Kataploom
Gold Member
It runs the same as PS5, Switch 2 is a beast!
Did the Switch 2 graphics capabilities meet your expectation?
- Thread starter LordOcidax
- Start date
- Poll
SweetTooth
Gold Member
WOW, by double the wattage you crawl above Steam deck. 8.6 TFlops APU vs Steam deck 1.6 TFlops
Good job!
Exactly like I said. Rog ally X is the representation of the wrong direction PC handhelds are taking. Trying to brute force its way with stupid ass big APUs that are bottlenecked and inefficient.
Its typical PC environment efficiency, imagine a Switch 2 with those specs? Console optimization will drive it even further but alas .. Nintendo opted for parts so cheap and off shelf they are on par with SD.
Zathalus
Member
ROG Ally X is not 8.6 TF, I'm not sure why that myth gets repeated. VOPD is useless for gaming, and the advertised GPU clock is almost never reached. At 25w it is 3.5 TF (2.3Ghz average clock), and at 15w you can expect around 1.69TF (1.1Ghz average clock). Memory bandwidth is only 17% higher vs the Deck, but in reality is around the same due to the Ally having 8 CPU cores using more memory bandwidth. If you install Bazzite and lock to 15w/17w then the Rog Ally X also has better battery life vs the Deck.
The Rog Ally X is engineered just fine. It's just a larger chip that needs 17w or more to start seeing meaningful performance lifts vs the Deck.
The Rog Ally X is engineered just fine. It's just a larger chip that needs 17w or more to start seeing meaningful performance lifts vs the Deck.
Last edited:
GigaBowser
The bear of bad news
games are nort optermized rog alley switch 2 performence will win
rodrigolfp
Haptic Gamepads 4 Life
In handheld mode is already losing if it is really 540p/25fps not ultra settings...
GigaBowser
The bear of bad news
Street Fighter 6 Runs Better on Nintendo Switch 2 Than on PS5
Good news for Street Fighter 6 fans. It looks like the Nintendo Switch 2 version runs better than Sony's consoles.
www.gfinityesports.com
Ainem Enamas
Member
Look at how great TOTK looks on S2
Buggy Loop
Gold Member
The DF marketing reel pixel count again? Rich said its from the shot where it slaps a video of cyberpunk playing and onto a render of switch for marketing purpose that they got that count. It's meaningless and even he questions it. But here we are. DF being that fucking dumb to drop resolutions without context is now ammo repeated constantly while the game is said to have handheld 1080p at launch.
Dr. Wilkinson
Member
Was expecting a generational leap over the original Switch. We clearly more than got that, it's not even a question.
rodrigolfp
Haptic Gamepads 4 Life
Epic failed marketing then if gave that impression.
Buggy Loop
Gold Member
I mean in any common sense you would not pixel count a render on a CG handheld. WTF is this. Why even do it. What's next, pixel count peoples holding an handheld on the roof commercial?
rodrigolfp
Haptic Gamepads 4 Life
540p makes some sense as Zelda TotK is 1440p and Cyberpunk is supposed to be much more demanding to run and is 1080p docked.
Zathalus
Member
In handheld mode I think the Rog Ally X at 20/25w will come out on top, but it is drawing way more power to do so and is almost twice the price, so the Switch 2 performing as well as it does is already a win.
Insane Metal
Member
From the little I saw it exceeded tbh. But so did the price. Ew.
CDPR stated that they're targeting 1080p/30fps for quality mode on handheld and 720p/40fps for performance mode.
rodrigolfp
Haptic Gamepads 4 Life
Then it doesn't make sense docked also be 1080p/30fps/40fps.
ValiantInstance
Member
You don't really need a 1080p resolution on such a small screen. It's nice to have but not really an issue. I think 720p might be slightly too low, 800 would be nice.
Doctor Hades
Member
I wasn't expecting much but Nintendo claim that the Switch 2 has 10X the performance of the Switch so on hearing that the screen was 1080p and the dock supported 4K I was kind of expecting more.
What irks me the most though is Nintendo's stubborn refusal to use DLSS or any kind of anti-aliasing in most of their games never mind anisotropic texture filtering, especially as I only buy Nintendo consoles for the first-party games. This is kind of a must for playing Switch games on a 4K TV so they don't look a jagged, shimmery mess while playing. I own a PS5 Pro and a PC (oh and something called an Xbox Series X, what a disappointment that is!) so I have zero-interest in playing third-party games on Switch or Switch 2 yet those are the games that are likely to embrace DLSS or use anti-aliasing etc etc.
