{"id":22887,"date":"2016-04-01T07:58:10","date_gmt":"2016-04-01T05:58:10","guid":{"rendered":"http:\/\/www.planet3dnow.de\/cms\/?p=22887"},"modified":"2016-04-01T07:58:10","modified_gmt":"2016-04-01T05:58:10","slug":"new-amd-firepro-s9300-x2-server-gpu-fuels-efforts-to-create-largest-map-of-the-universe","status":"publish","type":"post","link":"https:\/\/www.planet3dnow.de\/cms\/22887-new-amd-firepro-s9300-x2-server-gpu-fuels-efforts-to-create-largest-map-of-the-universe\/","title":{"rendered":"New AMD<\/span> FirePro S9300<\/span> x2 Server GPU<\/span> Fuels Efforts to Create Largest Map of the Universe"},"content":{"rendered":"

<\/p>

Rese\u00adar\u00adchers at Cana\u00addi\u00adan Hydro\u00adgen Inten\u00adsi\u00adty Map\u00adping Expe\u00adri\u00adment Sel\u00adect World\u2019s Fas\u00adtest Sin\u00adgle-Pre\u00adcis\u00adi\u00adon GPU<\/span> Acce\u00adle\u00adra\u00adtor for High-Per\u00adfor\u00admance Com\u00adpu\u00adting Pro\u00adject<\/b><\/center>SUNNYVALE<\/span>, CA<\/span> \u2014 (Mar\u00adket\u00adwired) \u2014 03\/31\/16 \u2014 AMD<\/span><\/a> (NASDAQ<\/span>: AMD<\/span>) today announ\u00adced that rese\u00adar\u00adchers at the Cana\u00addi\u00adan Hydro\u00adgen Inten\u00adsi\u00adty Map\u00adping Expe\u00adri\u00adment (CHIME<\/span>) will harness the AMD<\/span> Fire\u00adPro\u2122 S9300<\/span> x2 Ser\u00adver GPU<\/span>, the world\u2019s fas\u00adtest sin\u00adgle-pre\u00adcis\u00adi\u00adon GPU<\/span> acce\u00adle\u00adra\u00adtor1<\/em>, to ana\u00adly\u00adze extra\u00ador\u00addi\u00adna\u00adry amounts of data to help crea\u00adte a new, very detail\u00aded 3D<\/span> map of the lar\u00adgest volu\u00adme of the Uni\u00adver\u00adse ever obser\u00adved. Rather than using tra\u00addi\u00adtio\u00adnal dish-shaped telesco\u00adpes, CHIME<\/span> con\u00adsists of four 100-met\u00adre-long cylind\u00adri\u00adcal reflec\u00adtors which cover an area lar\u00adger than five pro\u00adfes\u00adsio\u00adnal hockey rinks and gathers signals for the cri\u00adti\u00adcal com\u00adpu\u00adta\u00adtio\u00adnal ana\u00adly\u00adses sup\u00adpli\u00aded by the AMD<\/span> Fire\u00adPro S9300<\/span> x2 GPU<\/span> cluster.\n

The CHIME<\/span> pro\u00adject was crea\u00adted to inves\u00adti\u00adga\u00adte the dis\u00adco\u00advery that the expan\u00adsi\u00adon of the Uni\u00adver\u00adse is spee\u00adding up rather than slo\u00adwing down. Using con\u00adsu\u00admer tech\u00adno\u00adlo\u00adgies simi\u00adlar to tho\u00adse found in com\u00admon radio recei\u00advers, the telescope coll\u00adects radio waves that have tra\u00advel\u00adled through space for up to 11 bil\u00adli\u00adon years and feeds them into a mas\u00adsi\u00adve super\u00adcom\u00adpu\u00adter powered by a series of AMD<\/span> Fire\u00adPro S9300<\/span> x2 GPUs. The inten\u00adse num\u00adber crun\u00adching requi\u00adred to map the Universe\u2019s expan\u00adsi\u00adon in this way was pre\u00advious\u00adly cost-pro\u00adhi\u00adbi\u00adti\u00adve, but is now being enab\u00adled by AMD<\/span> Fire\u00adPro GPUs. The anti\u00adci\u00adpa\u00adted results will help crea\u00adte a high\u00adly-detail\u00aded map show\u00ading the inten\u00adsi\u00adty of the hydro\u00adgen radia\u00adti\u00adon from bil\u00adli\u00adons of gala\u00adxies, which will help sci\u00aden\u00adtists under\u00adstand the acce\u00adle\u00adra\u00adting expan\u00adsi\u00adon of the Universe.<\/p>\n

