• 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.

The Curious Case of the Switch Foxconn Leak (Now a hardware fanfiction thread)

Here's a question. How far below 720p can a game's resolution in portable mode go before the IQ starts looking massively shitty?

I wanted to sort of test it on my current phone, which has a 5" 720p screen. But the pixel density of that is much higher than the Switch's screen. 480p wouldn't look great, but I wouldn't be surprised if someone attempted it.

How low did vita games go?
 
Dumb question here: when the Switch will be released, will it be possible to finally have precise numbers about its specs? Even the clockspeeds?
 

Zedark

Member
Dumb question here: when the Switch will be released, will it be possible to finally have precise numbers about its specs? Even the clockspeeds?
Clock speeds will requirements a jack most likely, but if we assume that either Eurogamer or Foxconn is right, we can most likely determine the type of cpu contents used and deduce from there which of the two is right. So, that requires only someone opening up the system and identifying the chips. (Please correct me if I am wrong here. I remember reading that identifying A72 vs. A57 cores was possible from the motherboard)
 
I think the biggest worry for me though is that even though docked mode will be 2.5x as more powerful as handheld, I hope that power won't just be dedicated to bumping the resolution instead of effects. As we know 720p to 1080p bump takes roughly 2.25x processing power.. I'd hope Nintendo allows devs to use remaining power would be for frame rate stability or graphical effects.. I'm more so curious if we get aAAA game like Call of Duty this Fall, and if handheld ends up being 720p(god forbid its less), then would devs make docked mode 1080p(just resolution bump) or if they could be given a choice to have it at same 720p resolution but with added graphical effects to match the PS4/Xbone version.

? I'd rather have cod on switch 720p on docked with the same graphical fidelity as Xbone/PS4 but with sacrificed resolution, than COD4 switch at 1080p resolution with considerably less graphical effects like polygons, textures, lighting, shadows, etc
Well, Zelda's docked resolution is to 900p, so it's unlikely that any dev is forced to use the extra power for just the resolution.
 
One thing interestingly nintendo quoted the battery life as between 2.5 and 6 hours and that zelda will manage about 3 so does that not suggest that some games may use a higher power mode than zelda
 
I think the biggest worry for me though is that even though docked mode will be 2.5x as more powerful as handheld, I hope that power won't just be dedicated to bumping the resolution instead of effects. As we know 720p to 1080p bump takes roughly 2.25x processing power.. I'd hope Nintendo allows devs to use remaining power would be for frame rate stability or graphical effects.. I'm more so curious if we get aAAA game like Call of Duty this Fall, and if handheld ends up being 720p(god forbid its less), then would devs make docked mode 1080p(just resolution bump) or if they could be given a choice to have it at same 720p resolution but with added graphical effects to match the PS4/Xbone version.

? I'd rather have cod on switch 720p on docked with the same graphical fidelity as Xbone/PS4 but with sacrificed resolution, than COD4 switch at 1080p resolution with considerably less graphical effects like polygons, textures, lighting, shadows, etc

Well, Zelda's docked resolution is to 900p, so it's unlikely that any dev is forced to use the extra power for just the resolution.

Yeah this is how I was gonna respond too. Zelda is kinda proof that they have no obligation to increase the resolution by 2.25x. I should also still note that docked mode isn't 100% confirmed to be a thing. It's unlikely, but it could be possible that Zelda runs at a smooth 720p in handheld mode and a choppy 900p in docked mode because the Switch doesn't (yet?) have the ability to change clock speeds depending on mode. Meaning, Zelda can render 900p at the Switch's base clock speed but doesn't use all that power when displaying to the handheld screen. This could also explain why Splatoon is still 720p in docked mode. Maybe the demo units don't have this capability? Again, this seems quite unlikely though, considering MK8 and Fast RMX, among others.

There's also Snake Pass to consider, where the visuals between the Switch and PS4 versions look pretty much identical outside of the frame rate. So a game like CoD could be the same way, though I do doubt that. UE4 games like DQXI should be like that though.
 
Well, Zelda's docked resolution is to 900p, so it's unlikely that any dev is forced to use the extra power for just the resolution.
I don't think devs are using Switch's full power for botw, and I think it's mainly because borw was made from the ground up for Wii u, and perhaps they wanted as much as an identical experience as possible on Wii u(and not have them feel shafted). up If Wii u and switch versions are identical in graphics(as Nintendo claims) and perhaps frame rate, 720p to 900p is using only a small fraction of switch's 4-5x power over Wii U's. But if it had more than enough power to run 1080p while docked, it is odd that it isn't. Perhaps there is a big bottle neck.
 
