{"id":52797,"date":"2019-11-15T14:27:20","date_gmt":"2019-11-15T13:27:20","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=52797"},"modified":"2019-11-15T14:27:20","modified_gmt":"2019-11-15T13:27:20","slug":"mindtech-announces-enhanced-scalability-support-for-chameleon","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/52797-mindtech-announces-enhanced-scalability-support-for-chameleon\/","title":{"rendered":"MINDTECH<\/span> ANNOUNCES<\/span> ENHANCED<\/span> SCALABILITY<\/span> SUPPORT<\/span> FOR<\/span> CHAMELEON<\/span>"},"content":{"rendered":"

LONDON<\/span>, Nov. 15, 2019 \/PRNewswire\/ \u2013 <\/b>Mind\u00adtech Glo\u00adbal Ltd, a UK<\/span> based start-up, has announ\u00adced con\u00adtai\u00adner sup\u00adport to allow run\u00adning mul\u00adti\u00adple ins\u00adtances of their Cha\u00adme\u00adle\u00adon syn\u00adthe\u00adtic data gene\u00adra\u00adti\u00adon tool on clus\u00adte\u00adred plat\u00adforms.  This approach enables full use of available com\u00adpu\u00adte resour\u00adces lea\u00adding to a signi\u00adfi\u00adcant reduc\u00adtion in trai\u00adning cycle times for neu\u00adral net\u00adworks using Syn\u00adthe\u00adtic data.<\/p>\n

Sca\u00adla\u00adbi\u00adli\u00adty by design<\/b><\/p>\n

Mindtech\u2019s Cha\u00adme\u00adle\u00adon was built from the out\u00adset with the capa\u00adbi\u00adli\u00adty to sca\u00adle for gene\u00adra\u00adting lar\u00adge amounts of data to meet the demands of trai\u00adning visu\u00adal neu\u00adral net\u00adworks for AI<\/span> sys\u00adtems.  The simu\u00adla\u00adtor incor\u00adpo\u00adra\u00adtes a scrip\u00adting engi\u00adne, enab\u00adling easy capa\u00adbi\u00adli\u00adties for repea\u00adted simu\u00adla\u00adti\u00adon runs, chan\u00adging sin\u00adgle varia\u00adbles at a time.  The sce\u00adna\u00adrio edi\u00adtor is inten\u00added to allow the quick and easy crea\u00adti\u00adon of mul\u00adti\u00adple dif\u00adfe\u00adrent sequen\u00adces, to crea\u00adte the data requi\u00adred for the inten\u00added use case.<\/p>\n

Exploi\u00adting scalability<\/b><\/p>\n

The Sce\u00adna\u00adrio edi\u00adtor and simu\u00adla\u00adti\u00adon scrip\u00adting engi\u00adne bet\u00adween them ensu\u00adre that the Cha\u00adme\u00adle\u00adon tools can easi\u00adly be used to gene\u00adra\u00adte huge amounts of data.  To allow the user to effec\u00adtively exploit this requi\u00adres that simu\u00adla\u00adti\u00adon and ren\u00adde\u00adring hard\u00adware be ful\u00adly uti\u00adli\u00adsed.  To achie\u00adve this the tools were desi\u00adgned to ful\u00adly employ all resour\u00adces available, taking advan\u00adta\u00adge of mul\u00adti-thread pro\u00adces\u00adsors and GPU<\/span> acce\u00adle\u00adra\u00adtors available on the sys\u00adtem hos\u00adting the simu\u00adla\u00adtor.  To fur\u00adther enhan\u00adce this capa\u00adbi\u00adli\u00adty, Mind\u00adtech has updated the simu\u00adla\u00adtor and encap\u00adsu\u00adla\u00adted it in a con\u00adtai\u00adne\u00adri\u00adsed form, to allow use of such tech\u00adno\u00adlo\u00adgies as Sin\u00adgu\u00adla\u00adri\u00adty and Docker.  This in turns allows deploy\u00adment within a clus\u00adte\u00adred envi\u00adron\u00adment like Kuber\u00adnetes.  This new func\u00adtion\u00ada\u00adli\u00adty allows for deploy\u00adment at ser\u00adver and cloud sca\u00adle, enhan\u00adcing the abili\u00adty to uti\u00adli\u00adse com\u00adpu\u00adte resour\u00adces available. <\/p>\n

