GeForce RTX 4080 Ti will be several times more productive GeForce RTX 3090
GeForce RTX 4080 Ti will be the best next-generation map – 5 nanometers and 18,500 Cuda cores
Information appeared on the network talking about the release of new video cards, the power consumption of which is much higher than that of predecessors. Experts gave different estimates, since 500 watts of consumption is a high indicator. Also, this seems especially strange when switching from the process at 8 nanometers from Samsung to 5 nanometers from TSMC. Sources confirm that the Ada Lovelace graphics chip has great potential. It should give a twofold increase in performance between the generation of 30 series and 40. We have already seen such a transition – from Maxwell to Pascal (a series of video cards GTX 900 to the GTX 1000 series). Today, no one needs a GTX 980 Ti video card, but the GTX 1080 Ti is still interesting to gamers and miners. If you believe the sources, then GeForce RTX 4080 Ti will inherit the huge potential of the GTX 1080 Ti card and revolutionize the gamer world.
With the release date of this map, good news should not be expected. The Ada Lovelace architecture will appear no earlier than the beginning of 2023. This suggests that the Ampere architecture will be the best for the next year and a half – everyone, even the giants, are experiencing a semiconductor crisis. it was the capacity deficiency that influenced such a positive output of a new tingling. Also, at about the same time, new video cards from AMD with the RDNA 3 architecture should also be published. They will have full competition of 4080 Ti, since the process time is absolutely the same – 5 nanometers.
Apparently, to achieve these indicators, it is not only a matter of a number of architectural improvements. According to some information, GeForce RTX 4080 Ti has to have very high clock frequencies. We think that the future graphics chip AD102, which will be based on 18,432 Cuda kernels can take a maximum of 2.5 gigahertz. Since the former GA 102 has only 10,752 Cuda cores on board, a 2-fold increase in power does not seem impossible.
According to sources, namely technical developers from 3DCenter, the architecture of the future AD102 will contain more than 72 clusters for texture processing – TPC, and more than 144 in-line multi-processors. This figure is 2 times the number of GPC clusters than the previous Turing architecture.
You should also understand that the graphics card is not only a performance for gamers and miners. A huge rating that affects its rating is the ratio of price to performance. For example, GeForce GTX 1080 Ti had a price of $700. Since then, prices have not been so low. GeForce RTX 2080 Ti cost $300 more, and GeForce RTX 3080 Ti peaked at $3,000. When NVIDIA makes this engineering breakthrough, we will see the best video card for the next five years.
It also remains a huge question – will NVIDIA continue to work with Samsung? Or go to TSMC? According to some reports, NVIDIA uses Samsung to sell small orders, and TSMC has already ordered 5 nanometer processors.
Follow all the news from the crypto industry with us. We publish only high-quality and interesting material. We wish you good luck and calm work!