{"id":68055,"date":"2023-08-20T12:09:26","date_gmt":"2023-08-20T10:09:26","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=68055"},"modified":"2023-08-20T12:14:18","modified_gmt":"2023-08-20T10:14:18","slug":"amd-showcases-leadership-cloud-performance-with-new-amazon-ec2-instances-powered-by-4th-gen-amd-epyc-processors","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/68055-amd-showcases-leadership-cloud-performance-with-new-amazon-ec2-instances-powered-by-4th-gen-amd-epyc-processors\/","title":{"rendered":"AMD<\/span> Showcases Leadership Cloud Performance with New Amazon EC2<\/span> Instances Powered by 4th Gen AMD<\/span> EPYC<\/span> Processors"},"content":{"rendered":"

\u2014Ama\u00adzon EC2<\/span> M7a ins\u00adtances offer up to 50 per\u00adcent more com\u00adpu\u00adte per\u00adfor\u00admance than M6a ins\u00adtances and bring an even broa\u00adder ran\u00adge of workloads to AWS<\/span>\u2014<\/p>\n

\u2014Ama\u00adzon EC2<\/span> Hpc7a ins\u00adtances, powered by 4th Gen AMD<\/span> EPYC<\/span> pro\u00adces\u00adsors, deli\u00adver up to 2.5x bet\u00adter per\u00adfor\u00admance com\u00adpared to Ama\u00adzon EC2<\/span> Hpc6a instances\u2014<\/p>\n

SANTA<\/span> CLARA<\/span>, Calif., Aug. 18, 2023 (GLOBE<\/span> NEWSWIRE<\/span>) \u2014 Today, AMD<\/span><\/a> (NASDAQ<\/span>: AMD<\/span>) announ\u00adced Ama\u00adzon Web Ser\u00advices (AWS<\/span>) has expan\u00added its 4th<\/sup> Gen AMD<\/span> EPYC<\/span>\u2122<\/em> pro\u00adces\u00adsor-based offe\u00adrings with the gene\u00adral avai\u00adla\u00adbi\u00adli\u00adty of Ama\u00adzon Ela\u00ads\u00adtic Com\u00adpu\u00adte Cloud (EC2<\/span>) M7a<\/a> and Ama\u00adzon EC2<\/span> Hpc7a ins\u00adtances<\/a>, which offer next-gene\u00adra\u00adti\u00adon per\u00adfor\u00admance and effi\u00adci\u00aden\u00adcy for appli\u00adca\u00adti\u00adons that bene\u00adfit from high per\u00adfor\u00admance, high through\u00adput and tight\u00adly cou\u00adpled HPC<\/span> workloads, respectively.<\/p>\n

\u201c<\/span>For cus\u00adto\u00admers with incre\u00adasing\u00adly com\u00adplex and com\u00adpu\u00adte-inten\u00adsi\u00adve workloads, 4th<\/sup> Gen EPYC<\/span> pro\u00adces\u00adsor-powered Ama\u00adzon EC2<\/span> ins\u00adtances deli\u00adver a dif\u00adfe\u00adren\u00adtia\u00adted offe\u00adring for cus\u00adto\u00admers,\u201d said David Brown, vice pre\u00adsi\u00addent of Ama\u00adzon EC2<\/span> at AWS<\/span>. \u201cCom\u00adbi\u00adned with the power of the AWS<\/span> Nit\u00adro Sys\u00adtem, both M7a and Hpc7a ins\u00adtances allow for fast and low-laten\u00adcy inter\u00adnode com\u00admu\u00adni\u00adca\u00adti\u00adons, advan\u00adcing what our cus\u00adto\u00admers can achie\u00adve across our gro\u00adwing fami\u00adly of Ama\u00adzon EC2<\/span> instances.\u201d<\/p>\n

\u201c<\/span>4th<\/sup> Gen EPYC<\/span> pro\u00adces\u00adsors have unmat\u00adched per\u00adfor\u00admance lea\u00adder\u00adship; and our col\u00adla\u00adbo\u00adra\u00adti\u00adon with AWS<\/span> is deli\u00adve\u00adring the full 4th<\/sup> Gen EPYC<\/span> per\u00adfor\u00admance per core in new M7a and Hpc7a ins\u00adtances,\u201d said Dan McNa\u00adma\u00adra, seni\u00ador vice pre\u00adsi\u00addent and gene\u00adral mana\u00adger, Ser\u00adver Busi\u00adness Unit, AMD<\/span>. \u201cAs a result, our cus\u00adto\u00admers can do more with less across a wide ran\u00adge of workloads, while simul\u00adta\u00adneous\u00adly opti\u00admi\u00adzing their businesses.\u201d<\/p>\n

