One of the weak points of the Nvidia GeForce RTX 30 was using Samsung Foundry’s 8nm manufacturing process versus TSMC’s 7nm that AMD used in its Radeon RX 6000 lineup, but now it seems like the company does not want to go back to repeat the same mistake. According to the industry sources, NVIDIA will use TSMC’s 5nm manufacturing process node, although the question is whether they will use the First Generation of the manufacturing process, referred to as N5, or its improved version N5P.
According to TSMC itself, its N5Ps, with respect to the 1st Generation of the 5nm (N5) manufacturing process, imply a reduction in power consumption by up to 10% together with a performance improvement of 5%, although TSMC may sacrifice the consumption improvements to add a further increase in performance if deemed appropriate.
On the other hand, a very reliable source in the industry, “kopite7kimi“, who correctly predicted the specifications of the entire line of graphics cards with Ampere architecture, months in advance, is not sure that Nvidia will use its Ada Lovelace architecture to the GeForce RTX 40 series. Nvidia’s strategy may hinge on the performance of competing chips, namely the RDNA3-based AMD Radeon RX 7000s, which are to be the first MCM-designed graphics chips for gamers.
Kopite claims that there is a possibility that Nvidia will launch a family of Graphics cards based on the Hooper architecture for gamers.
Finally, we add “Moore’s Law is Dead“, who says that Lovelace would be an exclusive architecture for future Nvidia SoCs like the one that would integrate the much-sought-after Nintendo Switch Pro, but they would not be available until the second quarter of 2022 (Q2 2022).