Ngokusho kwamahemuhemu, isizukulwane esilandelayo se-NVIDIA GPU sizobizwa ngokuthi i-Ampere, futhi namuhla i-WCCFTech yabelane ngengxenye enkulu yolwazi olungekho emthethweni mayelana nalawa ma-chips namakhadi wevidiyo asekelwe kuwo. I-NVIDIA kubikwa ukuthi yabelane ngolwazi olulandelayo nabalingani bayo, ngakho kufanele ithembeke impela.
Into yokuqala i-NVIDIA ehlela ukugxila kuyo ngama-Ampere GPUs wukulandelela ama-ray. Le nkampani ithembisa ukuthi amakhadi wehluzo we-GeForce RTX 30-series azohlinzeka ngentuthuko enkulu ekusebenzeni kokulandela umkhondo uma kuqhathaniswa nezixazululo zamanje ze-GeForce RTX 20-series. Ama-RT cores anesibopho sokulandelela ekwakhiweni kwe-Ampere azokhiqiza kakhulu futhi asebenzise amandla kakhudlwana, futhi kuzoba nje amaningi uma kuqhathaniswa ne-Turing.
Futhi ekwakhiweni kwe-Ampere, i-NVIDIA ifuna ukuthuthukisa ukusebenza kwe-rasterization. I-NVIDIA sekuyisikhathi eside inaka kakhulu le ndawo, ngenxa yokuthi ama-GPU ayo avame ukuba ngaphambi kwezixazululo ze-AMD lapho ecubungula i-geometry eyinkimbinkimbi. Ekuqaleni, ukugcizelelwa ekusebenzeni kwe-rasterization kwafakwa kuma-accelerator e-Quadro, kodwa manje amakhadi e-GeForce abathengi angathola ukuthuthuka okubalulekile kule ndawo.
Kuyaphawulwa ukuthi inkimbinkimbi yemidlalo yomhlaba iyakhula futhi ukusebenza okuthuthukile kwe-rasterization kuzovumela isizukulwane esilandelayo i-NVIDIA GPU ukuthi isebenze nabo ngempumelelo kakhudlwana. Sekukonke, kokubili i-rasterization nokulandelelwa kwe-ray kuzoba okubaluleke kakhulu emidlalweni ngemuva kokukhululwa kwama-consoles esizukulwane esilandelayo, ngakho-ke i-NVIDIA cishe ihamba ngendlela efanele.
Umthombo futhi uhlinzeka ngolwazi mayelana nezici zamakhadi wevidiyo wesikhathi esizayo, nakuba kuphela ngokwemibandela evamile, ngaphandle kwezinombolo ezithile. Okokuqala, kubikwa ukuthi i-Ampere GPUs izoba nebhafa yozimele enkulu uma iqhathaniswa ne-Turing. Okusho ukuthi, inani lememori yevidiyo lizokhula.
Okwesibili, ukushintshela kubuchwepheshe benqubo ye-7 nm (7 nm EUV) kuzonyusa imvamisa yama-chips cishe ngo-100–200 MHz. Futhi, ngenxa yokushintshela kubuchwepheshe benqubo obuncane, ama-GPU e-Ampere azosebenza ku-voltage ephansi, cishe ngaphansi kuka-1 V. Lokhu kungase kunciphise amandla okuba ngaphezulu kwama-chips. Kodwa ngasikhathi sinye, lokhu kuzophinde kwandise ukusebenza kahle kwamandla kwamakhadi amasha evidiyo.
Futhi ekugcineni, kubikwa ukuthi izindleko zamakhadi wevidiyo we-NVIDIA asekelwe ku-Ampere GPUs zizofana namakhadi wevidiyo asekelwe kuma-chips we-Turing. Kungenzeka ukuthi izixazululo ezindala, njenge-GeForce RTX 3080 kanye ne-RTX 3080 Ti, zingabiza kancane kunezandulelayo. Kodwa-ke, kubalulekile ukuqaphela ukuthi kusesekuseni kakhulu ukukhuluma ngezindleko, ngoba izici eziningi zingathonya. Amakhadi evidiyo esizukulwane se-Ampere kufanele akhishwe ngonyaka ozayo.
Source: 3dnews.ru