{"id":63399,"date":"2021-08-30T15:14:21","date_gmt":"2021-08-30T13:14:21","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=63399"},"modified":"2021-08-30T15:14:21","modified_gmt":"2021-08-30T13:14:21","slug":"amd-epyc-processors-picked-by-argonne-national-laboratory-to-prepare-for-exascale-future","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/63399-amd-epyc-processors-picked-by-argonne-national-laboratory-to-prepare-for-exascale-future\/","title":{"rendered":"AMD<\/span> EPYC<\/span>\u2122 Processors Picked by Argonne National Laboratory to Prepare for Exascale Future"},"content":{"rendered":"

- The new Pola\u00adris super\u00adcom\u00adpu\u00adter will opti\u00admi\u00adze AI<\/span>, engi\u00adnee\u00adring, and sci\u00aden\u00adti\u00adfic pro\u00adjects for forth\u00adco\u00adming exas\u00adca\u00adle super\u00adcom\u00adpu\u00adter at Argon\u00adne Natio\u00adnal Laboratory -<\/p>\n

SANTA<\/span> CLARA<\/span>, Calif., Aug. 30, 2021 (GLOBE<\/span> NEWSWIRE<\/span>) \u2014 AMD<\/span><\/a> (NASDAQ<\/span>: AMD<\/span>) announ\u00adced that the U.S. Depart\u00adment of Energy\u2019s (DOE<\/span>) Argon\u00adne Natio\u00adnal Labo\u00adra\u00adto\u00adry (Argon\u00adne)<\/a> has cho\u00adsen AMD<\/span> EPYC<\/span>\u2122 pro\u00adces\u00adsors to power a new super\u00adcom\u00adpu\u00adter, cal\u00adled Pola\u00adris, which will prepa\u00adre rese\u00adar\u00adchers for the forth\u00adco\u00adming exas\u00adca\u00adle super\u00adcom\u00adpu\u00adter at Argon\u00adne cal\u00adled Auro\u00adra. Pola\u00adris is built by Hew\u00adlett Packard Enter\u00adpri\u00adse (HPE<\/span>), will use 2<\/a>nd<\/sup><\/a> Gen EPYC<\/span> pro\u00adces\u00adsors<\/a> and then upgrade to 3<\/a>rd<\/sup><\/a> Gen AMD<\/span> EPYC<\/span> pro\u00adces\u00adsors<\/a>, and will allow sci\u00aden\u00adtists and deve\u00adlo\u00adpers to test and opti\u00admi\u00adze soft\u00adware codes and appli\u00adca\u00adti\u00adons to tack\u00adle a ran\u00adge of AI<\/span>, engi\u00adnee\u00adring, and sci\u00aden\u00adti\u00adfic projects.<\/p>\n

\u201c<\/span>AMD<\/span> EPYC<\/span> ser\u00adver pro\u00adces\u00adsors con\u00adti\u00adnue to be the lea\u00adding choice for modern HPC<\/span> rese\u00adarch, deli\u00adve\u00adring the per\u00adfor\u00admance and capa\u00adbi\u00adli\u00adties nee\u00added to help sol\u00adve the com\u00adplex pro\u00adblems that pre-exas\u00adca\u00adle and exas\u00adca\u00adle com\u00adpu\u00adting will address,\u201d said For\u00adrest Nor\u00adrod, seni\u00ador vice pre\u00adsi\u00addent and gene\u00adral mana\u00adger, Dat\u00ada\u00adcen\u00adter and Embedded Solu\u00adti\u00adons Busi\u00adness Group, AMD<\/span>. \u201cWe are extre\u00adme\u00adly proud to sup\u00adport Argon\u00adne Natio\u00adnal Labo\u00adra\u00adto\u00adry and their cri\u00adti\u00adcal rese\u00adarch into are\u00adas inclu\u00adding low car\u00adbon tech\u00adno\u00adlo\u00adgies, medi\u00adcal rese\u00adarch, astro\u00adno\u00admy, solar power and more as we draw clo\u00adser to the exas\u00adca\u00adle era.\u201d<\/p>\n

\n
 <\/div>\n<\/div>\n

Pola\u00adris will use the AMD<\/span> EPYC<\/span> 7532 and EPYC<\/span> 7543 pro\u00adces\u00adsors, and NVIDIA<\/span>\u00ae A100<\/span> Ten\u00adsor Core GPUs, to deli\u00adver appro\u00adxi\u00adm\u00adate\u00adly 44 peta\u00adflops of peak dou\u00adble pre\u00adcis\u00adi\u00adon per\u00adfor\u00admance, which is 4x fas\u00adter than Argonne\u2019s cur\u00adrent supercomputers.<\/p>\n

Initi\u00adal\u00adly, Pola\u00adris will be used by rese\u00adarch teams par\u00adti\u00adci\u00adpa\u00adting in initia\u00adti\u00adves such as the DOE<\/span>\u2019s Exas\u00adca\u00adle Com\u00adpu\u00adting Pro\u00adject and the ALCF<\/span>\u2019s Ear\u00adly Sci\u00adence Pro\u00adgram<\/a>. User com\u00admu\u00adni\u00adties within the DOE<\/span>\u2019s Exas\u00adca\u00adle Com\u00adpu\u00adting Pro\u00adject will also use Pola\u00adris for opti\u00admi\u00adzing engi\u00adnee\u00adring tasks for Argonne\u2019s forth\u00adco\u00adming exas\u00adca\u00adle super\u00adcom\u00adpu\u00adter, which includes sca\u00adling of com\u00adbi\u00adned CPU<\/span> and GPU-enab\u00adled sys\u00adtems and the inte\u00adgra\u00adti\u00adon of work\u00adflows com\u00adbi\u00adning mode\u00adling, simu\u00adla\u00adti\u00adon, AI<\/span> and other data-inten\u00adsi\u00adve components.<\/p>\n

Pola\u00adris is sche\u00addu\u00adled to be deli\u00adver\u00aded and instal\u00adled in August 2021 and will go into use start\u00ading ear\u00adly 2022. The broa\u00adder HPC<\/span> com\u00admu\u00adni\u00adty will access the sys\u00adtem in spring of 2022 to prepa\u00adre workloads for the next gene\u00adra\u00adti\u00adon of DOE<\/span>\u2019s HPC<\/span> resources.<\/p>\n

Sup\u00adport\u00ading Resources<\/strong><\/p>\n