No less than 400 watts, a figure that many fear, since it would represent a power supply of an amount of watts and of a quality that they cannot afford. This is the scenario that we are going to see in less than a year according to rumors and as such we have to investigate all the reasons we know today for trying to put white on black. Why is consumption not reduced if a new node opens?
TSMC 5nm node is not as good as they paint it, Intel has a chance
The truth is that AMD and NVIDIA are going to bet everything on the new 5nm of the Taiwanese because of the improvements that this new lithographic process of invoice brings, already with EUV to its credit. If the logic prevailed that consumption would return to the normality of 250-280 watts per card, reality will show the opposite.
There are two main reasons for both this and what is to come, at least to a greater or lesser extent in the years to come. Intel, TSMC, and Samsung are both focusing on specific areas of nodes due to the move from FinFET to GAA, which poses a challenge and, therefore, some improvements are needed in areas that do not prioritize the efficiency.
GPU consumption at 5nm, better node does not mean better consumption
To be precise, the Big Three are battling two issues and flaws:
- The new lithographic processes are not oriented to obtain a reduction in consumption or simply to maintain a consumption at the same gain as the space obtained. The density obtained by the reduction of the transistors is not equivalent or equitable with the consumption per mm2. The problem here is that AMD and NVIDIA are going to have to make chips without the so-called “dark silicones”, that is, without parts of the silicon that are not usable, so they need every mm2 at the bottom. edge of each die to introduce the greatest number of transistors.
- And here’s problem number two: competition between AMD and NVIDIA. Unlike other generations, AMD is now a rival to NVIDIA, so neither of them is going to keep anything up their sleeve and they are going to put whatever is available into their new architectures, to the limit. Such a sudden material leap (+ -70 / 80% more transistors
Therefore, Navi 31 and AD102 are going to change the paradigm again and the global GPU market, the issue for the user (price and availability aside) is going to be the PSU requirements for some monsters that are going to play. watts. Will we need sources of 1000 or 1200 watts to keep them at bay? Will the boards suffer from problems again due to the fast and aggressive spikes due to the power supplies not being ready for it?