{"id":52190,"date":"2019-11-05T17:45:08","date_gmt":"2019-11-05T16:45:08","guid":{"rendered":"https:\/\/www.planet3dnow.de\/cms\/?p=52190"},"modified":"2019-11-05T17:45:08","modified_gmt":"2019-11-05T16:45:08","slug":"globalfoundries-and-sifive-to-deliver-next-level-of-high-bandwidth-memory-on-12lp-platform-for-ai-applications","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/52190-globalfoundries-and-sifive-to-deliver-next-level-of-high-bandwidth-memory-on-12lp-platform-for-ai-applications\/","title":{"rendered":"GLOBALFOUNDRIES<\/span> and SiFive to Deliver Next Level of High Bandwidth Memory on 12LP<\/span> Platform for AI<\/span> Applications"},"content":{"rendered":"

Nov 05, 2019<\/p>\n

\n
\n
\n
\n
\n
\n
\n
\n

Next gene\u00adra\u00adti\u00adon high band\u00adwidth memo\u00adry solu\u00adti\u00adon based on GF<\/span>\u2019s most advan\u00adced Fin\u00adFET plat\u00adform aims to deli\u00adver capa\u00adci\u00adty, speed and power for cloud based AI<\/span> Applications<\/em><\/p>\n

San\u00adta Cla\u00adra Calif. and Hsin\u00adchu, Tai\u00adwan, Novem\u00adber 5, 2019 \u2013 GLOBALFOUNDRIES<\/span>\u00ae (GF<\/span>\u00ae) and SiFi\u00adve, Inc. announ\u00adced today at GLOBALFOUNDRIES<\/span> Tech\u00adno\u00adlo\u00adgy Con\u00adfe\u00adrence (GTC<\/span>) in Tai\u00adwan that they are working to extend high DRAM<\/span> per\u00adfor\u00admance levels with High Band\u00adwidth Memo\u00adry (HBM2E<\/span>) on GF<\/span>\u2019s recent\u00adly announ\u00adced 12LP<\/span>+<\/a>  Fin\u00adFET solu\u00adti\u00adon, with 2.5D<\/span> pack\u00ada\u00adging design ser\u00advices to enable fast time-to-mar\u00adket for Arti\u00adfi\u00adci\u00adal Intel\u00adli\u00adgence (AI<\/span>) applications.<\/p>\n

In order to achie\u00adve the capa\u00adci\u00adty and band\u00adwidth for data-inten\u00adsi\u00adve AI<\/span> trai\u00adning appli\u00adca\u00adti\u00adons, sys\u00adtem desi\u00adgners are chal\u00adlen\u00adged with squeezing more band\u00adwidth into a smal\u00adler area while main\u00adtai\u00adning a reasonable power pro\u00adfi\u00adle. SiFive\u2019s cus\u00adto\u00admizable high band\u00adwidth memo\u00adry inter\u00adface on GF<\/span>\u2019s 12LP<\/span> plat\u00adform and 12LP<\/span>+ solu\u00adti\u00adon will enable easy inte\u00adgra\u00adti\u00adon of high band\u00adwidth memo\u00adry into a sin\u00adgle Sys\u00adtem-on-Chip (SoC) solu\u00adti\u00adons to deli\u00adver fast, power-effi\u00adci\u00adent data pro\u00adces\u00adsing for AI<\/span> appli\u00adca\u00adti\u00adons in the com\u00adpu\u00adting and wired infra\u00adstruc\u00adtu\u00adre markets. <\/p>\n

As a part of the col\u00adla\u00adbo\u00adra\u00adti\u00adon, desi\u00adgners will also have access to SiFive\u2019s RISC<\/span>\u2011V IP<\/span> port\u00adfo\u00adlio and DesignSha\u00adre IP<\/span> eco\u00adsys\u00adtem<\/a>, which will levera\u00adge GF<\/span>\u2019s 12LP<\/span>+ Design Tech\u00adno\u00adlo\u00adgy Co-Opti\u00admiza\u00adti\u00adon (DTCO<\/span>), enab\u00adling them to signi\u00adfi\u00adcant\u00adly increase sili\u00adcon spe\u00adcia\u00adliza\u00adti\u00adon, impro\u00adve design effi\u00adci\u00aden\u00adcy and deli\u00adver dif\u00adfe\u00adren\u00adtia\u00adted SoC solu\u00adti\u00adons quick\u00adly and cost-effectively. <\/p>\n

