• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

What is the actual power of the Nintendo Switch?

beril

Member
Can you explain me why in a simple way?

Also, what do you think about the difference between a PC processor and a tweaked mobile processor?

TechGAF, explain to me.

Unless you code in assembler language the compiler takes care of all that.

Compiling my engine for x86 or ARM required exactly 0 changes.

For PPC on WiiU I changed two lines of codes to do an endian switch when loading data files

Of course if you want to do very low level optimizations you have to tailor it to the CPU to some extent; but that's true for different cores with the same instruction sets as well. The best way to optimize for on an i7 might not be at all the same as for a jaguar.
 

The Wii U number on Wikipedia is a common misconception that actually began here if I recall correctly, and it should be half that number. I'll try to find a source which clarifies that. Here is a source for the Switch numbers:

http://www.eurogamer.net/articles/d...-boosts-handheld-switch-clocks-by-25-per-cent

To calculate flops you multiply the number of cores (256 on a TX1) by 2, then by the clock rate based on 1GHz =1. So you get 256*2*0.3072 (or 0.384 or 0.768 depending on undocked, boosted or docked) which winds up as 157, 192, or 390 Gflops (in full precision) respectively.
 

Costia

Member
Unless you code in assembler language the compiler takes care of all that.

Compiling my engine for x86 or ARM required exactly 0 changes.

For PPC on WiiU I changed two lines of codes to do an endian switch when loading data files

Of course if you want to do very low level optimizations you have to tailor it to the CPU to some extent; but that's true for different cores with the same instruction sets as well. The best was to optimize for on an i7 might not be at all the same as for a jaguar.

It doesn't affect how the code is written, but it does affect how long it takes to execute.
ARM is a RISC cpu, x86 isn't. In other words something that is a single instruction that takes a single cycle in x86 be multiple instructions that take multiple cycles on a RISC cpu.

The Wii U number on Wikipedia is a common misconception that actually began here if I recall correctly, and it should be half that number. I'll try to find a source which clarifies that. Here is a source for the Switch numbers:

http://www.eurogamer.net/articles/d...-boosts-handheld-switch-clocks-by-25-per-cent

To calculate flops you multiply the number of cores (256 on a TX1) by 2, then by the clock rate based on 1GHz =1. So you get 256*2*0.3072 (or 0.384 or 0.768 depending on undocked, boosted or docked) which winds up as 157, 192, or 390 Gflops respectively.
I guess they mixed Fp16 and FP32 in the table.
 
I guess they mixed Fp16 and FP32 in the table.

If you're talking about the Wii U number I don't think so, since the Wii U wasn't capable of 2x FP16 processing like the Switch now is.

I think the problem was based on the die shots people saw two times the amount of something (not sure what) and factored that into equations before that could be corrected, so a lot of websites took that 352 number and went with it without bothering to check the corrections.

Someone who was here back then can clarify that better than I can.
 
We should be judging the Switch by its undocked mode, because undocked will always hold docked back. In terms of development time devs will always go for something simple like a resolution or frame rate upgrade. It makes no sense to make the games substantially different in the different modes, and its against what Nintendo wants. A seamless experience.

I don't understand why people feel the need to act as mini marketing departments for Nintendo (and by that I mean spouting meaningless bullshit that doesn't relate to real world application). The Switch is an incredible piece of hardware, not just as a handheld but as a console. We don't need to pretend its "nearly as powerful as an xbox one". Be real.
Why the fuck do you think I would be marketing Nintendo? Lol

Also it makes no difference to judge the switch in undocked mode. Devs will still push it as much as they can for a 720p(or even less like 540p undocked and 720p docked)screen in graphical fidelity and upres it to 1080p in docked mode if they see fit.

And the fact that those games like Minecraft exist, doesn't invalidate what I said earlier. Nintendo may nit care about doing it for their games, but third oartyDevs will do whatever the hell they want with it. Who is to say its a waste of time? Are you a dev? And how would it not be seem less? Lol

And yes, saying that the switch is only 2 to 2.5x weaker in GPU is perfectly fine to say. It falls around that ball park, especially from the ports we've seen. The only thing that could really be holding switch back in poets is the bandwidth, but they could work around it to some degree.
 

