Nintendo shares first official look at the Switch 2
- Kangal
- HX6
- 07 Feb 2025
Anonymous, 07 Feb 2025The videos are about Native 4K TAA and DLSS 4. The DLSS 4 p... moreYes, I did.
Did you read my comment properly?
TAA is a filter that removes jaggies, but introduces shimmers, artefacts, and a lot of blur. It is definitely NOT a native image / quality. I agree with most people, I would rather "take the jaggies" and have a clean native image, rather than a problematic one. Or to be more balanced to go with MSAA x3 at 1080p, or 1440p Native. So it is not a fair comparison to use "Native-TAA" to compare to DLSS-upscaling. You need to use Regular Native, or at least with an AA that doesn't degrade image quality (SMAA has pro/con, MSAA basically is all Pro when it comes to image quality, while SSAA is too biased/favourable for Native image quality).
It is important to mention the different DLSS technologies. Early on it was heavily reliant on TAA and fast-quality SMAA rendering. These were calculated natively using RISC-V cores in the dGPU that Nvidia dubbed Tensor Core. Kind of like PhysX all over again. It was NOT worth using DLAA or DLSS during the early days for the above reason. They made huge improvements in v2.1 and then a decent progress with the current v3.5 and another notable upgrade with the upcoming v4 (transformer) standard. I believe they are using a pre-filter technique now, when it comes to the temporal sampling. So no more TAA !!!
From what we know, Native image is still superior. The DLAA v4 is good. And some games can/will get updates to include the newer upscaler. However, some Nvidia Sponsored games just get stuck on whatever DLSS version that came with it. So it is important to know and understand how these work. In the future, we can get AI-Based filtering that is superior to Native Rendering. I explained how in my previous post. But we are not there yet. The Nvidia cards are just too slow, bar the RTX-5090. So we're still 5+ years away. Herk we do not even have ray-path tracing done properly yet, we will have to wait for the era of the PS6/XB6 in 2028 and later, to optimise it, and bring it to the mainstream.
- ?
- Anonymous
- 70d
- 07 Feb 2025
Kangal, 06 Feb 2025That's not an equal comparison. Maybe if you were doi... moreThe videos are about Native 4K TAA and DLSS 4. The DLSS 4 produces better images and has much better performance. I don't understand why you talk about DLSS 2.1 and DLSS 3.5 when I talk about DLSS 4. The article you provided is an old article that is about DLSS 3.5.10 and older. DLSS 4 Transformer Model is MUCH better than DLSS 3.5. Have you watched the videos?
- Kangal
- Hqj
- 06 Feb 2025
Anonymous, 04 Feb 2025I don't understand what's with the moderation. Th... moreThat's not an equal comparison.
Maybe if you were doing a Native "TAA" vs DLSS v2.1 (or FSR 2). That was the "early days" and it was not a good showing for AMD or Nvidia.
What we want to see is FSR 4.0 or DLSS 3.5 (or later). That is indicative of current technology. And we want to compare to actual Native Resolution. Adding TAA, if you don't understand the technology, greatly blurs and softens the image, and itself can produce artefacts. I call it the big blunder of modern gaming. Most including myself would rather "take the jaggies" than the obvious downgrade of TLAA/TAA. If you want something a bit less drastic, there still is the old FXAA. And yes I understand most people want some sort of AA when at resolutions below 1440p. I would recommend SMAA/MSAA instead. It's a big performance hit, but at least, it is worth it.
https://www.techspot.com/article/2803-dlss-vs-native-1080p/#:~:text=When%20we%20tested%20DLSS%20across,better%20results%20with%20native%20rendering.
Now there is DLAA. That is slightly new, and basically it is Nvidia's DLSS technology but without the downscaling. So you're getting the image sharpener at the base-resolution, and it's stretched to your monitor, without the higher (AI-rendered) resolution. All games that have DLSS actually support it, but few rarely give you the option. In fact, most games don't let you choose the AA these days. It's locked to TAA setting, sometimes there is FXAA. But it is getting rarer to see option for SMAA or Disabling-AA entirely. This is more of an UE5 thing than anything else, as developers are getting lazy.