The definitive Nintendo game experience is still on PC in my experience. I have played over 80 hours of Zelda: Breath of the Wild on my PC via the Cemu emulator with enhancements that make the game look far superior to the Switch 2 version and runs at 120 fps. Yet that game is the Wii U version (and, yes, I do still own boxed Wii U and GameCube games). Also, Mario Kart 8 on PC looks better than Mario Kart World.
The nail in the coffin was hearing that GameCube games would only run at 1280x960 on Switch 2 despite the fact that the NVIDIA Shield, which uses the same chipset as the Switch, runs GameCube games at higher resolutions according to Digital Foundry (although this is only for the Chinese market, I believe). So you would think that GameCube games would run at higher resolutions on the more powerful Switch 2, right? 1280x960 might be OK for the 1080p handheld screen but I'm guessing the games will not look good on a 4K TV.
What irks me the most though is Nintendo's stubborn refusal to use DLSS or any kind of anti-aliasing in most of their games never mind anisotropic texture filtering, especially as I only buy Nintendo consoles for the first-party games. This is kind of a must for playing Switch games on a 4K TV so they don't look a jagged, shimmery mess while playing. I own a PS5 Pro and a PC (oh and something called an Xbox Series X, what a disappointment that is!) so I have zero-interest in playing third-party games on Switch or Switch 2 yet those are the games that are likely to embrace DLSS or use anti-aliasing etc etc.
The definitive Nintendo game experience is still on PC in my experience. I have played over 80 hours of Zelda: Breath of the Wild on my PC via the Cemu emulator with enhancements that make the game look far superior to the Switch 2 version and runs at 120 fps. Yet that game is the Wii U version (and, yes, I do still own boxed Wii U and GameCube games). Also, Mario Kart 8 on PC looks better than Mario Kart World.
The nail in the coffin was hearing that GameCube games would only run at 1280x960 on Switch 2 despite the fact that the NVIDIA Shield, which uses the same chipset as the Switch, runs GameCube games at higher resolutions according to Digital Foundry (although this is only for the Chinese market, I believe). So you would think that GameCube games would run at higher resolutions on the more powerful Switch 2, right? 1280x960 might be OK for the 1080p handheld screen but I'm guessing the games will not look good on a 4K TV.
Baemono
Member
I played Hogwarts on the Deck from start to finish, it was worse than the Switch 2 footage.
Buggy Loop
Gold Member
Daniel Thomas MacInnes
GAF's Resident Saturn Omnibus
As far as I'm concerned, videogame graphics peaked with Soul Calibur on Sega Dreamcast. Everything that has followed since is icing on the cake. Heck, to us kids who grew up in the 1980s, the movie Tron was considered the peak of computer graphics. We couldn't imagine videogames ever looking better than that. We passed that milestone decades ago, hah.
Everything today is amazing and fantastic. Better than I could hope or ask for.
Radical_3d
Member
Team forties! This and modern gaming sucks.
dgrdsv
Member
VOPD is useless in general I'd say but it does provide a theoretical flops peak which is correct when counted as a doubling of non-VOPD flops.
In any case math flops of all things aren't the main limiting factor for RDNA3 performance in a 15W power envelope.
Lysandros
Member
Do you happen to know how Nvidia Ampere Teraflops are calculated? Are double rate fp32 calculations included in a similar fashion? I am reading 'ampere compute figures being inflated' left and right but couldn't find a definitive answer after doing a bit of research.
Ampere really does have double FP32, though accessing it requires sacrificing the use of the integer unit. The 3080 has the same number of SMs as the 2080 Ti, but gets 30% more performance on average. Compare that to RDNA 3, where you get maybe 5% better performance.
Variahunter
Member
Yes but Deck has been released 3 years ago, something most people tend to forget in comparisons.
Get out of here with your first hand experience!
Last edited:
Variahunter
Member
Could you please precise what are those settings and the frame rate on the deck ?
Switch 2 really has a huge upgrade over Steam Deck graphically.
Because it could just be a fanboy puttings the lowest settings on the deck just to make an unfair comparison.
Not saying it's the case here, but that wouldn't surprise me honestly.
And again, Deck is a 2022 device.
What really annoys me is Nintendo and Nvidia not giving the exact specs.
