{"id":66018,"date":"2022-09-12T17:10:42","date_gmt":"2022-09-12T15:10:42","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=66018"},"modified":"2022-09-12T17:10:42","modified_gmt":"2022-09-12T15:10:42","slug":"amd-joins-new-pytorch-foundation-as-founding-member-to-promote-broader-adoption-of-artificial-intelligence-and-machine-learning-capabilities","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/66018-amd-joins-new-pytorch-foundation-as-founding-member-to-promote-broader-adoption-of-artificial-intelligence-and-machine-learning-capabilities\/","title":{"rendered":"AMD<\/span> Joins New PyTorch Foundation as Founding Member To Promote Broader Adoption of Artificial Intelligence and Machine Learning Capabilities"},"content":{"rendered":"

AMD<\/span> will provide expertise with diverse compute engine capabilities and open software, to help drive leadership performance for accelerated workloads<\/h3>\n

SANTA<\/span> CLARA<\/span>, Calif., Sept. 12, 2022 (GLOBE<\/span> NEWSWIRE<\/span>) \u2014 AMD<\/span><\/a> (NASDAQ<\/span>: AMD<\/span>) today announ\u00adced it is joi\u00adning the new\u00adly crea\u00adted PyTorch Foun\u00adda\u00adti\u00adon as a foun\u00adding mem\u00adber. The foun\u00adda\u00adti\u00adon, which will be part of the non-pro\u00adfit Linux Foun\u00adda\u00adti\u00adon<\/a>, will dri\u00adve adop\u00adti\u00adon of Arti\u00adfi\u00adci\u00adal Intel\u00adli\u00adgence (AI<\/span>) too\u00adling by fos\u00adte\u00adring and sus\u00adtai\u00adning an eco\u00adsys\u00adtem of open source pro\u00adjects with PyTorch, the Machi\u00adne Lear\u00adning (ML<\/span>) soft\u00adware frame\u00adwork ori\u00adgi\u00adnal\u00adly crea\u00adted and fos\u00adte\u00adred by Meta.<\/p>\n

As a foun\u00adding mem\u00adber, AMD<\/span> joins others in the indus\u00adtry to prio\u00adri\u00adti\u00adze the con\u00adtin\u00adued growth of PyTorch\u2019s vibrant com\u00admu\u00adni\u00adty. Sup\u00adport\u00aded by inno\u00adva\u00adtions such as the AMD<\/span> ROCm\u2122 open soft\u00adware plat\u00adform<\/a>, AMD<\/span> Instinct\u2122 acce\u00adle\u00adra\u00adtors, Adap\u00adti\u00adve SoCs and CPUs, AMD<\/span> will help the PyTorch Foun\u00adda\u00adti\u00adon by working to demo\u00adcra\u00adti\u00adze sta\u00adte-of-the-art tools, libra\u00adri\u00ades and other com\u00adpon\u00adents to make the\u00adse ML<\/span> inno\u00adva\u00adtions acces\u00adsi\u00adble to everyone.<\/p>\n

\u201c<\/span>Open soft\u00adware is cri\u00adti\u00adcal to advan\u00adcing HPC<\/span>, AI<\/span> and ML<\/span> rese\u00adarch, and we\u2019re rea\u00addy to bring our expe\u00adri\u00adence with open soft\u00adware plat\u00adforms and inno\u00adva\u00adti\u00adon to the PyTorch foun\u00adda\u00adti\u00adon,\u201d said Brad McCre\u00addie, cor\u00adpo\u00adra\u00adte vice pre\u00adsi\u00addent, Data Cen\u00adter and Acce\u00adle\u00adra\u00adted Pro\u00adces\u00adsing, AMD<\/span>. \u201cAMD<\/span> Instinct acce\u00adle\u00adra\u00adtors and ROCm soft\u00adware power important HPC<\/span> and ML<\/span> sites around the world, from exas\u00adca\u00adle super\u00adcom\u00adpu\u00adters at rese\u00adarch labs to major cloud deploy\u00adments show\u00adca\u00adsing the con\u00adver\u00adgence of HPC<\/span> and AI<\/span>\/ML<\/span>. Tog\u00ade\u00adther with other foun\u00adda\u00adti\u00adon mem\u00adbers, we will sup\u00adport the acce\u00adle\u00adra\u00adti\u00adon of sci\u00adence and rese\u00adarch that can make a dra\u00adma\u00adtic impact on the world.\u201d<\/p>\n