ViolentP

Member
Powerful enough to play Bomberman at 60fps, not powerful enough to play Binding of Isaac without hitting single-digit frames.
 
Powerful enough to play Bomberman at 60fps, not powerful enough to play Binding of Isaac without hitting single-digit frames.

Errr... when you get Binding of Isaac to single digit frames on Switch (which has happened once in my 150 hours) would that really be 60fps on other consoles or PC? I imagine that's a limitation of the game, not the hardware since you basically have to break the game to get there.
 

beril

Member
It doesn't affect how the code is written, but it does affect how long it takes to execute.
ARM is a RISC cpu, x86 isn't. In other words something that is a single instruction that takes a single cycle in x86 be multiple instructions that take multiple cycles on a RISC cpu.

Well yes, but the timing of individual instructions is still irrelevant to anyone but the most hardcore assembler coders.
A57 beats jaguar hands down when running at the same clocks.

The Switch CPU isn't as powerful as Xbox One or PS4 and no one has claimed that it is; it has fewer cores and lower clockspeed. Obviously some games would have to scale back a few things because of the power difference; but the fact that it's ARM instead of x86 really has minimal, if any effect at all on porting
 

ViolentP

Member
Errr... when you get Binding of Isaac to single digit frames on Switch (which has happened once in my 150 hours) would that really be 60fps on other consoles or PC? I imagine that's a limitation of the game, not the hardware since you basically have to break the game to get there.

I own BoI on PC, PS4, Vita, and Switch and the Switch definitely has the biggest framerate issues. And it's certainly not from "breaking" the game.

That being said, I would wager there is much more going on in Bomberman than BoI so I would peg the game's performance issues on the port job itself. But to answer OP's question, in my experience thus far with the Switch, this has been the biggest disparity between performance in games I have experienced personally.
 

bomblord1

Banned
Nice write up but there is a correction to be made. The jump the new 3ds made from og 3ds was CPU related, going from 2x arm 11 @ 233 mhz to 4x arm 11 @ 804 mHz (or thereabouts). The gpu didn't get touched.

I was referring to the power leap in a general sense.

ex: New 3DS is is to the 3DS as the Switch GPU is to the WiiU GPU
 

Izcarielo

Banned
I own BoI on PC, PS4, Vita, and Switch and the Switch definitely has the biggest framerate issues. And it's certainly not from "breaking" the game.

That being said, I would wager there is much more going on in Bomberman than BoI so I would peg the game's performance issues on the port job itself. But to answer OP's question, in my experience thus far with the Switch, this has been the biggest disparity between performance in games I have experienced personally.

Is BoI Switch's version worse than the Vita one? Because ive extensively played the Vita port and I dont even want to start imagining a worse framerate than that
 

ViolentP

Member
Is BoI Switch's version worse than the Vita one? Because ive extensively played the Vita port and I dont even want to start imagining a worse framerate than that

It's been a while since I've played the Vita copy but I don't recall it being this bad. In fairness, I have likely already put in more time in the Switch version. The game itself runs well for the most part, but certain loadouts make the game take a serious hit. This is illustrative of my experiences with the issue: https://www.youtube.com/watch?v=6WukOc2DgZw
 

Easy_D

never left the stone age
It lags and chugs along while I play Binding of Isaac when I get too many spiders or flies on the screen. That said I still love the console. Wish it had more power, though.

Something tells me that's not really the Switch's fault lol. It's not really CPU starved

From slightly less than Wii U to 1.5x in undocked mode + more RAM

How can it be weaker than the Wii U in portable mode if it can do Mario Kart 8 in portable mode without any downgrades whatsoever compared to the Wii U lol
 
Errr... when you get Binding of Isaac to single digit frames on Switch (which has happened once in my 150 hours) would that really be 60fps on other consoles or PC? I imagine that's a limitation of the game, not the hardware since you basically have to break the game to get there.

