How Will be Switch 2 Performance Wise?

[HoloDust]

Biggerboat1 said:

HoloDust said:

Fun fact - 4050 mobile vs 3050 mobile (2560 version, so both with same 2560:80:32:80:20 config) is some 30% faster on average for same power draw.

That's what 4nm for 4050 vs 8nm for 3050 does, basically allowing for much higher clocks in 4050. So yeah, it would be really nice if Nintendo went with smaller node, which is not very likely.

Exactly. If the OG Switch had been 30% stronger then perhaps I and others wouldn't have held off on purchasing TOTK & other software and Nintendo would have had more money on the back-end to offset the higher initial investment (not saying that they had an option to achieve this with OG Switch, just pointing out S2 could suffer a similar fate).

My fear is that DLSS may be too expensive in a lot of situations for S2 (this was touched on by DF in their Cyberpunk analysis), which would be a major bummer...

30% more grunt would have given them the extra headroom and then some.

Yeah, those 30% would be really nice...I know many people won't like this, but when you do napkin math by specs alone for RTX 3000 series, and compare that to actual RL benchmarks, it's really, really close. So that same napkin math says that cut down, low clocked 3050 mobile in SW2 is above PS4, but not by as much as some might think. Of course, much newer architecture and DLSS helps a lot, but they could've gone with 4nm for quite a bit better performance, and charge the same $450, albeit for lower profit - that is, alas, not what Nintendo is about.