Both AMD and Nvidia’s existing graphics architectures are likely to be manufactured on a 5nm TSMC process and, independently of it, will be optimized to improve power consumption / performance. However, even with regard to the performance shift on the one hand and the limited decrease in consumption of the 5nm TSMC process on the other hand, it cannot be expected that the consumption of the new top models will be similar to the current high-end.
The information is so far strongly fragmented, but practically all fragments agree that the energy requirements of these products will be unprecedented. Leaker Greymon55 indicated by flame icons that the high-end built by AMD RDNA 3 will have high energy requirements and high-end built on Nvidia Lovelace even higher.
There was also an indication that it would be a top model with Nvidia architecture Loveleace / GeForce RTX 4090 could have a consumption of up to 600 watts, which, however, still needs to be taken with a grain of salt. On the other hand, if we took Greymon’s contribution literally (or rather illustratively), then 600 watts for the GeForce RTX 4090 would mean 400 watts for the Radeon RX 7900 XT (?).
The GeForce RTX 4090 is expected to perform approximately 2.0-2.2 times higher performance than the current GeForce RTX 3090 and the Radeon RX 7900 XT (or how the top model with RDNA 3 will eventually be christened) could offer performance. to 2.5-2.7 times higher than the Radeon RX 6900 XT.
According to current information, the GeForce RTX 4090 will be built on a monolithic GPU and the release is expected plus or minus in the winter of 2022/2023. The Radeon RX 7900 XT comes out of the GPU Navi 31 composed of 2-3 chiplets (depending on whether the active interposer counts as a chiplet) and release is expected around the fall of 2022. In both cases, these are 5-6 quarters distant products. It still takes quite a while for the release, during which the targeted performance and release date can be slightly adjusted.
–