The original also had frame drops during explosions in PC. There were able to fix it
 

jts

...hate me...
Unlimited power.

Halfway between PS3 and PS4 which makes it look more on par with current gen than any other previous .5 gen console could - dem diminishing returns are real.
 

FinalAres

Member
but, it is, on paper. Switch shares more commonalities with its console big brothers than Vita shared with PS3.
Could you explain this? I'm genuinely curious.

This is my limited understanding.
Vita was billed as being able to do ps3 lite graphics. I assume that was marketing tosh but the fidelity of the graphics didn't seem that different. Vita looks to be sub PS3 but better than ps2 by a long shot so not a generation behind.

Switch is a super up WiiU essentially, that's a whole generation behind.

I have numbers that back up my understanding but I won't pretend to get the ins and outs of what they mean, so I thought I'd be honest and present my layman's thinking.
 
It's been a while since I've played the Vita copy but I don't recall it being this bad. In fairness, I have likely already put in more time in the Switch version. The game itself runs well for the most part, but certain loadouts make the game take a serious hit. This is illustrative of my experiences with the issue: https://www.youtube.com/watch?v=6WukOc2DgZw

That's just Toxic Shock causing that?? I've had that several times and never seen a slowdown nearly that bad. Some of the "fog" effects give it a very minor slowdown but nothing that bad.

I had a single digit framerate the other day but that was because I had Death's Touch + Proptosis + Chocolate Milk which gave me sickles so big a single one filled an entire 2x2 room. I assume that type of thing would cause a slowdown on any platform.
 

Nheco

Member
Wii: 2 Gamecubes duct taped together
WiiU : 16 Gamecubes duc taped together
Switch: 24 Gamecubes duct taped together

Way more precise than flops.
 

bomblord1

Banned
Could you explain this? I'm genuinely curious.

This is my limited understanding.
Vita was billed as being able to do ps3 lite graphics. I assume that was marketing tosh but the fidelity of the graphics didn't seem that different. Vita looks to be sub PS3 but better than ps2 by a long shot so not a generation behind.

Switch is a super up WiiU essentially, that's a whole generation behind.

I have numbers that back up my understanding but I won't pretend to get the ins and outs of what they mean, so I thought I'd be honest and present my layman's thinking.

Vita was initially billed as a PS3 lite and in the end it ended up being even less than that. It got several cross-platform games but all of them were either 1. Held back by the Vita (see trials of cold Steel dev comments on things like world size) or 2. Ran absolutely horridly (see Atelier games, Ar No Surge, Jack Trilogy, etc). None of these were even close to system pushing games either.

The Vita's initial specs showing the CPU could hit I believe it was a max of 2ghz ended up being absolutely bologna the final clocks were 333mhz (or 444mhz in a mode that disabled the wifi and some background stuff). The gflops were about 30 meaning it wasn't even 1/4 of the PS3's GPU and then the CPU wasn't even in the same ballpark. I believe it was also held back by mobile graphics API's that were around at the time of the hardware being released.
 

jts

...hate me...
It's basically an Xbox 360 which was about an Xbox 1.5 and that makes it a Wii, which was a supercharged Gamecube. Which was not that much stronger than a DreamCast which in turn was almost a PS1, which, let's face it, it was just the SNES with a CD add-on.

I guess what I'm saying is that the Switch has less the power than a NES. And that's when docked.
 

Doctre81

Member
Its around 50% of the raw performance of a vanilla xbox one which given its size and the fact that it can be taken anywhere is pretty amazing.
 

orioto

Good Art™
The numbers are not that mysterious i mean, i'm surprised we're still debating that as some amazing secret...
Now between what a platform can do and what wil be done with it..

The thing with Nintendo platform is, Nintendo first party are the more talented with them, yet they always have really particular art direction that makes comparison difficult. Would be really funny to see BotW ported on PS3 to see how it runs, or even MK8d.