\u201c<\/span>We are exci\u00adted to have AMD<\/span> join the PyTorch Foun\u00adda\u00adti\u00adon and bring its exten\u00adsi\u00adve exper\u00adti\u00adse in HPC<\/span>, AI<\/span> and ML<\/span> to our mem\u00adbers,\u201d said San\u00adtosh Janard\u00adhan, VP<\/span>, Infra\u00adstruc\u00adtu\u00adre at Meta. \u201cAMD<\/span> has con\u00adtin\u00adued to sup\u00adport PyTorch with its inte\u00adgra\u00adti\u00adon on ROCm open soft\u00adware plat\u00adform and has work\u00aded exten\u00adsi\u00adve\u00adly with the open-source com\u00admu\u00adni\u00adty and other foun\u00adda\u00adti\u00adon mem\u00adbers to advan\u00adce per\u00adfor\u00admance of ML<\/span> and AI<\/span> workloads. The col\u00adla\u00adbo\u00adra\u00adti\u00adve sup\u00adport offe\u00adred by AMD<\/span> con\u00adti\u00adnues our enga\u00adge\u00adment across broad indus\u00adtry initia\u00adti\u00adves for glo\u00adbal impact.\u201d<\/p>\n

AMD<\/span>, Advan\u00adcing AI<\/span> and ML<\/span><\/strong>
\nAMD<\/span> is uni\u00adque\u00adly posi\u00adtio\u00adned with its broad pro\u00adduct and soft\u00adware port\u00adfo\u00adlio to help cus\u00adto\u00admers and part\u00adners deve\u00adlop and deploy appli\u00adca\u00adti\u00adons with mul\u00adti\u00adple forms of AI<\/span> from the cloud and enter\u00adpri\u00adse, to the edge and end\u00adpoints. With a diver\u00adse set of hard\u00adware inclu\u00adding AMD<\/span> Instinct and Alveo acce\u00adle\u00adra\u00adtors, adap\u00adti\u00adve SoCs and CPUs, AMD<\/span> can sup\u00adport a wide varie\u00adty of per\u00adva\u00adsi\u00adve AI<\/span> and ML<\/span> models, from small edge points to lar\u00adge sca\u00adle out trai\u00adning and infe\u00adrence workloads.<\/p>\n

AMD<\/span> also works exten\u00adsi\u00adve\u00adly with the AI<\/span> open com\u00admu\u00adni\u00adty to pro\u00admo\u00adte and extend machi\u00adne and deep lear\u00adning capa\u00adbi\u00adli\u00adties and opti\u00admiza\u00adti\u00adons. Vitis AI<\/span><\/a> pro\u00advi\u00addes a com\u00adpre\u00adhen\u00adsi\u00adve AI<\/span> infe\u00adrence deve\u00adlo\u00adp\u00adment plat\u00adform for AMD<\/span> adap\u00adti\u00adve SoCs and Alveo data cen\u00adter acce\u00adle\u00adra\u00adtors. Vitis AI<\/span> plugs into com\u00admon soft\u00adware deve\u00adlo\u00adper tools and uti\u00adli\u00adzes a rich set of opti\u00admi\u00adzed open-source libra\u00adri\u00ades to empower soft\u00adware deve\u00adlo\u00adpers with machi\u00adne lear\u00adning acce\u00adle\u00adra\u00adti\u00adon as part of their soft\u00adware code.<\/p>\n

The ROCm\u2122 open soft\u00adware plat\u00adform is con\u00adstant\u00adly evol\u00adving to meet the needs of the AI<\/span>\/ML<\/span> and HPC<\/span> com\u00admu\u00adni\u00adty. With the latest release of ROCm 5.0, deve\u00adlo\u00adpers have access to turn-key AI<\/span> frame\u00adwork con\u00adtai\u00adners on AMD<\/span> Infi\u00adni\u00adty Hub<\/a>, advan\u00adced tools, stream\u00adli\u00adned instal\u00adla\u00adti\u00adon, and can expect to expe\u00adri\u00adence redu\u00adced ker\u00adnel launch times and fas\u00adter appli\u00adca\u00adti\u00adon per\u00adfor\u00admance. As well, with the latest PyTorch 1.12 release, AMD<\/span> ROCm\u2122 sup\u00adport has moved from beta to sta\u00adble. You can read more about that here<\/a>.<\/p>\n

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