I-CEO ye-NVIDIA uJensen Huang ubonise amakhadi evidiyo exesha elide elindelwe kwisizukulwana esilandelayo ekhitshini lakhe. Njengoko bekulindelekile, izisombululo ezindala zabhengezwa namhlanje: I-GeForce RTX 3090, i-GeForce RTX 3080 kunye ne-GeForce RTX 3070. Amakhadi evidiyo akhiwe kwi-Ampere isizukulwana se-GPU esenziwe ngokusebenzisa iteknoloji yenkqubo ye-Samsung ye-8nm, ngelixa i-Turing isizukulwana sangaphambili saveliswa ngokusebenzisa iteknoloji ye-12nm TSMC.
GeForce RTX 3090
Ukutshintshwa kwenkqubo entsha, encinci yezobugcisa yenze ukuba kube lula ukubeka ii-transistors ezininzi kwi-chip, ngenxa yokuba inani leebhloko ezisebenzayo landa kakhulu. Ngelishwa, uJensen Huang akazange achaze ulungelelwaniso lwe-GPU, kwaye wathiya kuphela iiparamitha zenkqubo esezantsi yememori phakathi kweempawu. I-GeForce RTX 3090 ixhotyiswe nge-24 GB yememori ye-GDDR6X enesantya esisebenzayo sewotshi ye-19,5 GHz. Kanye kunye ne-384-bit yebhasi ebanzi, oku kunika i-memory subsystem bandwidth ye-936 GB / s. I-GeForce RTX 2080 Ti, khumbula, yayine-11 GB yememori ye-GDDR6 kuphela malunga namaxesha anesiqingatha esingaphantsi kwe-bandwidth.
Incopho yokusebenza kwe-GPU yethiyori yekhadi elitsha levidiyo yi-36 TFLOPS "ngokwakhiwa komfanekiso wendabuko usebenzisa i-shading (shaders)" (Shader-TFLOPS). Ukusebenza xa usebenza nge-ray tracing kuchazwe kwi-69 TFLOPS (RT-TFLOPS), kunye nokusebenza kwimisebenzi usebenzisa i-tensor cores ifikelela kwi-285 TFLOPS (Tensor-TFLOPS).
Ngokutsho kwe-NVIDIA, eli khadi levidiyo lenzelwe ukudlala kwisisombululo se-8K, apho ikwazi ukubonelela nge-60 FPS ezinzileyo. Kwakhona kuphawulwe ukuba le "rhamncwa" malunga ne-50% ngokukhawuleza kwisisombululo se-4K kune-Titan RTX.
GeForce RTX 3080
Intloko ye-NVIDIA yabiza i-GeForce RTX 3080 ikhadi levidiyo i-flagship entsha, idlula umbuzo wokuba yintoni i-GeForce RTX 3090 kule meko, nangona kunjalo, i-GeForce RTX 3080 ibonakala "njengeflegi," zombini imbonakalo kunye nokusebenza.
Phakathi kweempawu, ubukho be-10 GB yememori ye-GDDR6X ene-frequency esebenzayo ye-19 GHz. Imemori ixhunyiwe ngebhasi ye-320-bit, ekugqibeleni ibonelela ngokugqithiswa kwe-760 GB / s. Masikhumbule ukuba i-GeForce RTX 2080 yangaphambili yayine-8 GB yememori ye-GDDR6 kuphela kunye ne-bandwidth ye-448 GB / s.
Ukusebenza kweGeForce RTX 3080 kunye nokunikezelwa kwendabuko usebenzisa i-shaders yi-30 TFLOPS (Shader-TFLOPS). Xa kusetyenzwa i-ray tracing, ikhadi levidiyo libonelela nge-58 TFLOPS (RT-TFLOPS), kwaye ekusebenzeni kwi-tensor cores inikezela ukuya kwi-238 TFLOPS (Tensor-TFLOPS). Ngokwentloko ye-NVIDIA, ikhadi levidiyo le-GeForce RTX 3080 linemveliso kabini kunangaphambili, iGeForce RTX 2080.
GeForce RTX 3070
Ikhadi elilula kakhulu le-trio enikezelweyo, i-GeForce RTX 3070, njengangaphambili i-GeForce RTX 2070, ine-8 GB yememori ye-GDDR6.
Umgangatho wokusebenza we-GeForce RTX 3070 i-accelerator ye-graphics accelerator yayingu-20 TFLOPS kunye nokunikezelwa kwendabuko ngee-shaders (Shader-TFLOPS). Ukusebenza xa usebenza nge-ray tracing kuchazwe kwi-40 TFLOPS (RT-TFLOPS), kunye nokusebenza kwimisebenzi kunye ne-tensor cores ifikelela kwi-163 TFLOPS (Tensor-TFLOPS). I-NVIDIA ibanga ukuba iGeForce RTX 3070 entsha inemveliso ngakumbi kuneflegi yangaphambili - iGeForce RTX 2080 Ti.
Eyokuqala yeemveliso ezintsha iya kuba yiGeForce RTX 3080. Oku kuya kwenzeka kule nyanga-Septemba 17. Iindleko ziya kuhlala kwinqanaba langaphambili - i-$ 699 (eRashiya - i-ruble ye-63). I-GeForce RTX 500 efikelelekayo ngakumbi iya kubonakala kwiishelufu ngo-Oktobha, kwaye ixabiso layo liya kuba yi-$ 3070 (i-ruble ye-499) - kwakhona, kwinqanaba langaphambili. Ekugqibeleni, i-GeForce RTX 45 iya kuthengiswa ngoSeptemba 500, ifumaneka ngokukhawuleza kwi-NVIDIA kunye namaqabane ayo, ngexabiso ... $ 3090 (24 ruble). Bekungekho kunyuka kwamaxabiso apha.
umthombo:
umthombo: 3dnews.ru