I wonder what game will be a really show off for the platform, at least something that is obviousy far away from last gen, even for non believers.
 

orioto

Good Art™
This. It's a decent handheld, but it won't get the games that gaf wants and that's the reality of the situation.

And it has absolutely nothing to do with its power by the way. All those games would be prefectly enjoyable with downgraded graphics and modern tech make downgrading assets easy. And you know what, they would actually still look pretty good after the downgrade.
 

OmegaFax

Member
Vita was initially billed as a PS3 lite and in the end it ended up being even less than that. It got several cross-platform games but all of them were either 1. Held back by the Vita (see trials of cold Steel dev comments on things like world size) or 2. Ran absolutely horridly (see Atelier games, Ar No Surge, Jack Trilogy, etc). None of these were even close to system pushing games either.

The Vita's initial specs showing the CPU could hit I believe it was a max of 2ghz ended up being absolutely bologna the final clocks were 333mhz (or 444mhz in a mode that disabled the wifi and some background stuff). The gflops were about 30 meaning it wasn't even 1/4 of the PS3's GPU and then the CPU wasn't even in the same ballpark. I believe it was also held back by mobile graphics API's that were around at the time of the hardware being released.

The Vita's CPU, GPU, and Bus could can all be set to different clock speeds. I honestly don't know why the 444mhz mode disabled WiFi on some earlier games like Gravity Rush. I don't think the 2000 series model generates enough or any heat at that mode to need to disable it it. Maybe it was a precaution for the 3G model they decided to mandate across the board. 3G radios tended to get a bit on the warm side on their own.

2Ghz was such a weird comment Sony made. At the time the Vita came out, to put it into perspective, the iPad 3 w/ Retina Display just came out and had the same-ish GPU w/ a dual core CPU setup. It had a CPU clocked at 1Ghz and I think the GPU anywhere from 200mhz-400mhz off the top of my head and that thing pushed way more pixels and tended to get hot during more demanding games. The aluminum spread the heat and was probably part of the heatsink.

The Vita did a pretty good job given what developers had to work with. The screen's native resolution, I think, never hit its maximum potential because games like FFX/X-2 and MGS2/3 would shave off some pixels for performance.

Tearaway is one of the few games I've seen that totally shines at native w/ 60fps (I could be mistaken but it looks like the cleanest/fluid 3D game I've played on Vita yet. Killzone Mercenary, I don't think was native but damn, was it good.

If you play something like Uncharted Golden Abyss and then set Remote Play to 540p60 on Vita and play Uncharted Collection, the screen is just surprisingly more crisp.

Anyway. Early on ... I'm glad developers are trying to keep to a 720p60 experience at a bare minimum for Switch portable. The screen and games look amazing.
 

Oregano

Member
Vita was initially billed as a PS3 lite and in the end it ended up being even less than that. It got several cross-platform games but all of them were either 1. Held back by the Vita (see trials of cold Steel dev comments on things like world size) or 2. Ran absolutely horridly (see Atelier games, Ar No Surge, Jack Trilogy, etc). None of these were even close to system pushing games either.

The Vita's initial specs showing the CPU could hit I believe it was a max of 2ghz ended up being absolutely bologna the final clocks were 333mhz (or 444mhz in a mode that disabled the wifi and some background stuff). The gflops were about 30 meaning it wasn't even 1/4 of the PS3's GPU and then the CPU wasn't even in the same ballpark. I believe it was also held back by mobile graphics API's that were around at the time of the hardware being released.

The default Vita mode for Vita is actually closer to 15 GFlops which is less than the OG Xbox.
 
Wii-U to Switch looks almost exactly the equivalent jump from GC to Wii.

Seriously. It's actually shocking how Nintendo seems to have followed nearly the exact same playbook:

- 2.2X jump on RAM for GC > Wii, 3X jump from Wii-U >Switch
- 1.5X jump on GPU for GC > Wii1.4X jump from Wii-U > Switch
- 1.5X jump on CPU for GC > Wii1.3X jump from Wii-U > Switch

What numbers you are using to get this comparison for the Wii U vs Switch? Whatever it is from, it is flawed.

