{"id":13491,"date":"2014-11-21T18:02:46","date_gmt":"2014-11-21T17:02:46","guid":{"rendered":"http:\/\/www.planet3dnow.de\/cms\/?p=13491"},"modified":"2014-11-26T15:11:06","modified_gmt":"2014-11-26T14:11:06","slug":"penguin-computings-cluster-collaboration-with-advanced-micro-devices-makes-heterogeneous-system-architecture-clustering-a-reality","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/13491-penguin-computings-cluster-collaboration-with-advanced-micro-devices-makes-heterogeneous-system-architecture-clustering-a-reality\/","title":{"rendered":"Penguin Computing\u2019s Cluster Collaboration with Advanced Micro Devices Makes Heterogeneous System Architecture Clustering a Reality"},"content":{"rendered":"

Hete\u00adro\u00adge\u00adneous Sys\u00adtem Archi\u00adtec\u00adtu\u00adre Pro\u00adgress Pro\u00advi\u00addes Indus\u00adtry Lea\u00adder\u00adship for More Effi\u00adci\u00adent Computing<\/em><\/p>\n

New Orleans, LA-SC14-Nov. 20, 2014 \u2014 <\/strong>Pen\u00adgu\u00adin Com\u00adpu\u00adting, pro\u00advi\u00adder of high per\u00adfor\u00admance com\u00adpu\u00adting, enter\u00adpri\u00adse data cen\u00adter and cloud solu\u00adti\u00adons, today announ\u00adced the first appli\u00adca\u00adti\u00adon opti\u00admi\u00adzed acce\u00adle\u00adra\u00adted pro\u00adces\u00adsing unit (APU<\/span>) clus\u00adters, making seam\u00adless GPU<\/span> and CPU<\/span> memo\u00adry sha\u00adring on clus\u00adters a rea\u00adli\u00adty based on hete\u00adro\u00adge\u00adneous sys\u00adtem archi\u00adtec\u00adtu\u00adre (HSA<\/span>) from Advan\u00adced Micro Devices (NYSE<\/span>: AMD<\/span>). The shared memo\u00adry capa\u00adbi\u00adli\u00adty invol\u00adves very light\u00adweight con\u00adtext swit\u00adches to switch instanta\u00adneous\u00adly bet\u00adween the GPU<\/span> and CPU<\/span>, whi\u00adche\u00adver code runs best at a given moment.<\/p>\n

Today\u2019s appli\u00adca\u00adti\u00adons requi\u00adre power effi\u00adci\u00aden\u00adcy and high per\u00adfor\u00admance achie\u00adved through high\u00adly par\u00adal\u00adlel com\u00adpu\u00adting, howe\u00adver GPUs and CPUs pre\u00adsent a bot\u00adt\u00adlen\u00adeck to effi\u00adci\u00adent clus\u00adter func\u00adtion\u00ada\u00adli\u00adty due to sepa\u00adra\u00adte memo\u00adry space. This tra\u00addi\u00adtio\u00adnal archi\u00adtec\u00adtu\u00adre results in inef\u00adfi\u00adci\u00adent GPU<\/span>\/CPU<\/span> com\u00admu\u00adni\u00adca\u00adti\u00adon, which is a chall\u00adenge to scaling.<\/p>\n

\u201c<\/span>We are making the\u00adse machi\u00adnes imme\u00addia\u00adte\u00adly available for eva\u00adlua\u00adti\u00adon as a tre\u00admen\u00addous tool for soft\u00adware deve\u00adlo\u00adp\u00adment,\u201d said Phil Pokor\u00adny, Chief Tech\u00adno\u00adlo\u00adgy Offi\u00adcer, Pen\u00adgu\u00adin Com\u00adpu\u00adting. \u201cHSA<\/span> is a rea\u00adli\u00adty and our tech\u00adno\u00adlo\u00adgy is alre\u00ada\u00addy in the hands of major U.S. labs. Pen\u00adgu\u00adin Computing\u2019s exten\u00adsi\u00adve expe\u00adri\u00adence in APU<\/span> clus\u00adter deve\u00adlo\u00adp\u00adment and imple\u00admen\u00adta\u00adti\u00adon is instru\u00admen\u00adtal in this pro\u00adgress, in addi\u00adti\u00adon to clo\u00adse col\u00adla\u00adbo\u00adra\u00adte with AMD<\/span>.\u201d<\/p>\n