Early days of DLAA it was superior to DLSS, but not that great. And it did produce images better than Native vs Native-TAA. But this begs the question.... Is it even possible to produce BETTER images than totally rendered ones? In theory No.... the game you are playing and rendered as is, is what the game itself is. But in practice Yes! If you have enough information, you can actually know what the developers intended to show on-screen, and you can replace actual rendered images, with false/artificial image which is less accurate "in-game" but is closer to the target graphics that the developer intended. At lower resolutions, this is not possible/easy, but at 2160p or even higher it becomes easier/possible. That was the promise behind DLAA, but not what is delivered. Still it is a net positive, and I agree with it's use to improve gaming. Most games don't give you DLAA option but you can somewhat bypass that by going: eg 1440p Monitor, use 2160p DLSS Quality Mode, to get 1440p native rendering, upscaled to 2160p resolution, and your monitor then down-sampling it back down to 1440p for viewing.
What DLSS does is to get a Smaller Rendered image, reduce it's framerate/performance, upscale the image to a target resolution, and overlay a new image using DLAA to soften/sharpen as necessary. So a native 720p will be the fastest, but worst image quality. And DLSS-720p image will be 10% slower but image quality would be better due to DLAA upscaling to 1080p. Then a native 1080p image will be much slower but have great image quality. Lastly a 1080p image with DLAA (not DLSS) will be 5% or so slower, so basically the slowest, and have the best image quality. Unless you have MSAA as an option, which I highly suspect is not an option/coming back.
I would recommend you to experiment with older games. And also to read up on the different AA Technologies. Each technique has it's strengths and weaknesses. Off the top of my head: SSAA, MSAA, TxAA, SMAA, TrgAA, CMAA, TAA, FXAA. Outside of proprietary ones like DLAA / DSR, and RSI etc etc.
- ?
- Anonymous
- 70d
- 04 Feb 2025
Kangal, 04 Feb 2025Have you copium.
You just LOVE nintendo, very very biased.... moreI don't understand what's with the moderation. This is boring.
Anyway:
DLSS 4 is now BETTER than Native: https://youtu.be/5kI92zArqEA
Regarding the performance of DLSS 4: https://youtu.be/kg6wTWBFJTY
God of War Raknarok Native 4K: about 65 fps
God of War Raknarok 4K DLSS 4 Quality: about 82 fps
God of War Raknarok 4K DLSS 4 Performance: about 99 fps
The Last of Us Part 1 Native 4K: about 30 fps
The Last of Us Part 1 4K DLSS 4 Quality: about 58 fps
The Last of Us Part 1 4K DLSS 4 Performance: about 70 fps
Star Wars Jedi Survivor Native 4K: about 36 fps
Star Wars Jedi Survivor 4K DLSS 4 Quality: about 60 fps
Star Wars Jedi Survivor 4K DLSS 4 Performance: about 78 fps
Obviously, different games will have different results.
- Kangal
- Hqj
- 04 Feb 2025
Kangal, 04 Feb 2025Have you copium.
You just LOVE nintendo, very very biased.... moreSorry: the website timed out.
Here is my take, let's say I've got an RX-7900-XTX and you've got an equivalent RTX-4080. Let's make everything else identical. And say we are competing to show a certain game between us, as best as we can. You and I have different approaches: one is marketing wise, the other analytical.
You set your game to Ultra Settings, set Ray Tracing to Max Settings, and notice you're getting very bad performance. Then you lower your Resolution, set DLSS-4 to Performance Mode, and enable the Max Frame Insertion. You look at the game, it looks alright, and the framerate is reading ridiculously high.
Whereas me, I go a different route. I turn all the Settings to Medium. Then I experiment to see which ones tank the performance, and which ones actually improve the Image Quality. I end up with a Mixed Settings, mostly on High. I also do this with Ray Tracing, and notice the Minor Options don't affect performance much, and decide to leave them on. However, the Maximum Settings seem to have destroyed the performance, since these are not yet optimised. I've gone with 2160p native, and use some AA techniques that help sharpen or soften the image as necessary. It looks great. I decide to try FSR-4 and dial the resolution back to 1800p and it to Quality Upscaling mode, now it looks almost as good as before, and I've got more performance in the bank. Then I look at Frame Insertion, and notice the huge latency drop. I don't see a need for "smoothing" but I explore other options. I utilise Insertion Black-Frame, and this helps reduce flicker for certain humans. I test it out some more, and then I am satisfied.
We then come together and look at both games. Mine looks more crisp and accurate, whereas yours looks washed out. When playing the game, yours is reading 160fps but feels a bit slow. Mine is reading 120fps, and feels very fast, especially since I've considered VRR. A third-person also comes and tests both systems, we don't tell them which system is which, and they agree the AMD system is superior.
