Support NeoGAF

Skittzo0413 · Feb 22, 2017

HeelPower said:
Because it matters if Nintendo treats it as the 3DS successor.That would mean its gonna get an amazing stream of games.

As a home console,it has less chances of getting third parties due to power gap.

The thing is, it's going to get both. And that's why it's so exciting. As you can see from Snake Pass, ports (properly optimized like on UE4) will be fairly similar to the PS4/XB1 versions, though with some obvious downgrades. And as for handheld games once Pokemon drops the 3DS will likely see the end of its life, so all of those handheld style will be Switch exclusives.

That's why the hybrid idea is such a good one, at least in the long run. You build up a very large and very unique library.

BlastProcessing said:

It's been too long since NX *cough* I mean Switch threads were full of Simpsons memes.

Laevateinn · Feb 22, 2017

Genocyber said:
Is the Nintendo Switch physically the smallest home console?

PC Engine says hi

20cent · Feb 22, 2017

Fuchs said:
I've got no idea what all of this means.

We are amazed that a tablet-like device released in 2017 looks like a tablet on the inside once you open it.

Also that a 4310 mAh battery seems bigger than say a 2000mAh battery.

Wait for when we discover that they used a 6.2 inch panel for their 6.2in LCD screen.
Spoiler: That's probably the case.

Astral Dog · Feb 22, 2017

BDGAME said:
Good to know your true expectations.
In defense of myself, tell me, Do you never read people thinking it can use pascal and have 600 or more GFlops? Am I the first to tell about the possibility of it use A72 CPU? How many people do you seem hopping for a 30% better permanece in practice than in paper because Nvidia?

The only thing I do is put all these expectations together.

http://m.neogaf.com/showthread.php?t=1342575&page=100000

People assume it is 1080p 60fps on that thread and no one try to fix it. So I assume it's true.

Size? Really?

I mean, is there a weak argument than the size?

Come on of course size matters.

Switch is an incredible portable gaming device but its designed to work on a small battery and be smaller than a GamePad.

Xbox One was a huge power eating monster in comparison,just because it has been three years since it released doesn't mean mobile tech is close to surpass it,much less with how batteries have (not) improving.

Three years is not enough.

antonz · Feb 22, 2017

koziakauzu said:
We are amazed that a tablet-like device released in 2017 looks like a tablet on the inside once you open it.

Also that a 4310 mAh battery seems bigger than say a 2000mAh battery.

Wait for when we discover that they used a 6.2 inch panel for their 6.2in LCD screen.
Spoiler: That's probably the case.

Oh aren't you so funny. herp derp
You do seem to be jumping around a lot of threads just to leave piles of shit behind you.

gregor7777 · Feb 22, 2017

antonz said:
Oh aren't you so funny. herp derp
You do seem to be jumping around a lot of threads just to leave piles of shit behind you.

His post history is BRUTAL. I kinda feel bad for him.

Astral Dog · Feb 22, 2017

koziakauzu said:
We are amazed that a tablet-like device released in 2017 looks like a tablet on the inside once you open it.

Also that a 4310 mAh battery seems bigger than say a 2000mAh battery.

Wait for when we discover that they used a 6.2 inch panel for their 6.2in LCD screen.
Spoiler: That's probably the case.

Lol you would be amazed,GAF got very confused by that,check the thread

Astral Dog · Feb 22, 2017

HeelPower said:
Because it matters if Nintendo treats it as the 3DS successor.That would mean its gonna get an amazing stream of games.

As a home console,it has less chances of getting third parties due to power gap.

They don't treat it as a portable successor because they still need to sell those 3DSes,this holiday and next year you will see a much bigger focus on the Switch.

But its clear despite their messy messaging Switch will combine both typical handheld and console teams
If that makes it a success its to be seen after all Switch is much more expensive to make games for than 3DS ever was

BriareosGAF · Feb 22, 2017

The RSX was not a modern GPU, didn't it use fix functions and was a big issue with the PS3?

The RSX was not a fixed function GPU, it was a fully programmable vertex and fragment shader based GPU. The closest thing you could say to fixed about it was the fact that it did not have unified shader cores so you could not re-balance work between vertex and fragment dynamically; this was why so much effort was spent on improving vertex attribute fetch, culling, etc., because the fragment units were idle while you were processing vertex work.

I'm sure the PS3 was a huge problem, but I'm not sure you can compare it to a modern GPU doing fp16 code as these were designed around it.

Nor easily compare mobile titles to AAA, nor lighting setups from gen7 to gen8, etc. The point isn't that the PS3 was a huge problem, but that fp16 is not a keyword you sprinkle through your shader until it's "fast enough", or a radio box you toggle in a middleware package.