You guys heard about Unreal Engine 4.15 update? It includes the Nintendo Switch, and its pretty significant. I wonder how much it could help games.

Some of the big features include

Improved Texture Streaming: The Texture Streaming system has been optimized to reduce CPU usage, memory usage, and load times while eliminating low resolution artifacts and automatically handling limitations of varying memory budgets of different platforms.

Texture Memory Usage Improvements - The texture streaming system now handles per component visibility and computes (smaller) per material bounds for static geometry resulting in as much as a 40% reduction in texture memory usage.

Texture Loading Time Improvements - Texture load times are 2 to 3 times faster due to tracking visible mips, enabling the streamer to focus on visible textures followed by prefetch data and forced load data.

CPU Time Reduction

Game thread update time is reduced by 50% by moving dynamic component processing and other tasks to the async texture streaming task, enabling them to occur in parallel with other game thread tasks.

Texture processing stalls for streaming levels are reduced up to 98% due to amortized processing.

Low Resolution Artifacts Reduction - Mesh UV densities are now computed per material instead of per mesh and take into account LODs. Texture streaming is also now supported on a wider variety of Component types, including Particle Systems and Instanced Static Meshes.

Automatic Memory Budgeting - The Texture Streaming system now automatically adjust its behavior to deal with varying memory budgets without making manual tweaks. The system selects which textures need to be reduced using a variety of heuristics to maintain visual quality.

Also compile times are 50% faster for devs.

There's a whole bunch more that I bet the more tech savvy people can understand and comment on.

https://www.unrealengine.com/blog/unreal-engine-4-15-released
 

sits

Member
Question related to price. Not sure about USD prices, but in the UK the rumoured price was £250, and the announced price in January was £280.

Assuming there was some validity to the rumour of £250, but was raised by £30 in late-2016, could that coincide with the rumoured shift from A57 to A72?
 

Hermii

Member
I don't think devs are using Switch's full power for botw, and I think it's mainly because borw was made from the ground up for Wii u, and perhaps they wanted as much as an identical experience as possible on Wii u(and not have them feel shafted). up If Wii u and switch versions are identical in graphics(as Nintendo claims) and perhaps frame rate, 720p to 900p is using only a small fraction of switch's 4-5x power over Wii U's. But if it had more than enough power to run 1080p while docked, it is odd that it isn't. Perhaps there is a big bottle neck.

BotW runs a lot better on Switch, at least compared to the E3 build. I also believe the draw distance is significantly improved. Since its a Wii U game it probably don't utilize fp16 efficiently for example.
 

Mr Swine

Banned
So if the leaker said is true about the CPU/GPU was running at those speeds for that many days, can't it be on a 16nm node? I know it becomes increasingly hotter the smaller the CPU and GPU is when going down in a new node.
 

Zedark

Member
So if the leaker said is true about the CPU/GPU was running at those speeds for that many days, can't it be on a 16nm node? I know it becomes increasingly hotter the smaller the CPU and GPU is when going down in a new node.
Yes, that could indeed be the implication. If this leak is correct, it should be impossible to keep a system that has a form like the Switch does from malfunctioning when running for that long at those clock speeds unless the chips are placed on a more power efficient 16nm node. Also, the power consumption, even when on 16nm, of 4 A57 chips would exceed what we know the Switch can output in handheld mode (i.e. it would pass the 5.3W).

What the leak does not directly indicate, though, is that all cpu chips were running at that high clock speed at the same time. As a result, it could be possible that Nintendo is using one chip at high clock speed for things like VC emulation, while disabling some of the other CPU chips and (most of) the GPU. Unfortunately we can't say what is most likely, since each scenario has its merits.
 

PrimeBeef

Member
Question related to price. Not sure about USD prices, but in the UK the rumoured price was £250, and the announced price in January was £280.

Assuming there was some validity to the rumour of £250, but was raised by £30 in late-2016, could that coincide with the rumoured shift from A57 to A72?
There was no validity to the cheaper price. That was all speculation.
 
BotW runs a lot better on Switch, at least compared to the E3 build. I also believe the draw distance is significantly improved. Since its a Wii U game it probably don't utilize fp16 efficiently for example.
Comparisions with thrle e3 build mean nothing since so much can be improved. We won't know how Wii u version performs until launch. /:
 

Thraktor

Member
Maybe the 802.11ac Wifi support has more to do with local LAN multiplayer. I think for a game like Splatoon with 10 (apparently) players locally in a single game the bandwidth requirements might get fairly high. Also you're pushing over 10x more pixels per Switch than you were per 3DS, though I don't know if that has much of an affect on wifi bandwidth requirements.