RAM: 1GB for games vs 3.25GBs (3.25x)

GPU: 176 GFLOPS GPU based on the R700 series vs 196/393 GFLOPS GPU based on 2nd-gen Maxwell (Likely more than 1.4x undocked, and more than 2.8x docked, not including fp16)

CPU: 3x 1.25GHz CPU based on Gekko/G3s vs 4x 1.02GHz A57s (3x would be a modest estimate)
 
And which games are those?

Red dead 2, mass effect andromeda, destiny 2, battlefront2 to name a few.. I could go on.

And it has absolutely nothing to do with its power by the way. All those games would be prefectly enjoyable with downgraded graphics and modern tech make downgrading assets easy. And you know what, they would actually still look pretty good after the downgrade.

imo i believe if it had the power of the current gen consoles porting would be a piece of cake, just the minimum baseline of the Xbox one and I'm sure it would get those ports. However, you have to take into consideration handheld and dock mode.. so power is definitely a factor.
 
The Vita's CPU, GPU, and Bus could can all be set to different clock speeds. I honestly don't know why the 444mhz mode disabled WiFi on some earlier games like Gravity Rush. I don't think the 2000 series model generates enough or any heat at that mode to need to disable it it. Maybe it was a precaution for the 3G model they decided to mandate across the board. 3G radios tended to get a bit on the warm side on their own.

2Ghz was such a weird comment Sony made. At the time the Vita came out, to put it into perspective, the iPad 3 w/ Retina Display just came out and had the same-ish GPU w/ a dual core CPU setup. It had a CPU clocked at 1Ghz and I think the GPU anywhere from 200mhz-400mhz off the top of my head and that thing pushed way more pixels and tended to get hot during more demanding games. The aluminum spread the heat and was probably part of the heatsink.

The Vita did a pretty good job given what developers had to work with. The screen's native resolution, I think, never hit its maximum potential because games like FFX/X-2 and MGS2/3 would shave off some pixels for performance.

Tearaway is one of the few games I've seen that totally shines at native w/ 60fps (I could be mistaken but it looks like the cleanest/fluid 3D game I've played on Vita yet. Killzone Mercenary, I don't think was native but damn, was it good.

If you play something like Uncharted Golden Abyss and then set Remote Play to 540p60 on Vita and play Uncharted Collection, the screen is just surprisingly more crisp.

Anyway. Early on ... I'm glad developers are trying to keep to a 720p60 experience at a bare minimum for Switch portable. The screen and games look amazing.

I did always find that baffling too, Sony definitely made the mistake of overselling the Vita, especially around people's expectations. You had them show Metal Gear Solid 4 at the reveal event in January 2011, which made people think Vita could do MGS4...but it was actually the Vita doing a cutscene of MGS4 with no actual gameplay, and at a reduced framerate at that. It also didn't help that the screenshots of launch games Sony released to the press were at a much higher resolution than the Vita's screen, or what the games finally ended up rendering at. I was disappointed to find Everybody's Golf being so blurry (was one of the games that ran at 360p), as I always remembered it looking so crisp. Great game though, one of Vita's very best.

Tearaway was actually locked to 30fps, but it's still a beautiful game. At the time it just looked so clean, running at native resolution with antialiasing.
 

Astral Dog

Member
And it has absolutely nothing to do with its power by the way. All those games would be prefectly enjoyable with downgraded graphics and modern tech make downgrading assets easy. And you know what, they would actually still look pretty good after the downgrade.
Well it does have something to do with hardware power because that will determine the amount of cost and time in porting those assets.lets not pretend the system itself is not any inconvenience including the media cards.
But this argument always goes in cycles
 

Rodin

Member
Wii-U to Switch looks almost exactly the equivalent jump from GC to Wii.