The need for scalability<\/b><\/p>\n

The crea\u00adti\u00adon of the vir\u00adtu\u00adal worlds, used to crea\u00adte Mindtech\u2019s high qua\u00adli\u00adty syn\u00adthe\u00adtic data, requi\u00adres the use of sophisti\u00adca\u00adted 3D<\/span> ren\u00adde\u00adring tech\u00adni\u00adques, as well as signi\u00adfi\u00adcant pro\u00adces\u00adsor power, deli\u00adver\u00aded by GPU<\/span> acce\u00adle\u00adra\u00adtors, to help ensu\u00adre the vir\u00adtu\u00adal world beha\u00adves in a rea\u00adli\u00adstic, \u201clife-like\u201d man\u00adner.  Each trai\u00adning task for a neu\u00adral net\u00adwork will requi\u00adre hundreds of thou\u00adsands or even mil\u00adli\u00adons of images, each indi\u00advi\u00addu\u00adal\u00adly ren\u00adde\u00adred, and anno\u00adta\u00adted for the trai\u00adning task.  The use of con\u00adtai\u00adners allows the user to run mul\u00adti\u00adple simu\u00adla\u00adti\u00adon runs in par\u00adal\u00adlel, vast\u00adly redu\u00adcing the time to crea\u00adte data, and shor\u00adtening the over\u00adall trai\u00adning cycle for neu\u00adral networks.<\/p>\n

Says Ogi Brkic, Cor\u00adpo\u00adra\u00adte Vice Pre\u00adsi\u00addent and Gene\u00adral Mana\u00adger of the Data Cen\u00adter GPU<\/span> Busi\u00adness Unit at AMD<\/span>, Rade\u00adon Tech\u00adno\u00adlo\u00adgy Group: \u201cThe com\u00adbi\u00adna\u00adti\u00adon of the latest AMD<\/span> EPYC<\/span>\u2122 7002 Series Pro\u00adces\u00adsor and Rade\u00adon Instinct\u2122MI50<\/span> GPU<\/span> acce\u00adle\u00adra\u00adtor repres\u00adents the best of what\u2019s pos\u00adsi\u00adble for modern AI<\/span> workloads. Mindtech\u2019s Cha\u00adme\u00adle\u00adon lever\u00ada\u00adges this powerful com\u00adpu\u00adte plat\u00adform to acce\u00adle\u00adra\u00adte syn\u00adthe\u00adtic data gene\u00adra\u00adti\u00adon and to help redu\u00adce trai\u00adning cycle times with a con\u00adtai\u00adne\u00adri\u00adzed approach.\u201d<\/p>\n

Avai\u00adla\u00adbi\u00adli\u00adty<\/b><\/p>\n

Mindtech\u2019s Cha\u00adme\u00adle\u00adon AI<\/span> Tools<\/i> sca\u00adla\u00adbi\u00adli\u00adty using con\u00adtai\u00adners will be shown on the AMD<\/span> stand at Super\u00adcom\u00adpu\u00adting 2019, Novem\u00adber 18\u201321, Den\u00adver, Colorado. <\/p>\n

Cont\u00adact<\/b><\/p>\n

For fur\u00adther infor\u00adma\u00adti\u00adon, plea\u00adse cont\u00adact chris.longstaff@mindtech.global<\/a>, or https:\/\/mindtech.global<\/a><\/p>\n

AMD<\/span>, the AMD<\/span> logo, EPYC<\/span>, Rade\u00adon Instinct, and com\u00adbi\u00adna\u00adti\u00adons the\u00adreof are trade\u00admarks of Advan\u00adced Micro Devices, Inc.<\/u><\/p>\n

SOURCE<\/span> Mind\u00adtech Glo\u00adbal Ltd<\/p>\n

<\/p>\n

RELATED<\/span> LINKS<\/span><\/h4>\n

https:\/\/mindtech.global<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

LONDON<\/span>, Nov. 15, 2019 \/PRNewswire\/ \u2013 Mind\u00adtech Glo\u00adbal Ltd, a UK<\/span> based start-up, has announ\u00adced con\u00adtai\u00adner sup\u00adport to allow run\u00adning mul\u00adti\u00adple ins\u00adtances of their Cha\u00adme\u00adle\u00adon syn\u00adthe\u00adtic data gene\u00adra\u00adti\u00adon tool on clus\u00adte\u00adred plat\u00adforms.  This approach enables full use of available com\u00adpu\u00adte resour\u00adces lea\u00adding to a signi\u00adfi\u00adcant reduc\u00adtion in trai\u00adning cycle times for neu\u00adral net\u00adworks using Syn\u00adthe\u00adtic data. Sca\u00adla\u00adbi\u00adli\u00adty by design Mindtech\u2019s (\u2026) Wei\u00adter\u00adle\u00adsen \u00bb<\/a><\/p>\n","protected":false},"author":593,"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":[2116,966,2019,2115],"class_list":["post-52797","post","type-post","status-publish","format-standard","hentry","category-pressemitteilungen","tag-ai-training","tag-amd","tag-epyc-7002","tag-mindtech","entry"],"share_on_mastodon":{"url":"","error":""},"_links":{"self":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/52797","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\/593"}],"replies":[{"embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/comments?post=52797"}],"version-history":[{"count":1,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/52797\/revisions"}],"predecessor-version":[{"id":52798,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/52797\/revisions\/52798"}],"wp:attachment":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/media?parent=52797"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/categories?post=52797"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/tags?post=52797"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}