Three will currently be in charge of bringing the new NVIDIA gaming GPUs to life for this 2021: RTX 3060 Mobile, RTX 3070 Mobile and RTX 3080 Mobile. Early comparative data shows numbers as high as double digits, which is good enough considering that Ampere has significantly reduced the number of Shaders compared to its desktop versions.
NVIDIA RTX 30 Ampere Mobile: this is the estimated performance compared to Turing
The leak is very clear in this regard, and while those numbers are under 3DMark and with several of its tests, the reality in games shouldn’t be too out of tune with what’s on offer above. As we can see there will be quite a difference in performance if we compare the desktop RTX 3080 with the Mobile, where the average is around a 24%
The RTX 3070 also marks the distance, but much less than that of its older sister. And it is that between the two models of GPU there would be on average only one 11% of GAP and the Mobile version would not be far from the performance of the desktop RTX 2080 SUPER, which already tells us a lot about the progress of NVIDIA in notebooks of this Ampere range.
Even the RTX 3060 Mobile would give the war, since it would be ahead of any desktop RTX 2060 SUPER and above the RTX 2070 SUPER Mobile, of which there is nothing. In general, this is a fairly important step forward if you take into account the consumption demonstrated by NVIDIA.
What average performance can we expect on Turing Mobile GPUs?
All in all, improving these RTX 3000 Mobile would be the worst case scenario of 7%, while the best case scenario is over 30%, the architecture average being around 18%, will of course depend on the title chosen. . .
This is based on the RTX 2080 SUPER Mobile, as a benchmark to try, i.e. the RTX 3060 Mobile wouldn’t be far from the first in most games. Of course, this is something that must be very nuanced and that has its logic.
And it is that NVIDIA gives great freedom to manufacturers in the Max-P and Max-Q models in terms of TDP. Adjusting each card’s TDP for each laptop and its cooling is critical to ensuring the integrity of the chip and the equipment itself, but this will determine performance, as frequencies will vary.
Of course, this is not all new as such, because in Turing it was done in the same way, so the averages are also valid, as they were made with the stock specs that NVIDIA marks in all. the cases, but when buying a model we will have to be careful with this.