Raytracing is going through a few more passes on Series S than Switch 2. Switch 2 looks a lot darker in areas because of less temporal accumulation. You can also see it in the eyes of characters. Also more noise in shadows on Switch 2. And of course a moderate reduction in geometric/mesh complexity on Switch 2. Otherwise they're pretty close. 

Obviously there are going to be some differences, the fact that the lighting engine is largely intact on the Switch 2 and has any ray tracing whatsoever is fucking bonkers on a portable 10-15 watt console in a big open world 9th gen game to begin with.

I remember a year ago even people in the optimist camp thought maybe ray tracing would be reserved for really low end Switch 2 games like basically Switch 1 tier graphics only and something you see very rarely.  

I think honestly this puts to bed whether or not this is impressive hardware or not, this is very obviously already on the high end of what people were saying might be possible for this system or even I think some people were saying performance like this would be flat out impossible. And the system is only a few weeks old and just barely getting its first wave of 3rd party games, these are the first baby steps for the console. 

I also think it may be time for some people to acknowledge Nintendo's younger hardware engineers alongside Nvidia did a heckuva job here, this ain't 2011 anymore. It wasn't as simple as looking at a chip's node process and determining performance just from that, but the fact that they're able to get this level of performance from 8nm is impressive. They must have really worked hard on this chip design and/or Samsung's process is better than its being given credit for. If Nintendo got a good price on the chip as a result, I would have to say they made a smart choice in that case. They got a good, very capable chip for a reasonable cost it looks like. 

Seems like a fair assessment.  Docked the resolution seems fair.  Based on a few videos I watched, handheld and the resolution is definitely rough.  

Behind the Seris S, and of course ps5, but for a handheld quite punchy.  But all that was largely expected.  For a handheld the S2 is really good.  

Edit

Lighting jumps out to me as the biggest difference between the two versions.

Not sure on the exact technical terms but there seems to be a difference in volumetric effects as well.  Stuff like fog and haze.  

But definitely very playable.  

This is night and day better than things like The Witcher 3 port Switch 1 Vs. XBox One.

Like that isn't even close.

The gap has closed significantly.

It's also obvious with things like 120 FPS, 4K resolution emphasis right in the marketing and even in the Nintendo Welcome Tour game ... this is a different Nintendo today. The hardware division is largely newer, younger employees, the days of Genyo Takeda are over, obviously Iwata is not there anymore either. Furukawa Era is here, better get used to it being different from the past. 

I was actually one of a few who were expecting Switch 2 to have more examples of ray-tracing than Series S, when it became clear it had a T239. Switch 2 is roughly half as powerful as Series S for raster-loads (when docked), but unit per unit Ampere GPU's perform +30-50% compared to RDNA2 in RT workloads, and the more intensive the RT load the more that gap widens. Basically was expecting developers to go for a locked 30fps in games the Series S would target 60fps but with better RT effects, etc. But I also thought at the time it would have faster unified memory, more usable ram (only 1.5GB-2GB allocated to OS), and higher CPU clocks/more CPU cores than it does. 

But as far as I am aware, this game isn't using all of the T239's RT features anyway. Like ray reconstruction isn't used for example, and that can drastically improve quality with relatively minimal performance impact. 

It was said for years here, 1536 CUDA cores right off the bat is a fuckton of CUDA cores for a device that's supposedly a borked, low performance "don't get your hopes up cuz Nintendo!" system. 

That should have right there and then been a tip off that this wasn't going to be poor hardware, but some people really really wanted to keep banging their head against a brick wall and not listen. 

Finally, I think we should acknowledge Furukawa's time as Nintendo president will likely lead Nintendo is a new and different era. Maybe not bleeding edge hardware, but certainly not afraid of relatively modern tech either. Which is actually more in line with actually Nintendo's core history in hardware under Yamauchi anyway. The NES, SNES, N64, and GameCube were all relatively good to great hardware for their time of release given that they were all designed to be relatively affordable also. 

The people who really operate under the idea of Nintendo = Iwata's time as president (2003-2015) only are ignorant and don't have the gaming knowledge they think they do. Nintendo was making game consoles for decades prior to that. 

Now that we've seen more, this is a fantastic turnout for a port of such a demanding current gen game to a handheld machine, especially so early in the system's life.

Based on the little footage we had from trailers and such, I was kinda expecting we'd be looking at an unstable framerate and much steeper visual cutbacks, so seeing it hold up so well is a great sign as far as the viability of bringing high end PS5 games to Switch 2; if this can run in a decent shape, then most games should be doable.

Not sure about that "minimal performance impact" - don't have the game, so I can't test it myself, but on Ampere it's around 25-30%.

Everything I have read and experienced on the topic has been a few % in either direction depending on the game's denoisers. 

Sometimes there is a performance uplift because multiple shader-based denoisers that consume CUDA-core resources are replaced by the RR model that utilizes tensor cores. Other times there is a penalty if the denoiser it replaces was already light-weight. 

Cyberpunk 2077 used to require DLSS SR to be active to use RR, so I can see there being a heavy performance penalty there. But they updated that IIRC?

Haven't really tested this myself, so pulling data from net, but it seems that RTX 40 & 50 are those few %. Ampere gets hit much more.