\u201c<\/span>We\u2019\u00adre proud to par\u00adti\u00adci\u00adpa\u00adte with the CHIME<\/span> pro\u00adject in this cos\u00admic oppor\u00adtu\u00adni\u00adty to map the evo\u00adlu\u00adti\u00adon of the Uni\u00adver\u00adse. The appli\u00adca\u00adti\u00adons of instinc\u00adti\u00adve com\u00adpu\u00adting are expan\u00adding rapidly as seen in this fasci\u00adna\u00adting CHIME<\/span> pro\u00adject, through the use of powerful gra\u00adphics com\u00adpu\u00adte capa\u00adbi\u00adli\u00adties,\u201d said Greg Stoner, seni\u00ador direc\u00adtor of Rade\u00adon Open Com\u00adpu\u00adte, Rade\u00adon Tech\u00adno\u00adlo\u00adgy Group, AMD<\/span>. \u201cNot only is the new AMD<\/span> Fire\u00adPro S9300<\/span> x2 Ser\u00adver GPU<\/span> the world\u2019s fas\u00adtest sin\u00adgle-pre\u00adcis\u00adi\u00adon GPU<\/span> acce\u00adle\u00adra\u00adtor, it is also the world\u2019s first pro\u00adfes\u00adsio\u00adnal GPU<\/span> acce\u00adle\u00adra\u00adtor to be equip\u00adped with High Band\u00adwidth Memo\u00adry and the first to ful\u00adly sup\u00adport the Rade\u00adon Open Com\u00adpu\u00adte plat\u00adform \u2014 an inte\u00adgral ele\u00adment of the GPUO\u00adpen com\u00adpu\u00adte initiative.\u201d<\/p>\n

\u201c<\/span>CHIME<\/span> faces an extra\u00ador\u00addi\u00adna\u00adry com\u00adpu\u00adting task, but with the onset of pre\u00advious\u00adly unavailable tech\u00adno\u00adlo\u00adgy and tools such as the AMD<\/span> Fire\u00adPro S9300<\/span> x2 GPU<\/span>, we now have the com\u00adpu\u00adta\u00adtio\u00adnal power, band\u00adwidth, and effi\u00adci\u00aden\u00adcy to make this data ana\u00adly\u00adsis pos\u00adsi\u00adble to stu\u00addy the impact of dark ener\u00adgy on our Uni\u00adver\u00adse,\u201d said Pro\u00adfes\u00adsor Keith Van\u00adder\u00adlin\u00adde, of the Dun\u00adlap Insti\u00adtu\u00adte for Astro\u00adno\u00admy &<\/span> Astro\u00adphy\u00adsics at the Uni\u00adver\u00adsi\u00adty of Toron\u00adto. \u201cOur rese\u00adarch is the result of the com\u00adbi\u00adned efforts of dedi\u00adca\u00adted hard\u00adware, com\u00adpu\u00adte soft\u00adware and sci\u00aden\u00adti\u00adfic rese\u00adarch. We\u2019\u00adre very app\u00adre\u00adcia\u00adti\u00adve of the sup\u00adport\u00ading efforts from AMD<\/span> in this project.\u201d<\/p>\n

