A few days ago, some information emerged focused on the GeForce RTX 40 and Radeon RX 7000 that ensured that both will use the TSMC 4N node, and unfortunately this gave rise to many incorrect interpretations that assumed that we were talking about 4nm. Not true, both series will use TSMC’s 5nm node.
I know what you are thinking, why do we talk about node 4N then? Well, it’s very easy to explain, because that nomenclature actually refers to an improved 5nm node, that is, to an optimized version of TSMC’s original 5nm node. This is not new, more than one semiconductor manufacturer has used this approach on more than one occasion to create more interesting products for their customers, but they continue to generate misunderstandings among consumers.
Semiconductor manufacturing nodes, or manufacturing processes, whatever you want to call them, play a very important role because the most advanced are precisely the smallest. Thus, the 7nm node is more advanced than the 10nm node, and the 5nm node is also above the 7nm node. Downsizing allows for more powerful and more efficient chips to be created, and also reduces their impact on the wafer, as they end up taking up less space.
That simple explanation allows us to understand the stir that had been formed with that information. To think that the GeForce RTX 40 and Radeon RX 7000 could use a 4nm node it was a significant leap from the 5nm nodebut in the end everything has been a mistake resulting from an erroneous interpretation, as we have already mentioned.
GeForce RTX 40 and Radeon RX 7000: What does it mean to use a custom 5nm node?
It means that both lines of graphics cards will use a custom manufacturing process tailored to your needs. This adds a layer of specialization that, in the end, can have very positive consequences both in terms of performance and optimization, and get closer to the benefits that we would obtain by using a smaller node.
However, we must bear in mind that this improved 5nm node may not be present in all GeForce RTX 40, and not in all models of the Radeon RX 7000 series. According to some sources that I have been able to consult, both NVIDIA and AMD they could reserve that node for high-end models, and resort to other less advanced nodes for mid-range and low-end models.
In the case of the GeForce RTX 40, it is said that only the GeForce RTX 4080 and GeForce RTX 4090 will use that node, while in the case of AMD this will be limited to the Radeon RX 7900XT and Radeon RX 7800XT, which will be precisely the ones that will use a modular GPU with two chiplets. The rest of the models will be manufactured under the 6nm node, at least in theory. Hopper, however, is manufactured in 4nm, as we told you at the time in this article.