The reference of 70% of an ex ubisoft developer's code using fp16 is his own statement. UE4 also does support FP16 and AMD is pushing FP16 heavy with VEGA, so expect to use it more in AAA products if that is where you work, because the entire industry is about to start heavy pushing.

Unsurprisingly, I am well aware of this. We regularly meet with the IHV hardware architects and discuss these issues. Our team provides microcode analysis to some IHVs about our shader workloads, for instance, so they understand ISA usage in the wild, and have even provided information about our historical use of FP16 in our titles when discussing newer architectural revs.

I'm pretty sure local promotion would mean using FP32 in intermediate steps (that is, input and output are FP16 but the calculations inbetween might have to be higher precision to avoid inaccuracy).

Close, but there are two separate idioms. One is at the source level when authoring: "most of the lighting math is running in half but there are a few areas like the Schlick approximation for the fresnel that we promote to float, along with some parts of the filmic tonemapper." The other form is at the ISA level, where even when you want some maths to execute at half precision, the ISA may or may not support all the operations you want in half mode, and so the compiler may need to marshall types at the register level, which can reduce effective throughput.

BDGAME · Feb 22, 2017

Skittzo0413 said:
Have you actually read that thread? As you can tell I posted in it quite a bit. There's only PS4 footage shown, not XB1. And most people seem to agree that the PS4 footage is not locked at 60fps.

EDIT:
Snake Pass to me is an example of a game that shows off how well the Switch will punch above its weight. You have the PS4 version struggling to maintain 60fps, and the Switch version locked at 30fps with essentially the same visuals (I've heard different lighting but I haven't seen any appreciable changes) which, theoretically would put the Switch version at >50% of the PS4 version.

Whereas people going by pure flops keep reminding us that the Switch has less than 25% the flops of the PS4 going by the July devkit specs and Eurogamer clocks. So performing >50% should indicate that flops tell a very small part of the overall story.

Yes I read.
The fact that the ps4 version is right now with unstable frame rate is not necessarily definitive, since the game is in development.

The rest I agree with you. Of course there are more than flops. The fact that Wii U is a 176 GFlop machine and more stronger than the Ps360 and their 230 GFlops shows that.

Sub Boss said:
Come on of course size matters.

Switch is an incredible portable gaming device but its designed to work on a small battery and be smaller than a GamePad.

Xbox One was a huge power eating monster in comparison,just because it has been three years since it released doesn't mean mobile tech is close to surpass it,much less with how batteries have (not) improving.

Three years is not enough.

Ah, OK.
But reading that give me the impression the true problem is the battery, not the size.

AzaK · Feb 22, 2017

Sub Boss said:
You aren't going to see high end stuff from Nintendo anymore (not since when,the 64?) But they didn't intentionally chose the poorest parts for the Switch,that is not logical this time.

I said acceptable system with the lowest options. They'll have at target and if someone presents them with options A and B, both with hit that target they'll go "Which is cheaper?" which tends to mean (Unless older has hit the rare phase) the less modern, less capable hardware.

ShadowFox08 · Feb 22, 2017

HeelPower said:
Because it matters if Nintendo treats it as the 3DS successor.That would mean its gonna get an amazing stream of games.

As a home console,it has less chances of getting third parties due to power gap.

Power gap isn't going to be that much of an issue. For one, the power gap is looking more like switch is only roughly half as powerful as xbone if we go by digital foundary leaks. The Power gap is much closer to its competitor than the past two generations. The architecture is also very modern, and its easy to port games from PS4. So that's not going to be a huge issue. Any developer that makes an excuse about excluding making switch games due to power, is bullshitting and is mainly just worried about making a profit. The wii got a surprisingly decent amount of AAA ports from its 360/PS3 brethren. Games like Cod that ported surprisingly well, despite the wii being like like 20x weaker than a 360.

Secondly, its more about how much the console and its games sell. I have little doubt that the switch will sell like hot cakes.. Surely far more than the Wii U. But Nintendo needs to regain trust from 3rd party devs, and one way is to selling a ton of consoles.. Second requirement is to sell a noticable amount of 3rd party games--at least to make some profit out of it. As long as devs take a chance and they make a profit, they will put games on the switch.

blu · Feb 22, 2017

BriareosGAF said:
Nor easily compare mobile titles to AAA, nor lighting setups from gen7 to gen8, etc. The point isn't that the PS3 was a huge problem, but that fp16 is not a keyword you sprinkle through your shader until it's "fast enough", or a radio box you toggle in a middleware package.

While I get your point (and nobody in this thread has been arguing for indiscriminate use of fp16 in contemporary shaders - on the contrary), what you give as an negative example is perfectly doable in a carefully-devised integration test scenario, where a shader running entirely at precision X is taken, and the individual computational statements in it are demoted one by one to precision Y and passed through an automated test harness that mimics the use-cases of the shader and compared for, say, MSE vs reference results. It's not exactly rocket science (yes, I've used such a pipeline for fp64 to fp32).