Seriously. It's actually shocking how Nintendo seems to have followed nearly the exact same playbook:

- 2.2X jump on RAM for GC > Wii, 3X jump from Wii-U >Switch
- 1.5X jump on GPU for GC > Wii1.4X jump from Wii-U > Switch
- 1.5X jump on CPU for GC > Wii1.3X jump from Wii-U > Switch

The only possible thing that makes this jump better than what happened with GC > Wii is that unlike that situation, Nintendo seems to be using completely different architecture for the CPU and GPU compared to its predecessor. And even then, that's not guarantee that it makes it a substantial jump in any way (it could be possible that the newer architecture simply does things that Wii-U's CPU/GPU did in a different, but not particularly more efficient manner).

Ignoring specs, and letting the games do the talking, there is not a single game on the Switch that makes one think it is significantly more powerful than Wii-U. Super Mario Odyssey looks about as improved from SM3DW as Super Mario Galaxy did from Sunshine. And it should be pointed out that even with SMG, as much as I loved that game and think it looks gorgeous (for a Wii title), part of its visual improvement was due to the additional power of Wii, but ALSO due to the fact that it was a completely different style of game than Sunshine which made it look more impressive than it would have. Galaxy probably could have ran on GC with minor downgrades here and there, just fine. Similarly, I wouldn't be surprised if SMO's different design style from SM3DL might give the illusion of it being more impressive and pushing more of the hardware than it's actually doing.

So until I have evidence otherwise, Switch is basically a Wii-U turbo.
dae.jpg
Not only a random early port of a Lego game proves you wrong about the bolded, but you gotta explain to me where did you get those numbers for the "improvements". Your ass isn't a good source.

Could you explain this? I'm genuinely curious.

This is my limited understanding.
Vita was billed as being able to do ps3 lite graphics. I assume that was marketing tosh but the fidelity of the graphics didn't seem that different. Vita looks to be sub PS3 but better than ps2 by a long shot so not a generation behind.

Switch is a super up WiiU essentially, that's a whole generation behind.

I have numbers that back up my understanding but I won't pretend to get the ins and outs of what they mean, so I thought I'd be honest and present my layman's thinking.

This is some nice revisionism. Even by just looking at the same numbers you're mentioning (the actual ones though), Switch is much, much closer to current consoles than Vita was compared to PS3. The incredible overestimation of Vita hardware is only rivaled by the ridiculous downplay of Switch's specs. I thought we were over this.
 
What numbers you are using to get this comparison for the Wii U vs Switch? Whatever it is from, it is flawed.

RAM: 1GB for games vs 3.25GBs (3.25x)

GPU: 176 GFLOPS GPU based on the R700 series vs 196/393 GFLOPS GPU based on 2nd-gen Maxwell (Likely more than 1.4x undocked, and more than 2.8x docked, not including fp16)

CPU: 3x 1.25GHz CPU based on Gekko/G3s vs 4x 1.02GHz A57s (3x would be a modest estimate)

Yeah we actually have numbers this time. Not sure why there's so much misinformation going around, these are the actual numbers.
 

ViolentP

Member
That's just Toxic Shock causing that?? I've had that several times and never seen a slowdown nearly that bad. Some of the "fog" effects give it a very minor slowdown but nothing that bad.

I had a single digit framerate the other day but that was because I had Death's Touch + Proptosis + Chocolate Milk which gave me sickles so big a single one filled an entire 2x2 room. I assume that type of thing would cause a slowdown on any platform.

There are certainly some overkill loadouts that bring things to a crawl. That is totally understandable because things in the game can get hectic. It seems as if some effects however are not as refined in this version of the game. Still my favorite game on the device.
 

Rncewind

Member
And it has absolutely nothing to do with its power by the way. All those games would be prefectly enjoyable with downgraded graphics and modern tech make downgrading assets easy. And you know what, they would actually still look pretty good after the downgrade.

wow so you actually belive there is magic port button in games

i thought it was obvious that the gifs were memes
 
Top Bottom