{"id":68207,"date":"2024-11-18T20:13:49","date_gmt":"2024-11-18T19:13:49","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=68207"},"modified":"2024-11-18T20:15:25","modified_gmt":"2024-11-18T19:15:25","slug":"amd-accelerates-exascale-computing-to-new-heights-powering-the-fastest-supercomputer-ever-el-capitan","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/68207-amd-accelerates-exascale-computing-to-new-heights-powering-the-fastest-supercomputer-ever-el-capitan\/","title":{"rendered":"AMD<\/span> Accelerates Exascale Computing to New Heights Powering the Fastest Supercomputer Ever, El Capitan"},"content":{"rendered":"

\u2500 El Capi\u00adtan, powered by the AMD<\/span> Instinct MI300A<\/span> APU<\/span>, beco\u00admes the second AMD<\/span> super\u00adcom\u00adpu\u00adter to sur\u00adpass the Exas\u00adca\u00adle bar\u00adri\u00ader, pla\u00adcing #1 on the Top500 list \u2014<\/em><\/p>\n

\u2500 AMD<\/span> con\u00adti\u00adnues set\u00adting the stan\u00addard for HPC<\/span>, powe\u00adring 50 per\u00adcent of the top ten fas\u00adtest and 40 per\u00adcent of the ten most ener\u00adgy effi\u00adci\u00adent super\u00adcom\u00adpu\u00adters in the world\u2014<\/em><\/p>\n

\u2500 <\/em>IBM<\/span> and AMD<\/span> announ\u00adce col\u00adla\u00adbo\u00adra\u00adti\u00adon to deploy AMD<\/span> Instinct MI300X<\/span> acce\u00adle\u00adra\u00adtors as a ser\u00advice on IBM<\/span> Cloud \u2500<\/em><\/p>\n

 <\/p>\n

SANTA<\/span> CLARA<\/span>, Calif., Nov. 18, 2024 (GLOBE<\/span> NEWSWIRE<\/span>) \u2014 Today, AMD<\/span><\/a> (NASDAQ<\/span>: AMD<\/span>) show\u00adca\u00adsed its ongo\u00ading high per\u00adfor\u00admance com\u00adpu\u00adting (HPC<\/span>) lea\u00adder\u00adship at Super\u00adcom\u00adpu\u00adting 2024 by powe\u00adring the world\u2019s fas\u00adtest super\u00adcom\u00adpu\u00adter for the sixth straight Top500<\/a> list.<\/p>\n

The El Capi\u00adtan super\u00adcom\u00adpu\u00adter, housed at Law\u00adrence Liver\u00admo\u00adre Natio\u00adnal Labo\u00adra\u00adto\u00adry (LLNL<\/span>), powered by AMD<\/span> Instinct\u2122 MI300A<\/span> APUs<\/a> and built by Hew\u00adlett Packard Enter\u00adpri\u00adse (HPE<\/span>), is now the fas\u00adtest super\u00adcom\u00adpu\u00adter in the world with a High-Per\u00adfor\u00admance Lin\u00adpack (HPL<\/span>) score of 1.742 exa\u00adflops based on the latest Top500 list<\/a>. Both El Capi\u00adtan and the Fron\u00adtier sys\u00adtem at Oak Ridge Natio\u00adnal Lab clai\u00admed num\u00adbers 18 and 22, respec\u00adtively, on the Green500 list, show\u00adca\u00adsing the impres\u00adsi\u00adve capa\u00adbi\u00adli\u00adties of the AMD<\/span> EPYC<\/span> pro\u00adces\u00adsors and AMD<\/span> Instinct GPUs to dri\u00adve lea\u00adder\u00adship per\u00adfor\u00admance and ener\u00adgy effi\u00adci\u00aden\u00adcy for HPC<\/span> workloads.<\/p>\n

\u201c<\/span>We are thril\u00adled to see El Capi\u00adtan beco\u00adme the second AMD<\/span> powered super\u00adcom\u00adpu\u00adter to break the exa\u00adflop bar\u00adri\u00ader and beco\u00adme the fas\u00adtest super\u00adcom\u00adpu\u00adter in the world. Show\u00adca\u00adsing the incre\u00addi\u00adble per\u00adfor\u00admance and effi\u00adci\u00aden\u00adcy of the AMD<\/span> Instinct MI300<\/span> APUs, this ground\u00adbrea\u00adking machi\u00adne is a tes\u00adta\u00adment to the dedi\u00adca\u00adted work bet\u00adween AMD<\/span>, LLNL<\/span> and HPE<\/span>,\u201d said For\u00adrest Nor\u00adrod, exe\u00adcu\u00adti\u00adve vice pre\u00adsi\u00addent and gene\u00adral mana\u00adger, AMD<\/span>. \u201cAt AMD<\/span>, we are dri\u00adving the future of com\u00adpu\u00adting with lea\u00adder\u00adship per\u00adfor\u00admance and capa\u00adbi\u00adli\u00adties that will con\u00adti\u00adnue to defi\u00adne the con\u00adver\u00adgence of HPC<\/span> and AI<\/span> for years to come.\u201d<\/p>\n