Close, but there are two separate idioms. One is at the source level when authoring: "most of the lighting math is running in half but there are a few areas like the Schlick approximation for the fresnel that we promote to float, along with some parts of the filmic tonemapper." The other form is at the ISA level, where even when you want some maths to execute at half precision, the ISA may or may not support all the operations you want in half mode, and so the compiler may need to marshall types at the register level, which can reduce effective throughput.

Architectural peculiarities are just that - architectural peculiarities. If op X on arch Y cannot be reliably executed at the code-requested precision, then clearly the compiler is free to use whichever means necessary to achieve it, be that type promotion or refinement methods as Newton-Raphson. That does not mean that was not the right precision to request, or that the hw of another vendor is not perfectly capable of doing it right at the original precision. And in general, many fp ops at an uarch level can happen at a higher-than-requested precision, before returning the requested precision in the output register.

z0m3le · Feb 22, 2017

Going through that beyond3D thread and they bring up a good point about Q&A SoC testing not being done during the production stage, and the Foxconn leak's clocks and demos likely come from random sample testing on release specs.

One of the posters there seems to have some knowledge of how productions like this work.

Anyways, just thought it was interesting.

Matt · Feb 22, 2017

AzaK said:
I said acceptable system with the lowest options. They'll have at target and if someone presents them with options A and B, both with hit that target they'll go "Which is cheaper?" which tends to mean (Unless older has hit the rare phase) the less modern, less capable hardware.

You have just described engineering. It's all about the cheapest option that hits your target.

Broken Hope · Feb 22, 2017

Why does Snakepass keep coming up in comparisons other than it's coming out on both the PS4 and Switch? Some people are acting like because the graphics look to be the same on both but the PS4 is struggling to do a locked 60 fps that the Switch is somehow not that weak?

Snakepass isn't pushing the PS4 in any way and I still can't comprehend how a machine that can do a locked 30 with graphics like Horizon's can't do Snakepass at 60 fps.

Smurfman256 · Feb 22, 2017

Broken Hope said:
Why does Snakepass keep coming up in comparisons other than it's coming out on both the PS4 and Switch? Some people are acting like because the graphics look to be the same on both but the PS4 is struggling to do a locked 60 fps that the Switch is somehow not that weak?

Snakepass isn't pushing the PS4 in any way and I still can't comprehend how a machine that can do a locked 30 with graphics like Horizon's can't do Snakepass at 60 fps.

If I were to guess, it's because of CPU limitations (apparently rendering the snake in a realistic manner is hell on console CPUs). On the plus side, it does speak somewhat well of the ARM CPUs in the Switch if they can keep up with the CPUs in the other two biggies.

Pancake Mix · Feb 22, 2017

BDGAME said:
Lots of talk about how switch can be stronger than we think, but for me, that game give us a good idea of what is possible:

Snake Pass is 1080p 30fps on switch and 1080p 60fps on Xone.

That game beauty, but it's slow, with not much thing on screen and a repetitive scenery.

If Switch can't make a game like that run in the same quality of the Xbox one, then, it is weaker to that machine, don't matter how much FP16 and Nvidia GFlops help it.

As gorgeous as that game is, the fact that it's 1080p on either (and for that matter both), is not a sign of an insanely demanding game. You're right. It's a very modern machine, but it is weaker.

As long as you don't expect the big demanding (and high file size) third-party titles, then you'll probably like what you see.

Broken Hope · Feb 22, 2017

Smurfman256 said:
If I were to guess, it's because of CPU limitations (apparently rendering the snake in a realistic manner is hell on console CPUs). On the plus side, it does speak somewhat well of the ARM CPUs in the Switch if they can keep up with the CPUs in the other two biggies.

I still think it's most probably the fact that Sumo haven't shown themselves to be the most technically proficient studio in the past, I mean Little Big Planet 3 wasn't 60 fps either and I never understood why.

Pancake Mix · Feb 22, 2017

Smurfman256 said:
If I were to guess, it's because of CPU limitations (apparently rendering the snake in a realistic manner is hell on console CPUs). On the plus side, it does speak somewhat well of the ARM CPUs in the Switch if they can keep up with the CPUs in the other two biggies.

Yeah, for what they are, the CPUs of the PS4 and XB1 are certainly not the best. They're essentially a unique 8-core (doubled, so made up of two quad-core modules) tablet CPU from 2013. Those Jaguar CPUs were designed to be lower power (as in low consumption of power, like for a mobile device), so Switch should fit right in.