Lysandros
Member
I see, so T239's stated 3TF docked figure is with this feature factored in? Like in a sense it's 1.5 ('base TF' without double rate fp32)×1.3=1.95 TF, so about 2TF is the closer to real throughput, less theoretical figure? It is indeed inflated in that sense then, confirming what i have been reading. Another analogy (i know this isn't 1:1) might be PS4 PRO's RPM feature. PS4 PRO was theoretically 8.4 TF at half/fp16 precision. We know that RPM isn't useless and it can bring a significant uplift in real world compute throughput (maybe about 20% with good use). We could factor that fact in using the same mentality like this 4.2×1.2=5 TF to reach a higher figure for PS4 PRO, but we of course aren't doing it. All this say to say that it may be fairer to account for this aspect when comparing Switch 2 to another system like PS4 regarding theoretical compute ceilings. It isn't quite 3>1.84 really, there are nuances there.
Last edited:
BlueManifest
Neo Member
It surpassed my expectations by a little, I wasn't expecting 120 fps and gysnc in handheld mode
Last edited:
rodrigolfp
Haptic Gamepads 4 Life
Handhelded??
Switch 2 really has a huge upgrade over Steam Deck graphically.
That's true but doing that calculation gives you the equivalent of a 1.95 TF Turing card when in reality Turing was slightly faster than RDNA 2 per teraflop, and RDNA 2 is 1.25X GCN. The multiplier I use is 0.7 to get the roughly equivalently performing RDNA 2 part (~33.6 TF 3080 competes with ~23.8 TF 6900 XT) and 0.875 to get the equivalently performing GCN part (0.7 * 1.25 = 0.875). So by this metric, 3.1 Ampere TF would roughly equal 2.17 RDNA 2 TF or 2.7 GCN TF. That is, a bit more than half the power of the Xbox Series S, or in between the PS4 and the PS4 Pro. Other than a few possible outliers, so far it looks like this matches what we see in Switch 2 games.
Last edited:
Lysandros
Member
I see, thanks. To clarify, i was only assessing/comparing theoretical compute ceilings in isolation amongst other metrics/indicators since it's the most overused one which show up the most often in arguments. Since compute isn't the only indicator of game performance and its efficiency/real world throughput depends on a lot of factors like bandwidth (be it caches or memory), async, custom bits, APIs etc. It's all the more difficult to access its veracity for comparing specific platforms. Then there is the fixed function/rasterisation part of the picture of course. I guess we'll see the released games running side by side in due time for a better assessment.
Well Ampere compute isn't "faked", so in a synthetic benchmark the Switch 2 should be able to hit that 3.1 TF figure.
Lysandros
Member
Of course, i never said that it was "faked" and couldn't reach it in synthetic benchmarks. I guess PS4 PRO could also reach 8.4 TF in a tailor made FP16 synthetic benchmark. I was only inquiring about the intricacies.
FingerBang
Member
I have tried it today. My answer is yes, fully. The games I tried look good on a TV screen and absolutely gorgeous on the Switch 2 screen. It's miles ahead of the first one (as expected) and I love the form factor. So much power in such a slick device.
Last edited:
Trilobit
Absolutely Cozy
In this case fps stands for fires per switch.
Gambit2483
Member
i didn't expect to see it damn near on par with PS5. Visually, games like FF7R, Cyberpunk 2077 and SF6 look only slightly less than their PS5 counterparts, at least from what I can see. That's extremely impressive for what is essentially a handheld.
I expected DLSS and 4k but I didn't expect VRR and a 120fps screen. Nintendo truly went above and beyond to future proof this thing!
I expected DLSS and 4k but I didn't expect VRR and a 120fps screen. Nintendo truly went above and beyond to future proof this thing!
Last edited:
RJMacready73
Simps for Amouranth
As a non Nintendo fan, I honestly couldn't see the difference, because every Nintendo game to me looks the same as the previous gen and I'm not 5 years old it was a major meh, I stopped giving a shit about Nintendo when my balls dropped and voice changed, they're like the Michael Jackson of videogames, never changes, all about the kids and something just off lying beneath the surface
Gambit2483
Member
Huh? Litterally no one ever expected that. Hell, no one even expects that on a PS5 Pro.
blacktout
Member
Seriously, like half of GAF believed that it wouldn't even be able to output in 4k, and even the biggest optimists, myself included, didn't think you would have games like Metroid Prime 4 hitting 4k/60.
BlackTron
Member
What I really didn't see coming was 120FPS screen, nevermind supporting it on TV. With 120FPS and mice they really broadened the scope of games I can take this seriously to play. Lots of PC games won't really need a PC for a similar experience.
Don't worry guys not all PC games. Put the pitchforks away.
Jinzo Prime
Member
Kataploom
Gold Member
Really, someone might have something wrong in their heads if they really thought it would be that powerful, thank god I haven't seen anyone
BlackTron
Member
To be fair. I wouldn't be able to tell the difference between a PS5 and PS4 title if I never played the PS4 game either.