Named J\u00e4\u00e4\u00adtik\u00adk\u00f6, or ice\u00adberg in Fin\u00adnish, the clus\u00adter is curr\u00adent\u00adly being demons\u00adtra\u00adted at AMD<\/span>\u2019s SC14<\/span> booth #839 and com\u00adbi\u00adnes 10 AMD<\/span> APU<\/span> com\u00adpu\u00adte nodes, plus head node based on Penguin\u2019s Alt\u00adus 2A30<\/span> deve\u00adlo\u00adp\u00adment plat\u00adform, with high per\u00adfor\u00admance ether\u00adnet using Penguin\u2019s Arc\u00adti\u00adca open ether\u00adnet switches.<\/p>\n

\u201c<\/span>Initi\u00adal feed\u00adback from ear\u00adly adop\u00adters rein\u00adforces our belief that this col\u00adla\u00adbo\u00adra\u00adti\u00adon with Pen\u00adgu\u00adin Com\u00adpu\u00adting is an important step for\u00adward for the indus\u00adtry,\u201d said Karl Freund, cor\u00adpo\u00adra\u00adte vice pre\u00adsi\u00addent, Pro\u00adduct Manage\u00adment and Mar\u00adket, Ser\u00adver Busi\u00adness Unit, AMD<\/span>. \u201cThe poten\u00adti\u00adal of modern hete\u00adro\u00adge\u00adneous archi\u00adtec\u00adtures is exci\u00adting, and col\u00adla\u00adbo\u00adra\u00adti\u00adons such as the\u00adse can result in signi\u00adfi\u00adcant steps for\u00adward in per\u00adfor\u00admance for a broad ran\u00adge of soft\u00adware applications.\u201d<\/p>\n

The oil and gas indus\u00adtry is an exam\u00adp\u00adle of a cus\u00adto\u00admer seg\u00adment that could expe\u00adri\u00adence signi\u00adfi\u00adcant bene\u00adfits from this capa\u00adbi\u00adli\u00adty. With the oil and gas sector\u2019s need for GPU<\/span> par\u00adal\u00adlel codes and sin\u00adgle-pre\u00adcis\u00adi\u00adon, APU<\/span> has almost a tera\u00adflop of sin\u00adgle pre\u00adcis\u00adi\u00adon floa\u00adting point performance.<\/p>\n

The com\u00adbi\u00adned Pen\u00adgu\u00adin Com\u00adpu\u00adting and AMD<\/span> solu\u00adti\u00adon is a very cost-effec\u00adti\u00adve way to get star\u00adted with one node or cluster.<\/p>\n

Visit http:\/\/www.penguincomputing.com\/products\/rackmount-servers\/altus\/altus-2a30\/ <\/a>to learn more about the Alt\u00adus 2a30.<\/p>\n","protected":false},"excerpt":{"rendered":"

New Orleans, LA-SC14-Nov. 20, 2014 \u2014 Pen\u00adgu\u00adin Com\u00adpu\u00adting, pro\u00advi\u00adder of high per\u00adfor\u00admance com\u00adpu\u00adting, enter\u00adpri\u00adse data cen\u00adter and cloud solu\u00adti\u00adons, today announ\u00adced the first appli\u00adca\u00adti\u00adon opti\u00admi\u00adzed acce\u00adle\u00adra\u00adted pro\u00adces\u00adsing unit (APU<\/span>) clus\u00adters, making seam\u00adless GPU<\/span> and CPU<\/span> memo\u00adry sha\u00adring on clus\u00adters a rea\u00adli\u00adty based on hete\u00adro\u00adge\u00adneous sys\u00adtem archi\u00adtec\u00adtu\u00adre (HSA<\/span>) from Advan\u00adced Micro Devices (NYSE<\/span>: AMD<\/span>). The shared memo\u00adry capa\u00adbi\u00adli\u00adty invol\u00adves very light\u00adweight con\u00adtext swit\u00adches to switch instanta\u00adneous\u00adly bet\u00adween the GPU<\/span> and CPU<\/span>, whi\u00adche\u00adver code runs best at a given moment. (\u2026) Wei\u00adter\u00adle\u00adsen \u00bb<\/a><\/p>\n","protected":false},"author":8,"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":[],"class_list":["post-13491","post","type-post","status-publish","format-standard","hentry","category-pressemitteilungen","entry"],"share_on_mastodon":{"url":"","error":""},"_links":{"self":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/13491","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\/8"}],"replies":[{"embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/comments?post=13491"}],"version-history":[{"count":3,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/13491\/revisions"}],"predecessor-version":[{"id":13494,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/13491\/revisions\/13494"}],"wp:attachment":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/media?parent=13491"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/categories?post=13491"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/tags?post=13491"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}