Now to top it off, my system is cheaper. I've done my research and picked up parts when they were at a good price. It's likely a 7800x3D full AMD, and doesn't look the best, but it hides underneath the desk away from things. Yours is probably an i9-14900k and is in a huge box with fancy RGB lighting. I can see the appeal. But for the money I've saved, I manage to buy an ergonomic chair, a good-value monitor, and a decent enough sound system. Your setup uses an uncomfortable "gaming" chair, an overpriced monitor, and a sound system that costs triple of mine.
You understand my point. You can replace the "you" or "me" with whatever you like, as the point made, is that this is the difference between Analytical Thinking versus Marketing Orientated. If you can understand the merits of such argument, we can move forwards, as to how this affects the topic on hand.
Which is to be open to constructive criticism, and battle of ideas, as to the capabilities of the Switch-2. I actually don't know squat because I do not have one physically in my hands to test it myself. I am going off the leaks. I have seen some "fanbois" claim that the Nvidia Tegra T925 chipset is superior to the Qualcomm Snapdragon 8 Elite. And I've seen some bashers claim it is still in the same performance category of the last-last generation. Personally I do not know. However, I postulate from those recent leaks that it is very similar in capability to the 2020-esque QSD 870 chipset (think Retroid Pocket 5). I think bumping it up to the QC 7g2+ for performance would be generous (think AYN Odin2).
PS: the Nintendo Switch-2 is basically NOT competing against the XB1, PS4, PS4 Pro, XB1X. Nor is it competing against the XSS, PS5, XSX, PS5 Pro. Nor is it competing against the Retroid Pocket 5, or AYA Neo DMG, or AYN Odin2. Nor is it competing against the MSi Claw 8+, ASUS ROG Ally, Lenovo Legion Go.
It stands alone.
The closest competitor conceivable would be the Valve SteamDeck, as that is a console, and runs a few first-party games, and has many third-party games, is the same portable form-factor, and they fall in the same price and performance bracket. Again, this is not a perfect comparison. The Valve handheld is struggling to hit the 10 Million Units sold mark, while Nintendo is poised to do x10 as much sales. The Nintendo unit will be heavily marketed, featured mainstream, and will rely on its first-party catalogue of games. So major differences there, but their core values are similar.
- Kangal
- Hqj
- 04 Feb 2025
Anonymous, 02 Feb 2025LOL. Look up DLSS 4. As I said, it's unlikely that the... moreHave you copium.
You just LOVE nintendo, very very biased. I am not for them, I am for you (consumer). That is why we are at cross-roads so often. Also you are not willing to concede when you are wrong, whereas I'm cut from a different cloth. I like when people challenge me, because it's an opportunity to learn, to teach, and expand my communication skills. I ponder that you merely get a kick out of the activity, a sort of high, by looking down on others who don't conform to your biases.
Where is the misinformation?
Well, your most recent one claiming that Nvidia's technology can somehow upscale a 720p game, into a 1440p game. And claim that it looks better than a full-1440p game rendered natively. And then somehow they will also double the framerate too!
The truth is that the latest DLSS doesn't make games look better, it makes them look worse. It's using a TLAA technique, and there are still artefacts in random areas. I will grant that it has improved notably since v2.1 which itself was a major improvement over v1.0 back in the RTX-2060 days.
I mean even Digital Foundry and their testing is flawed to a degree. They're pretty biased for Nintendo and Nvidia in their conclusions sometimes. But even they don't spew absurdities like you have. Look at multiple sources. Do your own Hands-On testing. Ask in some forums. Dig deeper.
Here is my take, let's say I've got an RX-7900-XTX and you've got an equivalent RTX-4080-Super.
- ?
- Anonymous
- 70d
- 02 Feb 2025
Kangal, 27 Jan 2025This is marketing fud. Stop spreading misinformation. Typic... moreLOL. Look up DLSS 4. As I said, it's unlikely that the Nintendo Switch 2 supports DLSS 4. Which part of my post is misinformation? I won't be surprised if you don't even know what DLSS is. It's you who always talk nonsense, and when I call you out on it, you respond by disliking my posts. You know that you're not equipped to argue with me.
- Kangal
- Hqj
- 27 Jan 2025
Anonymous, 24 Jan 2025Apparently DLSS 4 is capable of upscaling 720p images and m... moreThis is marketing fud. Stop spreading misinformation. Typical of your posts.