\u201c<\/span>El Capi\u00adtan is cru\u00adcial to the Natio\u00adnal Nuclear Secu\u00adri\u00adty Administration\u2019s core mis\u00adsi\u00adon and signi\u00adfi\u00adcant\u00adly bols\u00adters our abili\u00adty to per\u00adform lar\u00adge ensem\u00adbles of high-fide\u00adli\u00adty 3D<\/span> simu\u00adla\u00adti\u00adons that address the intri\u00adca\u00adte sci\u00aden\u00adti\u00adfic chal\u00adlenges facing the mis\u00adsi\u00adon,\u201d said Rob Nee\u00adly, direc\u00adtor of LLNL<\/span>\u2019s Advan\u00adced Simu\u00adla\u00adti\u00adon and Com\u00adpu\u00adting program.<\/p>\n

Bro\u00adnis R. de Supin\u00adski, LLNL<\/span>\u2019s chief tech\u00adno\u00adlo\u00adgy offi\u00adcer for Liver\u00admo\u00adre Com\u00adpu\u00adting adds, \u201cLever\u00adaging the AMD<\/span> Instinct MI300A<\/span> APUs, we\u2019\u00adve built a sys\u00adtem that was once uni\u00adma\u00adgi\u00adnable, pushing the abso\u00adlu\u00adte boun\u00adda\u00adries of com\u00adpu\u00adta\u00adtio\u00adnal per\u00adfor\u00admance while main\u00adtai\u00adning excep\u00adtio\u00adnal ener\u00adgy effi\u00adci\u00aden\u00adcy. With AI<\/span> beco\u00adming incre\u00adasing\u00adly pre\u00adva\u00adlent in our field, El Capi\u00adtan allows us to inte\u00adgra\u00adte AI<\/span> with our tra\u00addi\u00adtio\u00adnal simu\u00adla\u00adti\u00adon and mode\u00adling workloads, ope\u00adning new ave\u00adnues for dis\u00adco\u00advery across various sci\u00aden\u00adti\u00adfic disciplines.\u201d<\/p>\n

AMD<\/span> \u2013 Powe\u00adring HPC<\/span> and AI<\/span><\/strong>
\nAMD<\/span> com\u00adpu\u00adte engi\u00adnes con\u00adti\u00adnue to power the most important super\u00adcom\u00adpu\u00adters and deli\u00adver excep\u00adtio\u00adnal per\u00adfor\u00admance for tech\u00adni\u00adcal com\u00adpu\u00adting for enter\u00adpri\u00adses and natio\u00adnal labo\u00adra\u00adto\u00adries around the world.<\/p>\n

The latest AMD<\/span> EPYC<\/span>\u2122 9005 Series pro\u00adces\u00adsors<\/a> are the best ser\u00adver CPUs for enter\u00adpri\u00adse, AI<\/span> and cloudi<\/sup>, pro\u00advi\u00adding up to 37 per\u00adcent bet\u00adter gene\u00adra\u00adtio\u00adnal IPC<\/span> per\u00adfor\u00admance for HPC<\/span> and AI<\/span> workloadsii<\/sup>. The\u00adse pro\u00adces\u00adsors also pro\u00advi\u00adde up to 3.9X<\/span> fas\u00adter time to insights com\u00adpared to the com\u00adpe\u00adti\u00adti\u00adon for sci\u00adence and HPC<\/span> appli\u00adca\u00adti\u00adons that sol\u00adve the world\u2019s most chal\u00adlen\u00adging pro\u00adblemsiii<\/sup>.<\/p>\n

AMD<\/span> Instinct acce\u00adle\u00adra\u00adtors<\/a> pro\u00advi\u00adde lea\u00adder\u00adship per\u00adfor\u00admance for the data cen\u00adter, at any scale\u2014from AI<\/span> solu\u00adti\u00adons to Exas\u00adca\u00adle-class super\u00adcom\u00adpu\u00adters. The AMD<\/span> Instinct MI300X<\/span> and MI325X<\/span> acce\u00adle\u00adra\u00adtors enable lea\u00adder\u00adship AI<\/span> per\u00adfor\u00admance and memo\u00adry capa\u00adbi\u00adli\u00adties, while the AMD<\/span> Instinct MI300A<\/span> APU<\/span> puts CPU<\/span> and GPU<\/span> cores, and sta\u00adcked memo\u00adry tog\u00ade\u00adther into a sin\u00adgle packa\u00adge, enab\u00adling new levels of effi\u00adci\u00aden\u00adcy and per\u00adfor\u00admance for HPC<\/span> and AI<\/span> workloads.<\/p>\n

Addi\u00adtio\u00adnal\u00adly, AMD<\/span> EPYC<\/span> Pro\u00adces\u00adsors and AMD<\/span> Instinct acce\u00adle\u00adra\u00adtors are being used to power many new super\u00adcom\u00adpu\u00adting and AI<\/span> pro\u00adjects and deploy\u00adments including:<\/p>\n