NVIDIA and source manufacturers faced by the RTX 40

If there is something that is clear, and it is not something that we like, it is that the energy consumption of tower PCs is going to rise in the coming years. And it is that despite the fact that the consumption performance of the new architectures is better than the previous one, the different manufacturers have decided to pull up in terms of electrical power. The last? Apparently the NVIDIA RTX 40 and power suppliesat least some models, would not get along with each other.

One of the dilemmas for graphics hardware designers at NVIDIA has been the following: maintain consumption and increase performance by 50% or double it by increasing consumption. This has led them to bet on the new 12+4 pin connector. The associated problem? We will have to buy a new source for the top-of-the-range RTX 40 and apparently there are problems between both parties.

The NVIDIA RTX 40 could have problems with certain sources

One of the biggest challenges facing future graphics cards are those that will use the 12+4-pin PCI Express connector. At the moment, of the three manufacturers of this component and officially, it seems that only NVIDIA will bet on this connector while waiting to receive information from AMD and Intel. Specifically, it will do it with the AD102 chip, the most powerful used in its RTX 40 and apparently, they are having some friction with power supply manufacturers.

We have to start from the fact that a graphics card takes the electrical energy to work from the PCI Express port itself, which provides 75 W of power in combination with a double or simple 6 or 8-pin connector. The former provides an additional 75 W and the latter 150 W. So so far the graphics cards consume 375 W at most. However, cWith the arrival of the new 12+4 pin connector, consumption can go up to 675 W with a single connector or if it is double up to 1275 W. Although at the moment there is no graphics chip or video memories that excessively support such consumption.

This translates into new challenges, especially in the relationship between the computer’s power supply and the graphics card. And it is that, apparently, one of the reasons why the RTX 40 would have been delayed is for he misunderstanding with font manufacturers feeding and NVIDIA

The problem of consumption peaks

There are times when a graphics card can consume twice what it normally consumes. These are small spikes with a few milliseconds duration, but at the same time they accelerate the increase in temperature and that can lead to the dry stop of said card in the event that we have a power supply of poorer quality.

Pîcos Consumption Graph

To understand how it works, we must start from the fact that there is a maximum safety temperature, which is lower than that supported by the component with the least heat tolerance on the card. When a very high consumption peak suddenly appears, then the security system interprets that the upload speed has accelerated and causes a dry stop.

The Consumption spikes occur when there is a sudden change in voltage or amperage. In the case at hand, it is the amperage that rises and, therefore, the current. so heThe power supply must have the capacity to withstand such sudden surges. Which last very little in time, since they can last at most thousandths of a second and in the best of cases millionths.

What does this mean for the most powerful RTX 40?

Well, something that many of you may have realized a long time ago is that it will not be possible to install a top-of-the-range graphics card that uses the new PCIe Gen 5 connector on power supplies that do not provide a minimum power of 1000 W or even more.

However, consumption peaks can be mitigated by using quality components in power supplies that allow us to use power supplies that do not consume such high amounts. This will be key so that the electricity bill does not rise, of course, these components are expensive and many power supplies, no matter how much they give the necessary power, can end up on a blacklist by NVIDIA, which has caused a internal controversy in which both parties are blaming each other.

Keep in mind that if the RTX 4090 suddenly starts crashing with sporadic crashes, then that will be a problem for NVIDIA and their custom card manufacturers and quite a serious one at that. Hence, what seems like a minor problem may have become a nightmare for Jen Hsun Huang.

Who is right, NVIDIA or the font manufacturers?

Once we have already commented on the problem, we are going to go to the information that has apparently been leaked in various media in recent weeks. Currently NVIDIA would have finished several of the Founders Edition of its future RTX 40 based on the AD102 GPU and, therefore, on the most powerful GPU in the family. Well, the fact is that the RTX 40 would be testing them with different power supplies with a 12+4-pin connector.

RTX 40 900W

What has been the surprise? Without disclosing marks, NVIDIA has apparently found that during peak consumption the card suddenly stops working. This causes crashes in games and what is worse, it can make your PC’s power supply useless if it uses poor quality materials. Which has made them prepare a list of sources that are recommended. On the other hand, this has caused the different PSU manufacturers to throw their hands on their heads. Many of them are already in manufacturers’ inventories and others in stores. So it is a huge problem. The way to fix it? Apparently, the RTX 40 would have been delayed for at least a month for this reason.

In any case, do not be surprised to see some models of the RTX 4090 that are sold in a pack together with the recommended power supply for the system. Especially with some manufacturers of custom graphics cards, since they are also PSU manufacturers and it is with them that they will be able to test them with total safety.