To expo\u00adse cut\u00adting-edge com\u00adpu\u00adte capa\u00adbi\u00adli\u00adties of the AMD<\/span> Fire\u00adPro S9300<\/span> x2 GPU<\/span> and sup\u00adport uni\u00adque oppor\u00adtu\u00adni\u00adties such as the CHIME<\/span> pro\u00adject, AMD<\/span> deve\u00adlo\u00adped the Rade\u00adon Open Com\u00adpu\u00adte plat\u00adform (based on the Boltz\u00admann initia\u00adti\u00adve<\/a>), a Linux\u00ae-focused open source soft\u00adware par\u00adal\u00adlel com\u00adpu\u00adting plat\u00adform opti\u00admi\u00adzed for HPC<\/span> and Ultra-sca\u00adle com\u00adpu\u00adting. The Rade\u00adon Open Com\u00adpu\u00adte ker\u00adnel pro\u00advi\u00addes direct access to the gra\u00adphics hard\u00adware, offe\u00adring pro\u00adgramm\u00aders more con\u00adtrol over the code exe\u00adcu\u00adti\u00adon. Dis\u00adco\u00adver more at GPUOpen.com for the Rade\u00adon Open Com\u00adpu\u00adte plat\u00adform-opti\u00admi\u00adzed open source math libra\u00adri\u00ades, bench\u00admarks and appli\u00adca\u00adti\u00adons from AMD<\/span> and third-par\u00adty partners.<\/p>\n

The AMD<\/span> Fire\u00adPro S9300<\/span> x2 Ser\u00adver GPU<\/span> is a com\u00adpel\u00adling pro\u00adduct com\u00adbi\u00adning an open source com\u00adpi\u00adler, run\u00adtime and dri\u00adver full access, mas\u00adsi\u00adve com\u00adpu\u00adte per\u00adfor\u00admance, and pro\u00adces\u00adsor den\u00adsi\u00adty sup\u00adport\u00ading tight power and finan\u00adcial bud\u00adgets. For more infor\u00adma\u00adti\u00adon on this pro\u00adduct or how AMD<\/span> Fire\u00adPro\u2122 gra\u00adphics can be inte\u00adgra\u00adted into aca\u00adde\u00admic com\u00adpu\u00adte pro\u00adjects, cont\u00adact the AMD<\/span> Pro\u00adfes\u00adsio\u00adnal Gra\u00adphics Team by email: AMD<\/span>.ProfessionalGraphics@amd.com<\/a><\/p>\n

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

Based on the third gene\u00adra\u00adti\u00adon AMD<\/span> Gra\u00adphics Core Next (GCN<\/span><\/a>) archi\u00adtec\u00adtu\u00adre, the AMD<\/span> Fire\u00adPro S9300<\/span> x2 Ser\u00adver GPU<\/span> deli\u00advers up to 13.9 TFLOPS<\/span> of peak sin\u00adgle-pre\u00adcis\u00adi\u00adon floa\u00adting point per\u00adfor\u00admance \u2014 grea\u00adter than any other pro\u00adfes\u00adsio\u00adnal dual GPU<\/span> acce\u00adle\u00adra\u00adtor available on the mar\u00adket today for sin\u00adgle-pre\u00adcis\u00adi\u00adon com\u00adpu\u00adte1<\/em>. It is the first GPU<\/span> for data cen\u00adter use equip\u00adped with High Band\u00adwidth Memo\u00adry (HBM<\/span>). The 1TB<\/span>\/s memo\u00adry band\u00adwidth enab\u00adled by HBM<\/span> on the AMD<\/span> Fire\u00adPro S9300<\/span> x2 GPU<\/span> is 3.5X<\/span> the memo\u00adry band\u00adwidth of the clo\u00adsest com\u00adpe\u00adti\u00adti\u00adve solu\u00adti\u00adon2<\/em> while pro\u00advi\u00adding over 15X<\/span> the memo\u00adry band\u00adwidth and 12X<\/span> the peak sin\u00adgle-pre\u00adcis\u00adi\u00adon per\u00adfor\u00admance of com\u00adpe\u00adting CPU<\/span> solu\u00adti\u00adons, making it a com\u00adpel\u00adling addi\u00adti\u00adon to tra\u00addi\u00adtio\u00adnal x86 plat\u00adforms3<\/em>.<\/p>\n