\u201c<\/span>Exten\u00adding SiFive\u2019s refe\u00adrence IP<\/span> plat\u00adform, with HBM2E<\/span>, on GF<\/span>\u2019s best-in-class per\u00adfor\u00admance 12LP<\/span>+ solu\u00adti\u00adon\u00adde\u00adli\u00advers new levels of per\u00adfor\u00admance and inte\u00adgra\u00adti\u00adon for next gene\u00adra\u00adti\u00adon SoCs and acce\u00adle\u00adra\u00adtors,\u201d said Mohit Gupta, vice pre\u00adsi\u00addent and gene\u00adral mana\u00adger, IP<\/span> Busi\u00adness Unit at SiFi\u00adve. \u201cDeploy\u00adment of high\u00adly opti\u00admi\u00adzed sili\u00adcon requi\u00adres high\u00adly cus\u00adto\u00admizable capa\u00adbi\u00adli\u00adties in order to rea\u00adli\u00adze the much-nee\u00added hig\u00adher TOPS<\/span> per mil\u00adli\u00adwatt with low laten\u00adcy per\u00adfor\u00admance requi\u00adred for AI<\/span>, while balan\u00adcing the needs for low power and smal\u00adler area footprints.\u201d<\/p>\n

\u201c<\/span>At GF<\/span>, we con\u00adti\u00adnue our com\u00admit\u00adment to pro\u00advi\u00adding dif\u00adfe\u00adren\u00adtia\u00adted Fin\u00adFET spe\u00adci\u00adfic appli\u00adca\u00adti\u00adon solu\u00adti\u00adons and IP<\/span> that allow our cli\u00adents to deve\u00adlop per\u00adfor\u00admance enhan\u00adced pro\u00adducts for AI<\/span> appli\u00adca\u00adti\u00adons,\u201d said Ted Leta\u00advic, CTO<\/span> of Com\u00adpu\u00adting and Wired Infra\u00adstruc\u00adtu\u00adre at GF<\/span>. \u201cTog\u00ade\u00adther, with GF<\/span>\u2019s most advan\u00adced Fin\u00adFET plat\u00adform and SiFive\u2019s uni\u00adque design metho\u00addo\u00adlo\u00adgy, we will deve\u00adlop a uni\u00adque high per\u00adfor\u00admance edge com\u00adpu\u00adting solu\u00adti\u00adon which empowers desi\u00adgners to take full advan\u00adta\u00adge of the data deluge.\u201d<\/p>\n

GF<\/span>\u2019s 12LP<\/span>+ an inno\u00adva\u00adti\u00adve new solu\u00adti\u00adon for AI<\/span> trai\u00adning and infe\u00adrence appli\u00adca\u00adti\u00adons, offers desi\u00adgners a high-speed, low-power 0.5Vmin SRAM<\/span> bit\u00adcell that sup\u00adports the fast, power-effi\u00adci\u00adent shut\u00adt\u00adling of data bet\u00adween pro\u00adces\u00adsors and memo\u00adry. Moreo\u00adver, a new inter\u00adpo\u00adser for 2.5D<\/span> packa\u00adges faci\u00adli\u00adta\u00adtes the inte\u00adgra\u00adti\u00adon of high-band\u00adwidth memo\u00adry with pro\u00adces\u00adsors for fast, power-effi\u00adci\u00adent data processing. <\/p>\n

SiFive\u2019s HBM2E<\/span> inter\u00adface and cus\u00adtom IP<\/span> solu\u00adti\u00adon on GF<\/span>\u2019s 12LP<\/span> and 12LP<\/span>+ are now under deve\u00adlo\u00adp\u00adment at GF<\/span>\u2019s Fab 8 in Mal\u00adta, New York. Cli\u00adents can start opti\u00admi\u00adzing their chip designs to deve\u00adlop dif\u00adfe\u00adren\u00adtia\u00adted solu\u00adti\u00adons for high per\u00adfor\u00admance com\u00adpu\u00adte and edge AI<\/span> appli\u00adca\u00adti\u00adons in 1H<\/span> 2020. <\/p>\n

