXVILLE<\/span>, Tenn.\u2014 The 58th annu\u00adal edi\u00adti\u00adon of the <span class=\"caps\">TOP500<\/span> saw litt\u00adle chan\u00adge in the Top10. The Micro\u00adsoft Azu\u00adre sys\u00adtem cal\u00adled Voy\u00ada\u00adger-EUS2 was the only machi\u00adne to shake up the top spots, clai\u00adming No. 10. Based on an <span class=\"caps\">AMD<\/span> <span class=\"caps\">EPYC<\/span> pro\u00adces\u00adsor with 48 cores and 2.45GHz working tog\u00ade\u00adther with an <span class=\"caps\">NVIDIA<\/span> <span class=\"caps\">A100<\/span> <span class=\"caps\">GPU<\/span> and 80 <span class=\"caps\">GB<\/span> of memo\u00adry, Voy\u00ada\u00adger-EUS2 also uti\u00adli\u00adzes a Mel\u00adlan\u00adox <span class=\"caps\">HDR<\/span> Infi\u00adni\u00adband for data transfer.&nbsp;<\/p>\n<p>While the\u00adre were no other chan\u00adges to the posi\u00adti\u00adons of the sys\u00adtems in the Top10, Perl\u00admut\u00adter at <span class=\"caps\">NERSC<\/span> impro\u00adved its per\u00adfor\u00admance to 70.9 Pflop\/s. Housed at the Law\u00adrence Ber\u00adke\u00adley Natio\u00adnal Labo\u00adra\u00adto\u00adry, Perlmutter\u2019s increased per\u00adfor\u00admance couldn\u2019t move it from its pre\u00advious\u00adly held No. 5&nbsp;spot.&nbsp;<\/p>\n<p>Fug\u00ada\u00adku con\u00adti\u00adnues to hold the No. 1 posi\u00adti\u00adon that it first ear\u00adned in June 2020. Its <span class=\"caps\">HPL<\/span> bench\u00admark score is 442 Pflop\/s, which excee\u00added the per\u00adfor\u00admance of Sum\u00admit at No. 2 by 3x. Instal\u00adled at the Riken Cen\u00adter for Com\u00adpu\u00adta\u00adtio\u00adnal Sci\u00adence (R\u2011<span class=\"caps\">CCS<\/span>) in Kobe, Japan, it was co-deve\u00adlo\u00adped by Riken and Fuji\u00adtsu and is based on Fujitsu\u2019s cus\u00adtom <span class=\"caps\">ARM<\/span> <span class=\"caps\">A64FX<\/span> pro\u00adces\u00adsor. Fug\u00ada\u00adku also uses Fujitsu\u2019s Tofu D inter\u00adcon\u00adnect to trans\u00adfer data bet\u00adween&nbsp;nodes.&nbsp;<\/p>\n<p>In sin\u00adgle or fur\u00adther-redu\u00adced pre\u00adcis\u00adi\u00adon, which are often used in machi\u00adne lear\u00adning and A.I. appli\u00adca\u00adti\u00adon, Fug\u00ada\u00adku has a peak per\u00adfor\u00admance abo\u00adve 1,000 PFlop\/s (1 Exaflop\/s). As a result, Fug\u00ada\u00adku is often intro\u00addu\u00adced as the first \u201cExas\u00adca\u00adle\u201d super\u00adcom\u00adpu\u00adter. (\u2026) <a class=\"moretag\" href=\"https:\/\/www.planet3dnow.de\/cms\/64035-still-waiting-for-exascale-japans-fugaku-outperforms-all-competition-once-again\/\">Wei\u00adter\u00adle\u00adsen&nbsp;\u00bb<\/a><\/p>\n","protected":false},"author":593,"featured_media":0,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"wp_typography_post_enhancements_disabled":false,"ngg_post_thumbnail":0,"footnotes":""},"categories":[22],"tags":[2659,2530,759],"class_list":["post-64035","post","type-post","status-publish","format-standard","hentry","category-pressemitteilungen","tag-58th","tag-fugaku","tag-top500","entry"],"share_on_mastodon":{"url":"","error":""},"_links":{"self":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/64035","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/users\/593"}],"replies":[{"embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/comments?post=64035"}],"version-history":[{"count":1,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/64035\/revisions"}],"predecessor-version":[{"id":64036,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/64035\/revisions\/64036"}],"wp:attachment":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/media?parent=64035"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/categories?post=64035"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/tags?post=64035"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}