If there's one area where Switch and the PS4/XB1 could be closer than expected, it's in CPU power.

However...as we've seen, it's amazing what you can do with a relatively weak CPU as long as your GPU is good and you have plentiful memory. XB1 and even moreso the PS4 have the edge there by a good margin, PS4 having the best memory (GDDR5, with a little DDR3 for downloading applications connected to the low-power ARM secondary processor) and GPU (it was solidly mid-range in 2013, meaty and nothing to scoff at) of the three. I'm not going to go into the mid-generation revisions.

ughkingdom · Feb 22, 2017

Broken Hope said:
I still think it's most probably the fact that Sumo haven't shown themselves to be the most technically proficient studio in the past, I mean Little Big Planet 3 wasn't 60 fps either and I never understood why.

Always assumed it was for backwards compatibility reasons with the previous games.

I bet if they re-made the engine for PS4 specifically, but to look largely the same as LBP3, then they could manage 60fps.

ShadowFox08 · Feb 22, 2017

EleventhHourSuperpower said:
Yeah, for what they are, the CPUs of the PS4 and XB1 are certainly not the best. They're essentially a unique 8-core (doubled, so made up of two quad-core modules) tablet CPU from 2013. Those Jaguar CPUs were designed to be lower power (as in low consumption of power, like for a mobile device), so Switch should fit right in.

If there's one area where Switch and the PS4/XB1 could be closer than expected, it's in CPU power.

However...as we've seen, it's amazing what you can do with a relatively weak CPU as long as your GPU is good and you have plentiful memory. XB1 and even moreso the PS4 have the edge there by a good margin, PS4 having the best memory (GDDR5, with a little DDR3 for downloading applications connected to the low-power ARM secondary processor) and GPU (it was solidly mid-range in 2013, meaty and nothing to scoff at) of the three. I'm not going to go into the mid-generation revisions.

speaking of RAM and xbone, I'm surprised Xbone's RAM is DDR3. hmm. And its believed that the Switch has DDR4 RAM, right? Which is actually faster and more efficient than DDR3. ..

Broken Hope · Feb 22, 2017

What's your point? It's not like the XB1 relies solely on its DDR3.

Ninja Dom · Feb 22, 2017

HeelPower said:
If Nintendo would treat this as a handheld first that you can play on a TV rather than a console that you can take on the go, their situation would be much better. Its an amazing handheld.

But they're going with "homeconsole first that you can take on the go" narrative,and that's why comparisons to competitors are warranted.

I wouldn't worry about it.

blu · Feb 22, 2017

Matt said:
You have just described engineering. It's all about the cheapest option that hits your target.

Haha. So true.

Cuningas de Häme · Feb 22, 2017

Scoobie said:
If this isn't a final retail unit, why would someone have taken it apart and put these pics up, purely to troll?

Oh, it seems I quoted this yesterday but forgot to answer...

What I've read it seems that the one who uploaded the pics is from a spare parts company, they literally tear these things apart and use the parts for repairs.

Especially if the console is broken, I believe that in China it is not uncommon that cunning entrepeneurs buy these broken electronic devices straight from the factory and then sell the parts online. I mean, there is A LOT of screens and whatnot buyable from E-bay and Aliexpress etc...

Switch is clearly powerful enough to get ports from 3rd parties, and that's all what matters. We will have Steep and NBA2k18 from West, numerous games from the East.

If a company sees value in developing for Switch, they will develop for that. Ubisoft develops always for Nintendo consoles because French actually buy Ninty stuff.

Japanese devs develop for Nintendo consoles because people buy them, especially mobile consoles like Switch.

Indies want to spread their games as far as possible and everything points to Nintendo making everything as easy as possible for them.

No one is thinking that Switch is close to dudebro consoles, we are trying to estimate how far from them Switch actually is. It is purely for the sake of discussion, way of stretching thinking muscles. And it is utterly fascinating.

EDIT.

Yeah, I am an engineer in training (automation), and all the time we are told that our job is to meet the target as cheap as possible while maintaining certain level of quality.

The story about that Indian car (the funny looking one) is an awesome example of that. The owner told his engineers to make a very cheap car but it had to meet all the possible standards and safety regulations. Can't remember the name from the top of my head, it was a quite confusing name.

luchadork · Feb 22, 2017

sorry for the complete drive-by, but will this look as good as a wii u on a tv?

Deleted member 10571 · Feb 22, 2017

luchadork said:
sorry for the complete drive-by, but will this look as good as a wii u on a tv?

We haven't seen more recent Wii U shots, but it's very likely that it looks better. The viewing distance seems much improved, performance is way better and resolution too, the only thing of "concern" is some lighting issues, but that might be a general change or a build thing.