- ?
- Anonymous
- 70d
- 24 Jan 2025
Apparently DLSS 4 is capable of upscaling 720p images and making them look better than native 1440p images and doubling the frame rate. It's unlikely that the Nintendo Switch 2 supports DLSS 4, though. It's a given that the Nintendo Switch 2 supports DLSS 3.8.
- 1
- 1989
- CbI
- 23 Jan 2025
Kangal, 23 Jan 2025TL:DR - Performance is important because it dictates what i... moreYeah sure bro i sure i read all that bs.
- Kangal
- HX6
- 23 Jan 2025
Kangal, 23 Jan 2025TL:DR - Performance is important because it dictates what i... more^Small Correction:
I am saying that on-average the underclocked ARM Chipset in the Switch-2 will be slightly less capable than the AMD Aerith chipset in the original Steam Deck, and even slower than the Z1e/6800u chipset of the ROG Ally. There will be Nintendo games that will optimise it very well for the Switch-2 and these will surpass the handhelds mentioned above. Overall they are in the same performance category.
It's just an on-average rating, and not a definitive one. Just like how on-average the XSX is faster than the PS5, but not from a definitive point. Some studios do spend extra money and time to optimise the PS5 version and it has higher performance than the XSX version.
Overall, I think my thoughts are clear about the systems. I am not a hater, I am an enthusiast. See the TL:DR that is listed above. Now go and start gaming :D
- Kangal
- HX6
- 23 Jan 2025
1989, 22 Jan 2025Bad as in low fps under 15w, yeah. Its just a repurposed la... moreTL:DR - Performance is important because it dictates what is possible. What is important is that the system you like supports the games you want. That is all. There is no need to bicker about which is better, because what is better for me is different to what is better for you. Just find the games, the system, and have fun!
Rant:
An old friend of mine does this full-time. Check out his channel ThePhawx on YouTube. The original Valve Steam Deck was great, but it got better with updates. The original ASUS ROG Ally was even better, and it seemed it would be like that, but it also got some updates and got better.
The new SteamDeck OLED uses 2020 technology, but is very very optimised and runs decent even today. The newer ASUS Ally X uses more like 2022 technology, and is decently optimised, and it's even better (try Bazzite).
Neither can quiet match the underclocked (Tegra X1) ARM chipset in the NSwitch OLED for battery life, and they likely won't match the Switch-2 in that regard too. Credit where due, well done Nintendo on the weight and battery life. Price wise we are expecting a USD $400 for the Switch-2 which is roughly same as the $350 NSwitch when adjusted for inflation. No details for the screen, but it seems it has not been upgraded to 1080p or OLED, and they are sticking with a cheap 720p LCD.
But I would NOT say the Switch-2 can match the performance of the SteamDeck, and even less-so for the ROG Ally. The original SteamDeck was gaming a bit above PS4-level back in 2022 (launch delayed due to pand emic), and the RoG Ally is pushing out nearly PS4 Pro levels in 2024. Important to note that both systems combined have sold 10 Million Units or less, so these are less mainstream and more niche products. Sidenote the Switch-2 is targeting 100 Million Units sold, which is very impressive, even if it is lower than the NSwitch with 144 Million Units sold roughly counting. Based on the accurate leaks, the Switch-2 does not seem impressive from a performance point. That 2018 technology of the RTX-2000 and Samsung-8nm is really going to hold it back closer to the Xbox One levels. We should get some "super" optimised games pushing it slightly past PS4 levels. But all those people online that claimed PS4 Pro level were wrong. Comparing with Teraflops is wrong. If you don't agree with the technical assessment, just wait, and see the performance after the Switch-2 release. I'm willing to bet money. I am not one to gatekeep, I just want the system to be powerful enough to support third-party games, in particular like current or upcoming games. There is only so much joy you can have from old games ported to new system (eg RDR1 on NSwitch).
