In autumn, the fight for the performance throne in the graphics cards will enter the next round – in all likelihood.
In all likelihood, it will be that time again in autumn: two years will have passed and we expect genuine new graphics cards from both Nvidia and AMD in the form of Geforce RTX 4000 and Radeon RX 7000 respectively.
Much has already been reported about the likely specifications for both model series and there is actually little doubt where the journey is headed. Nvidia relies on more cores, AMD also, but should also bring a so-called chiplet design with it and competitor Nvidia may be able to outsmart it.
Radeon RX 7000
AMD is said to have a decisive advantage over Nvidia
However, the latest reports see Nvidia at an advantage again. Known for his accurate and often accurate information, leaker Moore’s Law Is Dead tweeted:
link to Twitter content
Nvidia apparently uses better production
Nvidia Lovelace, which we will probably see under the name RTX 4000, should not rely on a 5-nanometer process, as was previously assumed, but switch directly to 4-nanometers.
To avoid confusion here: TSMC’s 4N is not a completely new so-called technology node, as would be the case with N4, but an improved 5-nanometer process. Also very important: The nanometer specifications no longer speak of actual structure widths, but are merely names. Nevertheless, production in 4N promises advantages in terms of energy efficiency and performance of the Lovelace GPUs.
AMD, on the other hand, seems to continue to rely on a mix of 5 and 6 nanometer processes to put at TSMC. The chiplets with the computing cores are to be manufactured in 5 nanometers, the I/O chip in 6 nanometers.
In the current generation in the form of the Radeon RX 6000, AMD has once again caught up with Nvidia in terms of performance in the high-end segment (raytracing excluded). This is mainly due to the significantly higher clock rate (about plus 300 to 500 megahertz) that the RX 6000 can drive compared to the RTX 3000. One reason for this is that Nvidia RTX 3000 is not manufactured in a 7 nanometer process at TSMC like AMD, but in 8 nanometers at Samsung. The process is simply not as mature as that of TSMC.
RTX 4000: Better efficiency and higher clock speeds
With RTX 4000, Nvidia not only seems to want to close this gap, but maybe even add a little something on top. At the same time, this could counter AMD’s rumored change to the chiplet design and somewhat reduce the problems with the high power consumption.
With the new generation of graphics cards, a major leap in performance is in the offing. Previous leaks and rumors speak of a theoretical increase in performance of 200 to 300 percent. Experience has shown that this cannot be applied one-to-one to the performance in games. But some of that should still stick in the form of more frames per second.
What do you think? Will Nvidia put AMD back in its place, or do you expect AMD to overtake Nvidia? And what about ray tracing – can AMD follow suit here? Feel free to write it in the comments!