JoshuaJSlone · Feb 22, 2017

Genocyber said:
Is the Nintendo Switch physically the smallest home console?

I was recently wondering how the tablet portion of Switch compared to PS1/PS2/Wii's mini revisions, this gives me a good excuse to check.

Switch without Joy-Cons
102mm x 167.2mm x 13.9mm = 237cm^3

PSone
38 mm × 193 mm × 144 mm = 1056cm^3

PS2 Slim
23 cm x 15.2 cm x 2.8 cm = 979cm^3

Wii Mini
1.9 inches x 6.3 inches x 7.5 inches = 1471cm^3

Significantly smaller than those.

sinxtanx said:
I'd think the PSTV is smaller still

PSTV
65.0 × 105.0 × 13.6 mm = 92.8cm^3

But yep, still significantly larger than PSTV.

Laevateinn said:
PC Engine says hi

PC Engine
14 cm×14 cm×3.8 cm = 744.8 cm^3

Sorry, too huge.

luchadork said:
sorry for the complete drive-by, but will this look as good as a wii u on a tv?

Same HDMI output, but more hardware power behind it and same or better resolutions used, so it should.

Turrican3 · Feb 22, 2017

luchadork said:
sorry for the complete drive-by, but will this look as good as a wii u on a tv?

We already know Mario Kart 8 Deluxe is 1080p @60fps, so...

Hermii · Feb 22, 2017

HeelPower said:
Because it matters if Nintendo treats it as the 3DS successor.That would mean its gonna get an amazing stream of games.

As a home console,it has less chances of getting third parties due to power gap.

Its going to get an amazing stream of software because it will unify Nintendos development. Pokemon is leaked, fire emblem is confirmed, Aonuma heavily hinted that the 2d Zelda team is working on a Switch game. The Switch is going to get traditional handheld games as well as traditional console game.

Hermii · Feb 22, 2017

EleventhHourSuperpower said:
Yeah, for what they are, the CPUs of the PS4 and XB1 are certainly not the best. They're essentially a unique 8-core (doubled, so made up of two quad-core modules) tablet CPU from 2013. Those Jaguar CPUs were designed to be lower power (as in low consumption of power, like for a mobile device), so Switch should fit right in.

If there's one area where Switch and the PS4/XB1 could be closer than expected, it's in CPU power.

However...as we've seen, it's amazing what you can do with a relatively weak CPU as long as your GPU is good and you have plentiful memory. XB1 and even moreso the PS4 have the edge there by a good margin, PS4 having the best memory (GDDR5, with a little DDR3 for downloading applications connected to the low-power ARM secondary processor) and GPU (it was solidly mid-range in 2013, meaty and nothing to scoff at) of the three. I'm not going to go into the mid-generation revisions.

The foxconn cpu would be very comparable, the eurogamer cpu would be much weaker.

Edit: sorry.

KingSnake · Feb 22, 2017

z0m3le said:
Going through that beyond3D thread and they bring up a good point about Q&A SoC testing not being done during the production stage, and the Foxconn leak's clocks and demos likely come from random sample testing on release specs.

One of the posters there seems to have some knowledge of how productions like this work.

Anyways, just thought it was interesting.

Would that mean that the Foxconn leak refers to a much earlier timeline than we thought?

z0m3le · Feb 22, 2017

KingSnake said:
Would that mean that the Foxconn leak refers to a much earlier timeline than we thought?

No, he was posting over multiple days, it happened in November iirc.

KingSnake · Feb 22, 2017

z0m3le said:
No, he was posting over multiple days, it happened in November iirc.

I'm talking about the benchmarks/QA of the SoC. Because by November the mass production surely started.

z0m3le · Feb 22, 2017

KingSnake said:
I'm talking about the benchmarks/QA of the SoC. Because by November the mass production surely started.

Right, he says in his first post that he was looking at the switch that day, and then goes onto describe the demo scene and talk about how noisy it was in the factory.

The point I was making is that if this was happening in November, it was probably random sample testing they were doing, and you aren't testing the design of the product at this point, because you are already in full production in October, so the clock speeds he should be seeing/hearing are from retail specs, this is what he is reporting though.

I mean at this point there is little reason to try and confirm the Foxconn leak, he could be making up the clocks but that sounds absurd to me, it's still a possibility however. I just think that waiting a couple weeks won't kill us and someone might come along and hack the device and give us some info. Ignoring the leak at this point is probably foolish? but there are people attacking me on twitter just because they don't want to get their expectations high or want to downplay any performance the Switch does have? I'm not sure what is going on with these people now and days, it's like 8 days from now that Switch will be in the wild, hardly the time to start needing confirmation about leaks from 4+ months ago.

