nvidia

ROGCBE' Alıntı:

Nvidia Ampere specs
The Nvidia Ampere GPUs are confirmed to use 7nm architecture, which is already an upgrade on the larger 12nm architecture found with the current Turing graphics cards.

Further specs details are far more uncertain though.

Bağlantıları görmek için lütfen Giriş Yap

suggests the Ampere GPUs will see more vRAM, higher clock speeds and improved power efficiency.

Other sources, including

Bağlantıları görmek için lütfen Giriş Yap

and Twitter account

Bağlantıları görmek için lütfen Giriş Yap

, have been far more specific with the specs, but aren’t as credible a source. If they are to be believed though, the Nvidia Ampere specs could be as follows (with comparisons):

	Nvidia RTX 3080 Ti	Nvidia RTX 2080 Ti
Architecture	Ampere	Turing
CUDA Cores	5376	4352
Memory size	12GB	11GB
Memory Bus	384-bit	352-bit

	Nvidia RTX 3080	Nvidia RTX 2080
Architecture	Ampere	Turing
CUDA Cores	3840	3072
Memory size	10GB	8GB
Memory Bus	320-bit	256-bit

	Nvidia RTX 3070	Nvidia RTX 2070
Architecture	Ampere	Turing
CUDA Cores	3072	2304
Memory size	8GB	8GB
Memory Bus	256-bit	256-bit

GA102 - 84 SMs / 5376 CUDA cores / 12GB GDDR6 / 384-bit bus - 40% faster than RTX 2080 Ti
GA103 - 60 SMs / 3840 CUDA cores / 10GB GDDR6 / 320-bit bus - 10% faster than RTX 2080 Ti
GA104 - 48 SMs / 3072 CUDA cores / 8GB GDDR6 / 256-bit bus - 5% slower than RTX 2080 Ti

Genişletmek için tıklayın ...