Kwintetho yasekuseni ye-AMD, i-CEO yenkampani uLisa Su ubonise ukusuka kwinqanaba iprosesa ye-graphics ye-7nm ye-Navi (RDNA), eya kuba sisiseko se-Radeon RX 5700 yamakhadi evidiyo aqaliswe ngoJulayi Ukuthatha iifoto ezicacileyo kumgama onjalo kwakuyingxaki , kodwa Khetha amalungu oshicilelo avunyelwe ukuba abambe le GPU ezandleni zabo. Owu, ayingabo bonke abaxhalabileyo malunga nobukhulu kangangokuba bahlala bephethe izixhobo zokulinganisa ezichanekileyo kunye nabo, kwaye i-AMD censors ayinakukwazi ukuvuma ukukhohlisa okunjalo ngeesampulu zeemveliso ezingekavezwa.
Kwaye kunjalo abameli besayithi Sikwazile ukufumana umbono ongalunganga wendawo yokufa yeNavi GPU. Ngokutsho kwabo, ayidluli 275 mm2. Nokuba sicinga ukuba olu lubalo olurhabaxa kakhulu, izibonelelo zokusebenzisa itekhnoloji yenkqubo ye-TSMC ye-7nm ziyabonakala apha. Njengoko bekutshiwo ngaphambili kwinkcazo-ntetho, isizukulwana sokuqala se-GPUs kunye ne-RDNA ye-architecture iphucula umlinganiselo wokusebenza-kuya kumandla ngama-50% xa kuthelekiswa noyilo lwe-GCN. Ukongeza, itekhnoloji yenkqubo ye-7-nm yenza kube lula ukuvelisa i-crystal compact ngokufanelekileyo.
Kumboniso wasekuseni, i-AMD yathelekisa ikhadi lemizobo ye-Radeon RX 5000 enemiqathango enekhadi lemifanekiso ye-NVIDIA GeForce RTX 2070, kunye ne-Strange Brigade, imveliso ene-architecture ye-Navi ubuncinane yayiyi-10% ngokukhawuleza. Ukuza kuthi ga ngoku, akukho datha ichanekileyo malunga neendleko zamakhadi evidiyo e-AMD amatsha, kodwa abaluleke ngakumbi "umda wokuqhuba ixabiso", kuba iprosesa yegraphics yeTU106 ephantsi kwemveliso yeNVIDIA iveliswa kusetyenziswa itekhnoloji ye-12-nm, kunye nendawo yayo yekristale. malunga ne-445 mm2. Xa sithetha nje, i-AMD ine-62% inzuzo yendawo.
Ewe kunjalo, ngaphandle kokwazi ama-nuances obudlelwane bekhontrakthi phakathi kwe-AMD kunye ne-NVIDIA kunye ne-TSMC, kunzima ukwenza izigqibo ezicacileyo malunga neendleko ze-7-nm GPUs zangaphambili kunye ne-12-nm ye-GPU yamva. Nangona kunjalo, sikhumbula inkomfa yamva nje yekota ye-NVIDIA yeengxelo ezinekratshi zomsunguli wenkampani, uJen-Hsun Huang, malunga nokungabikho kwesidingo sokutshintshela kwitekhnoloji yemveliso ye-7nm. Wachaza ukuba iminikelo ekhoyo ye-NVIDIA ayihambelani ngokubhekiselele ekusebenzeni kunye nokusetyenziswa kwamandla, nangona ziveliswa kwi-teknoloji ye-12nm, kwaye akukho sizathu sokuthelekisa kunye neemveliso ze-7nm ezikhuphisanayo. Masilinde uJulayi kwaye sibone ukuba i-rhetoric yentloko ye-NVIDIA itshintsha njani emva kokukhutshwa kophononongo oluzimeleyo lwamakhadi evidiyo e-AMD yamva nje...
umthombo: 3dnews.ru