Also, has anyone found a Nintendo logo on the tear down? do we know if Nintendo puts its logos on devkit hardware? because it seems odd for it to be missing from retail units if this is one. Might mean Nvidia ran away with the design if it is a retail unit and only worked within a budget from Nintendo? that seems very strange to me considering Nintendo has its own hardware team.

frankie_baby · Feb 22, 2017

Would the storage being on a seperate pcb allow some nutjob hacker to replace it and get some benchmarks of what its at least capable of? (even if not what nintendo is running it at)

z0m3le · Feb 22, 2017

frankie_baby said:
Would the storage being on a seperate pcb allow some nutjob hacker to replace it and get some benchmarks of what its at least capable of? (even if not what nintendo is running it at)

In the new translation from the foxconn leaker, he says that a new board came in for the Switch, there was no details on the changes, but it is possible that the flash storage is now soldered onto the board. Otherwise, yes something like that could be possible, but not sure it would have clock data without the storage anyways.

frankie_baby · Feb 22, 2017

z0m3le said:
In the new translation from the foxconn leaker, he says that a new board came in for the Switch, there was no details on the changes, but it is possible that the flash storage is now soldered onto the board.

Ah that is possible i guess

Skittzo0413 · Feb 22, 2017

BDGAME said:
Yes I read.
The fact that the ps4 version is right now with unstable frame rate is not necessarily definitive, since the game is in development.

The rest I agree with you. Of course there are more than flops. The fact that Wii U is a 176 GFlop machine and more stronger than the Ps360 and their 230 GFlops shows that.

By the same measure, the Switch framerate shouldn't be taken as definitive since the game is still in development and was first ported about a month ago.

Broken Hope said:
Why does Snakepass keep coming up in comparisons other than it's coming out on both the PS4 and Switch? Some people are acting like because the graphics look to be the same on both but the PS4 is struggling to do a locked 60 fps that the Switch is somehow not that weak?

Snakepass isn't pushing the PS4 in any way and I still can't comprehend how a machine that can do a locked 30 with graphics like Horizon's can't do Snakepass at 60 fps.

Because that's the only example of a multiplat game that isn't DQ Heroes which we can compare between the two consoles. The fact that it's near PS4 levels (or over 50% those levels) a month after it was first ported over is incredibly impressive for the Switch. If it's struggling to maintain 60fps on the PS4 then there's obviously some sort of bottleneck, be it GPU or CPU, and it's therefore maxing the hardware in some way.

z0m3le · Feb 22, 2017

Skittzo0413 said:
By the same measure, the Switch framerate shouldn't be taken as definitive since the game is still in development and was first ported about a month ago.

Because that's the only example of a multiplat game that isn't DQ Heroes which we can compare between the two consoles. The fact that it's near PS4 levels (or over 50% those levels) a month after it was first ported over is incredibly impressive for the Switch. If it's struggling to maintain 60fps on the PS4 then there's obviously some sort of bottleneck, be it GPU or CPU, and it's therefore maxing the hardware in some way.

I think it could be single threaded performance, which PS4 potentially has the weakest of the 3. Since it is the lead platform for most developers, this isn't really much of an issue, but with a simpler game like this, it makes sense that the single threaded performance would handicap the performance because you can't offload 1 thread to multiple threads, meaning that Switch is at least on par with PS4's single threaded performance based on X1 with Eurogamer's clocks and the 30fps would give the CPU plenty of room to keep up with the PS4.

This means that you might not be looking at the game correctly, as it isn't an indicator that the Switch's GPU performs 50% as well as the PS4, that can only be speculated. Likely the PS4's CPU is hitting a bottleneck before the GPU, causing the unstable frame rate.

What we can currently expect from the Switch is ~3 PS4 CPU cores of performance, if foxconn's clocks pan out though, we could be looking at ~6 CPU cores and if they moved to A72, you'd gain another couple cores (even with just the 3 enabled for Switch. In the end, single threaded performance is most important here, and 3 cores is currently the standard for the gaming industry for code requirements, with some games not working without 3+ physical cores, even if there is hyperthreading/SMT.

Elfotografoalocado · Feb 22, 2017

I don't understand why people are so sceptical about the Foxconn leak, it got absolutely everything right except some assumptions ("The screen looks good, it must be 1080p) and it got many more details than even Eurogamer did.
Then Eurogamer releases some admittedly old information about clock speeds, in the same context as "October devkits are more powerful than the July ones".
I think it makes a lot of sense that older devkits were using A57s and Maxwell cores at 20nm and the old frequencies, and they just shrinked the node and changed the CPU for A72s at the same frequencies, with the same RAM size (4GB) and bandwidth.
And even if the machine has a full core or part of it dedicated to the OS it still has close throughput to that of PS4, more single thread performance and a decent GPU that can run its games with reduced graphics.