About SiFi\u00adve<\/strong><\/p>\n

SiFi\u00adve is the lea\u00adding pro\u00advi\u00adder of mar\u00adket-rea\u00addy pro\u00adces\u00adsor core IP<\/span> and sili\u00adcon solu\u00adti\u00adons based on the free and open RISC<\/span>\u2011V ins\u00adtruc\u00adtion set archi\u00adtec\u00adtu\u00adre. Led by a team of sea\u00adso\u00adned sili\u00adcon exe\u00adcu\u00adti\u00adves and the RISC<\/span>\u2011V inven\u00adtors, SiFi\u00adve helps SoC desi\u00adgners redu\u00adce time-to-mar\u00adket and rea\u00adli\u00adze cost savings with cus\u00adto\u00admi\u00adzed, open-archi\u00adtec\u00adtu\u00adre pro\u00adces\u00adsor cores, and demo\u00adcra\u00adti\u00adzes access to opti\u00admi\u00adzed sili\u00adcon by enab\u00adling sys\u00adtem desi\u00adgners in all mar\u00adket ver\u00adti\u00adcals to build cus\u00adto\u00admi\u00adzed RISC<\/span>\u2011V based semi\u00adcon\u00adduc\u00adtors. With 15 offices world\u00adwi\u00adde, SiFi\u00adve has back\u00ading from Sut\u00adter Hill Ven\u00adtures, Qual\u00adcomm Ven\u00adtures, Spark Capi\u00adtal, Osa\u00adge Uni\u00adver\u00adsi\u00adty Part\u00adners, Cheng\u00adwei, Hua\u00admi, SK<\/span> Hynix, Intel Capi\u00adtal, and Wes\u00adtern Digi\u00adtal. For more infor\u00adma\u00adti\u00adon, www.sifive.com<\/a>.<\/p>\n

About GF<\/span><\/strong><\/p>\n

GLOBALFOUNDRIES<\/span> (GF<\/span>) is the world\u2019s lea\u00adding spe\u00adcial\u00adty foundry. We deli\u00adver dif\u00adfe\u00adren\u00adtia\u00adted fea\u00adture-rich solu\u00adti\u00adons that enable our cli\u00adents to deve\u00adlop inno\u00adva\u00adti\u00adve pro\u00adducts for high-growth mar\u00adket seg\u00adments. GF<\/span> pro\u00advi\u00addes a broad ran\u00adge of plat\u00adforms and fea\u00adtures with a uni\u00adque mix of design, deve\u00adlo\u00adp\u00adment and fabri\u00adca\u00adti\u00adon ser\u00advices. With an at-sca\u00adle manu\u00adfac\u00adtu\u00adring foot\u00adprint span\u00adning the U.S., Euro\u00adpe and Asia, GF<\/span> has the fle\u00adxi\u00adbi\u00adli\u00adty and agi\u00adli\u00adty to meet the dyna\u00admic needs of cli\u00adents across the glo\u00adbe. GF<\/span> is owned by Muba\u00add\u00ada\u00adla Invest\u00adment Com\u00adpa\u00adny. For more infor\u00adma\u00adti\u00adon, visit globalfoundries.com<\/a>.<\/p>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

Nov 05, 2019 Next gene\u00adra\u00adti\u00adon high band\u00adwidth memo\u00adry solu\u00adti\u00adon based on GF<\/span>\u2019s most advan\u00adced Fin\u00adFET plat\u00adform aims to deli\u00adver capa\u00adci\u00adty, speed and power for cloud based AI<\/span> Appli\u00adca\u00adti\u00adons San\u00adta Cla\u00adra Calif. and Hsin\u00adchu, Tai\u00adwan, Novem\u00adber 5, 2019 \u2013 GLOBALFOUNDRIES<\/span>\u00ae (GF<\/span>\u00ae) and SiFi\u00adve, Inc. announ\u00adced today at GLOBALFOUNDRIES<\/span> Tech\u00adno\u00adlo\u00adgy Con\u00adfe\u00adrence (GTC<\/span>) in Tai\u00adwan that they are working (\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":[1368,569,2094,2093],"class_list":["post-52190","post","type-post","status-publish","format-standard","hentry","category-pressemitteilungen","tag-12lp","tag-gloabalfoundries","tag-hbme2","tag-sifive","entry"],"share_on_mastodon":{"url":"","error":""},"_links":{"self":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/52190","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=52190"}],"version-history":[{"count":1,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/52190\/revisions"}],"predecessor-version":[{"id":52191,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/posts\/52190\/revisions\/52191"}],"wp:attachment":[{"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/media?parent=52190"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/categories?post=52190"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.planet3dnow.de\/cms\/wp-json\/wp\/v2\/tags?post=52190"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}