I don't think we will be getting an Xbox Handheld or a Portable PlayStation based on their current objectives. But there does seem to be a good probability for a SteamDeck-2 and ASUS ROG-2 maybe in 2027. So yay for competition. Those could be as good as a 2nm-TSMC silicon, Zen6 CPU, and RDNA-5 iGPU, for a speculative performance that is roughly equal to a PS5 at the 10W-20W thermal limit. At the worst (unlikely), they might go with 4nm-TSMC, Zen4+, RDNA-4 technologies, and require that 20W thermal limit to fall inbetween the XSS and PS5 performance. I guess the only "problem" is that the next-gen consoles of the XB6 and PS6 are inline to be released around the same time in 2027-2028 and likely use the same technology just at higher 200W thermals. But I imagine that a lot of next-gen games will get ported to the SteamDeck-2 with lower fidelity in such a situation, like what happened with Elden Ring for SteamDeck. At least until the late-age of the console lifecycle as the performance gap becomes too large to ignore. There would be exceptions to the rule, such as the "impossible ports" like Witcher 3, ARK, and Doom Eternal on the NSwitch.
Conclusion?
The reason why performance is important is because the 2017 NSwitch was stuck with old-gen titles from the 2005 Xbox 360 era. The new 2025 Switch-2 looks like it will be stuck with last-gen titles from the 2013 Xbox One era. The niche will be covered by the SteamDeck and ROG Ally, and not a Portable PlayStation or Xbox Handheld. The next handhelds from Valve and ASUS seem potent and likely to happen, and based on the roadmap from AMD, should match the current-gen PS5 consoles. The next-gen consoles of PS6 and XB6 will release around the similar time, which could pose an interesting challenge for developers, but still a win for consumers.
- 1
- 1989
- CbI
- 22 Jan 2025
Anonymous, 22 Jan 2025It's not a bad chip per se. It's just not efficie... moreBad as in low fps under 15w, yeah. Its just a repurposed laptop chip afterall.
- ?
- Anonymous
- 70d
- 22 Jan 2025
1989, 21 Jan 2025If switch 2 really is on 8nm then closer to 400 or if its a... moreIt's not a bad chip per se. It's just not efficient enough for handheld gaming devices. Ideally, the Asus ROG Ally should outperform the Steam Deck at any W. In reality, the effective 4.3 TFLOPS seems to be only for synthetic benchmarks and requires much power, which will make battery life a tragedy.
- 1
- 1989
- CbI
- 21 Jan 2025
Anonymous, 21 Jan 2025We'll know soon enough. If it's $399.99, the game... moreIf switch 2 really is on 8nm then closer to 400 or if its a better node closer to 500, thats what i think. Atleast they can justify it with a big battery life then.
Z1e is just bad at low watts, below 15 its much worse than deck.
- ?
- Anonymous
- 70d
- 21 Jan 2025
1989, 21 Jan 2025Yeah you didnt check, thats on you. The 12$ ugreen dock doe... moreWe'll know soon enough. If it's $399.99, the game will probably be over for all handheld gaming PCs. It will be a big win for consumers, though. The effective 4.3 TFLOPS of the AMD Ryzen Z1 Extreme is for marketing, by the way. In theory, the performance of the Asus ROG Ally should more than double that of the Steam Deck. In reality, it will probably take 30 W and minimal CPU load to reach that. The problem with this APU is that it's not efficient enough for handheld devices. At 15 W, it's only on par with the APU of the Steam Deck. This means that if the Steam Deck reaches 1.6 TFLOPS at 15 W, the Asus ROG Ally also reaches about 1.6 TFLOPS and at about 15 W. So much for the effective 4.3 TFLOPS.
- 1
- 1989
- CbI
- 21 Jan 2025
Anonymous, 20 Jan 2025I know for a fact that the Lenovo Legion Go in my region is... moreYeah you didnt check, thats on you. The 12$ ugreen dock does 100w. Getting another gamepad to play when docked is easy and cheap too. Anything you might have lying around will work with windows anyway. I think its gonna be closer to 400 than 500 though.
- ?
- Anonymous
- 70d
- 20 Jan 2025
1989, 20 Jan 2025See, you didnt even look at the price you can buy a z1e dev... moreI know for a fact that the Lenovo Legion Go in my region is over $700. IIRC, when I checked the price of the Asus ROG Ally in my region in the past, it was about $700. I didn't know that the Asus ROG Ally is discounted and is only over $500 in my region now. A dock is also for charging, by the way. A cheap dock probably won't provide fast charging. Anyway, this doesn't invalidate my point. The Nintendo Switch 2 has detachable Joy-Con controllers and comes with a Joy-Con grip and a dock. It should have better battery life than that of the Asus ROG Ally and the Levovo Legion Go. $499.99 isn't expensive for what it is. As I said, I'm not against $399.99, but in my opinion, that price is unlikely.