JonnyDBrit · Feb 22, 2017

Skittzo0413 said:
Because that's the only example of a multiplat game that isn't DQ Heroes which we can compare between the two consoles. The fact that it's near PS4 levels (or over 50% those levels) a month after it was first ported over is incredibly impressive for the Switch. If it's struggling to maintain 60fps on the PS4 then there's obviously some sort of bottleneck, be it GPU or CPU, and it's therefore maxing the hardware in some way.

Pretty much. Of course, that falls prey to the presumption that the two ports are equally demanding - something like the lava zone from the trailers would be more of a real stress test and eye opener - but since neither has come out yet, and we've only seen them in limited contexts, it's hard to completely, directly compare the two. Plus, more raw power doesn't always translate into significant scaling or smooth performance - just ask anyone trying to play the original Witcher game with today's high end hardware. If it's a case of the latter, then what it may suggest is that it's easier to fully utilise the Switch's capabilities, even if in terms of pure hardware it is significantly behind the likes of its competitor from Sony.

Skittzo0413 · Feb 22, 2017

z0m3le said:
I think it could be single threaded performance, which PS4 potentially has the weakest of the 3. Since it is the lead platform for most developers, this isn't really much of an issue, but with a simpler game like this, it makes sense that the single threaded performance would handicap the performance because you can't offload 1 thread to multiple threads, meaning that Switch is at least on par with PS4's single threaded performance based on X1 with Eurogamer's clocks and the 30fps would give the CPU plenty of room to keep up with the PS4.

This means that you might not be looking at the game correctly, as it isn't an indicator that the Switch performs 50% as well as the PS4, that can only be speculated. Likely the PS4's CPU is hitting a bottleneck before the GPU, causing the unstable frame rate.

What we can currently expect from the Switch is 3 PS4 CPU cores of performance, if foxconn's clocks pan out though, we could be looking at ~6 CPU cores and if they moved to A72, you'd gain another couple cores (even with just the 3 enabled for Switch. In the end, single threaded performance is most important here, and 3 cores is currently the standard for the gaming industry for code requirements, with some games not working without 3+ physical cores, even if there is hyperthreading/SMT.

That's a good point, though when I refer to 50% of PS4 performance I don't mean that's what the Switch will reach overall, just that it's clearly possible for it to hit that in some games like Snake Pass, which is a good sign. Maybe because Snake Pass is highly single threading dependent as you said.

People like to make comparisons about the hardware but we really only have this game and DQ Heroes to actually compare how modern games run between the two, and DQ Heroes looks a bit worse than a Wii U game in spots which heavily suggests it was a quick and unoptimized port. As we get more games announced and more footage (like DQXI) we'll be able to make better comparisons.

Elfotografoalocado said:
I don't understand why people are so sceptical about the Foxconn leak, it got absolutely everything right except some assumptions ("The screen looks good, it must be 1080p) and it got many more details than even Eurogamer did.
Then Eurogamer releases some admittedly old information about clock speeds, in the same context as "October devkits are more powerful than the July ones".
I think it makes a lot of sense that older devkits were using A57s and Maxwell cores at 20nm and the old frequencies, and they just shrinked the node and changed the CPU for A72s at the same frequencies, with the same RAM size (4GB) and bandwidth.
And even if the machine has a full core or part of it dedicated to the OS it still has close throughput to that of PS4, more single thread performance and a decent GPU that can run its games with reduced graphics.

Really the main reason is that the clocks presented by Eurogamer were apparently described by Nintendo themselves as "final for launch". It would be kind of strange for the clocks to be a good deal higher than that (especially the CPU clocks being 75% higher) when devs can only use far lower clocks for launch games. It's possible that they'll raise the clock speeds post launch but again, an extra 75% for the CPU is very odd.

Broken Hope · Feb 22, 2017

Skittzo0413 said:
If it's struggling to maintain 60fps on the PS4 then there's obviously some sort of bottleneck, be it GPU or CPU, and it's therefore maxing the hardware in some way.

So games like Mighty number 9 and Bro force are also maxing out the hardware since they also struggle to maintain 60 fps? I didn't realise the PS4 was so weak...

Except it's not and the PS4 struggling to maintain 60 fps in Snakepass likely has nothing to do with Sumo maxing out the PS4 and is just them not optimising correctly.

z0m3le · Feb 22, 2017

Skittzo0413 said:
Really the main reason is that the clocks presented by Eurogamer were apparently described by Nintendo themselves as "final for launch". It would be kind of strange for the clocks to be a good deal higher than that (especially the CPU clocks being 75% higher) when devs can only use far lower clocks for launch games. It's possible that they'll raise the clock speeds post launch but again, an extra 75% for the CPU is very odd.