AMD<\/span> and AWS<\/span> Expand EC2<\/span> for Grea\u00adter Per\u00adfor\u00admance and New Applications
\n<\/strong>Ama\u00adzon EC2<\/span> M7a ins\u00adtances, first pre\u00adview\u00aded<\/a> at this year\u2019s AMD<\/span> Data Cen\u00adter and AI<\/span> Tech\u00adno\u00adlo\u00adgy Pre\u00admie\u00adre<\/a>, cater to a wide ran\u00adge of workloads, pro\u00advi\u00adding cus\u00adto\u00admers with up to 50 per\u00adcent more com\u00adpu\u00adte per\u00adfor\u00admance than Ama\u00adzon EC2<\/span> M6a ins\u00adtances, while offe\u00adring new pro\u00adces\u00adsor capa\u00adbi\u00adli\u00adties, such as AVX3-512<\/span>, VNNI<\/span>, and BFloat16. In addi\u00adti\u00adon to high per\u00adfor\u00admance in gene\u00adral pur\u00adpo\u00adse workloads, the Ama\u00adzon EC2<\/span> M7a ins\u00adtances are ide\u00adal for appli\u00adca\u00adti\u00adons that bene\u00adfit from high per\u00adfor\u00admance and high through\u00adput such as finan\u00adcial appli\u00adca\u00adti\u00adons, simu\u00adla\u00adti\u00adon mode\u00adling, gam\u00ading, mid-sized data stores and more.<\/p>\n

Ama\u00adzon EC2<\/span> Hpc7a ins\u00adtances are desi\u00adgned for tight\u00adly cou\u00adpled high per\u00adfor\u00admance com\u00adpu\u00adting workloads and deli\u00adver 2.5x bet\u00adter per\u00adfor\u00admance com\u00adpared to Ama\u00adzon EC2<\/span> Hpc6a ins\u00adtances. The\u00adse ins\u00adtances meet the demands of incre\u00adasing workload com\u00adple\u00adxi\u00adty by offe\u00adring AWS<\/span> cus\u00adto\u00admers more com\u00adpu\u00adte, memo\u00adry and net\u00adwork per\u00adfor\u00admance. Powered by 300 Gb\/s EFA<\/span> net\u00adwork band\u00adwidth, the Hpc7a ins\u00adtances are ide\u00adal\u00adly sui\u00adted for com\u00adpu\u00adte-inten\u00adsi\u00adve, laten\u00adcy-sen\u00adsi\u00adti\u00adve workloads such as com\u00adpu\u00adta\u00adtio\u00adnal flu\u00adid dyna\u00admics, wea\u00adther fore\u00adcas\u00adting, mole\u00adcu\u00adlar dyna\u00admics, and com\u00adpu\u00adta\u00adtio\u00adnal che\u00admis\u00adtry. Ama\u00adzon EC2<\/span> Hpc7a ins\u00adtances also bring next-gene\u00adra\u00adti\u00adon ser\u00adver tech\u00adno\u00adlo\u00adgy like DDR5<\/span> memo\u00adry, which pro\u00advi\u00addes 50 per\u00adcent hig\u00adher memo\u00adry band\u00adwidth com\u00adpared to DDR4<\/span> memo\u00adry to enable high-speed access to data in memory.<\/p>\n

\u201c<\/span>To deli\u00adver high-level ope\u00adra\u00adtio\u00adnal intel\u00adli\u00adgence for wea\u00adther-depen\u00addent indus\u00adtries, DTN<\/span> deploys a suite of wea\u00adther data and models that deli\u00adver sophisti\u00adca\u00adted, high-reso\u00adlu\u00adti\u00adon out\u00adputs and requi\u00adre con\u00adti\u00adnu\u00adal pro\u00adces\u00adsing of vast amounts of data from inputs across the glo\u00adbe,\u201d said Lars Ewe, Chief Pro\u00adduct and Tech\u00adno\u00adlo\u00adgy Offi\u00adcer, DTN<\/span>. \u201cOur col\u00adla\u00adbo\u00adra\u00adti\u00adon with AWS<\/span> allows us to bet\u00adter ser\u00adve our cus\u00adto\u00admers with the most up-to-date wea\u00adther intel\u00adli\u00adgence that feed tho\u00adse ana\u00adly\u00adtic engi\u00adnes. We are exci\u00adted to see how the next gene\u00adra\u00adti\u00adon of Ama\u00adzon EC2<\/span> Hpc7a ins\u00adtances can poten\u00adti\u00adal\u00adly sup\u00adport our mis\u00adsi\u00adon to pro\u00advi\u00adde cus\u00adto\u00admers with the insights they need at the moment they are needed.\u201d<\/p>\n

AMD<\/span> began working with AWS<\/span> in 2018 and now pro\u00advi\u00addes more than 100 EPYC<\/span> pro\u00adces\u00adsor-based ins\u00adtances for gene\u00adral pur\u00adpo\u00adse, com\u00adpu\u00adte opti\u00admi\u00adzed, memo\u00adry opti\u00admi\u00adzed and high per\u00adfor\u00admance com\u00adpu\u00adting workloads. Cus\u00adto\u00admers such as DNT<\/span>, Sprinklr, and True\u00adCar have all bene\u00adfit\u00adted from signi\u00adfi\u00adcant cost and cloud uti\u00adliza\u00adti\u00adon opti\u00admiza\u00adti\u00adon with AMD<\/span> based Ama\u00adzon EC2<\/span> ins\u00adtances. With the\u00adse new ins\u00adtances, AWS<\/span> cus\u00adto\u00admers can con\u00adti\u00adnue to take advan\u00adta\u00adge of the excel\u00adlent per\u00adfor\u00admance, sca\u00adla\u00adbi\u00adli\u00adty, and effi\u00adci\u00aden\u00adcy offe\u00adred by AMD<\/span> EPYC<\/span> processors.<\/p>\n

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