If developers wanted to port over AAA titles that required more CPU performance, the change could have happened. I kind of agree that this doesn't make sense in a vacuum, the CPU boost would be hard to figure if it was just an internal decision, but if Capcom wanted to port a big game and couldn't because of CPU performance, would Nintendo change the clock for them? they already added 33% more memory and built the original classic controller with Capcom's help just for MHtri on Wii.

Skittzo0413 · Feb 22, 2017

Broken Hope said:
So games like Mighty number 9 and Bro force are also maxing out the hardware since they also struggle to maintain 60 fps? I didn't realise the PS4 was so weak...

Except it's not and the PS4 struggling to maintain 60 fps in Snakepass likely has nothing to do with Sumo maxing out the PS4 and is just them not optimising correctly.

You're right that it could be down to optimization, but the fact that it struggles to reach 60fps does indeed point to it maxing out some aspect of the hardware. That could be due to poor optimization but consider that the PS4 version has been under development for at least a year, whereas the Switch version was only ported over about a month ago.

So if we're assuming the PS4 version is not properly optimized what makes you think the Switch version is?

z0m3le said:
If developers wanted to port over AAA titles that required more CPU performance, the change could have happened. I kind of agree that this doesn't make sense in a vacuum, the CPU boost would be hard to figure if it was just an internal decision, but if Capcom wanted to port a big game and couldn't because of CPU performance, would Nintendo change the clock for them? they already added 33% more memory and built the original classic controller with Capcom's help just for MHtri on Wii.

Yeah I'm not saying it isn't possible, just that it would be a huge change which is why it's hard to just assume the Foxconn clocks will be retail. Those numbers kinda cause one to raise an eyebrow.

But yeah it could be possible and would be pretty huge if so.

z0m3le · Feb 22, 2017

Skittzo0413 said:
Yeah I'm not saying it isn't possible, just that it would be a huge change which is why it's hard to just assume the Foxconn clocks will be retail. Those numbers kinda cause one to raise an eyebrow.

But yeah it could be possible and would be pretty huge if so.

I'm not 100% sure about the clocks either, I figure we will know soon enough, so I'm not trying to confirm the leak clocks as retail, just fitting them into the conversation because ignoring them makes 0 sense, and is only for the trolls.

Having used my disclaimer above, the Foxconn leak is substantially correct about the facts it gave, for him to lie about the clocks makes no sense to me, but is a possibility no matter how confusing it is.

KingSnake · Feb 22, 2017

I don't know of any leaker who was 100% accurate. And it's not about lying, it can be about things changing over time, or some facts being misinterpreted or lost in communication / translation.

As I said previously, the leaker doesn't even say that the clocks are the clocks of to the retail unit per se. So even if those clocks are not in the retail unit he didn't lie.

z0m3le · Feb 22, 2017

KingSnake said:
I don't know of any leaker who was 100% accurate. And it's not about lying, it can be about things changing over time, or some facts being misinterpreted or lost in communication / translation.

As I said previously, the leaker doesn't even say that the clocks are the clocks of to the retail unit per se. So even if those clocks are not in the retail unit he didn't lie.

He did say that those clocks were the clocks of the Switch device though.

That is a pretty big stretch to say those clocks came from anywhere but the retail unit/spec sheet because that is all we know he had access to, and if he asked someone, that person would be giving us that information or lying to him. The reason that we can probably look at these clocks with more confidence is because they relate to the X1 chip's base frequency, the final hardware had a performance increase that is undetailed to us and battery life seems to have missed its target battery life by a very large amount, with multiple leakers stating 5 to 8 hours were the target battery life of the device and final hardware is 2.5 to 6 hours.

EDIT: "Here's some standard specs: CPU 1750 MHz, GPU 912 MHz, EMC 1600 MHz. (I remembered wrong, the GPU is actually 921 MHz. It sure looks like it [Ed: likely in reply to someone saying it's similar to some other chip], but this one is 16nm so it's probably a new product.)"

Support NeoGAF

A Nintendo Switch has been taken apart

Member

Member

Banned

Member

Member

Banned

Member

Member

Member

Member

Member

Member

Wants the largest console games publisher to avoid Nintendo's platforms.

Banned

Member

Member

Member

Copied someone else's pancake recipe

Member

Copied someone else's pancake recipe

Member

Member

Member

Member

Wants the largest console games publisher to avoid Nintendo's platforms.

Member

Member

Deleted member 10571

Unconfirmed Member

Member

Member

Member

Member

The Birthday Skeleton

Banned

The Birthday Skeleton

Banned

Member

Banned

Member

Member

Banned

Banned

Member

Member

Member

Banned

Member

Banned

The Birthday Skeleton

Banned

Similar threads