Resolution or graphics quality shouldn't really have any effect on bandwidth requirements in this scenario, all you need to transmit are the locations, velocities and so forth for movable in-game objects (i.e. players and physics objects). To give an example, the bandwidth requirements for multiplayer MK8D shouldn't be all that much higher than they were for MK7, which managed 8-player ad-hoc wireless multiplayer just fine on that 54Mb/s connection.

The wifi solution they're using is massive overkill for any local multiplayer scenario. It's possible that they want to enable really fast game downloads for people with high-speed internet connections, but MicroSD write speeds would probably be a bigger bottleneck than the wireless interface by that point.

And regarding the July devkits, if they were labeled as 2GHz max and 1GHz max for the CPU and GPU on the hardware side, and TBD on the "for applications" side, how would developers know what the max for their games should be? Would the devkits be incapable of actually reaching those max clock speeds? I guess that would be hard for us to know without hearing from a developer who used those devkits back then.

Having to deal with varying clock speeds would be pretty common for developers that far before the launch of new hardware, I would think. I would imagine that, if clock speeds are changing on a relatively regular basis, the SDK or firmware would indicate the clock speed, rather than having it in printed material (which would then go quickly out of date).

Anyway, if the 4k SCD dock is a thing (which I'm still unsure about) then I don't think 3x A57s at 1GHz would be a very suitable CPU for running 4k applications. I guess I don't know how CPU requirements scale with GPU power but I would imagine for 4k that you'd need something a bit closer to what the PS4Pro and Scorpio will have (aka what the PS4 and XB1 have).

In theory CPU requirements shouldn't scale at all with resolution. The only difference on the CPU front is sending a "render to a 3840*2160 target" command rather than a "render to a 1920*1080 target" command. If you were designing a console to render at that resolution from the start you may want a more powerful CPU to, for example, calculate more detailed physics simulations (e.g. use a denser mesh for cloth simulation, as it's more visible), but if your intent was purely to play existing games at a higher resolution, then you wouldn't need a CPU any more powerful than what your existing device already has. Hence why Switch doesn't (as far as we know) clock up the CPU in docked mode, because the higher resolution doesn't change the CPU workload.

That said, if the 200mm² chip is merely a stand-in (as I suspect it is), then the final device could use an SoC which also includes, say, an octo-core A73 CPU, or some number of the new cores Nvidia are designing for Xavier. If what they're designing is intended to be able to operate on its own then it needs a CPU, and there's nothing stopping them including a CPU as powerful as they deem necessary.

I'll just say this.... something is off. No way this is the console devs are saying Nintendo worked with third parties with on developing. We still haven't heard any major negative tone from developers about the specs. Something is missing. No way a stock X1 or even worse is in the switch and it is a developers dream and running full unreal 4 games in a week. Something is missing... what is the most "logical" answer? my personal opinion like I said before is whatever they put in the switch is nothing to write home about... but it does its job. It can get ports of xb1 and ps4 games without much hassle.

I honestly don't think most developers are quite as concerned with all-out performance as many people tend to assume. Once you've got a relatively sensibly designed architecture without any major bottlenecks I'd say their main concern is how good the development environment, tools and support are, and by all accounts Nintendo has improved on those fronts massively with Switch.
 
Resolution or graphics quality shouldn't really have any effect on bandwidth requirements in this scenario, all you need to transmit are the locations, velocities and so forth for movable in-game objects (i.e. players and physics objects). To give an example, the bandwidth requirements for multiplayer MK8D shouldn't be all that much higher than they were for MK7, which managed 8-player ad-hoc wireless multiplayer just fine on that 54Mb/s connection.

The wifi solution they're using is massive overkill for any local multiplayer scenario. It's possible that they want to enable really fast game downloads for people with high-speed internet connections, but MicroSD write speeds would probably be a bigger bottleneck than the wireless interface by that point.

Wouldn't download play (multiple local Switches playing a game with only one copy of the game being bought) have some pretty intense wifi bandwidth requirements? I'm fairly sure we've had confirmation that some Switch games will use download play. I know the 3DS had this but Switch game files are obviously much larger.
 

nordique

Member
I'll just say this.... something is off. No way this is the console devs are saying Nintendo worked with third parties with on developing. We still haven't heard any major negative tone from developers about the specs. Something is missing. No way a stock X1 or even worse is in the switch and it is a developers dream and running full unreal 4 games in a week. Something is missing... what is the most "logical" answer? my personal opinion like I said before is whatever they put in the switch is nothing to write home about... but it does its job. It can get ports of xb1 and ps4 games without much hassle.

Or maybe the X1 was underestimated/underutilized to begin with?

The unreal 4 tech demo for it looked pretty damn good to me when nvidia revealed it way back when
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
I'll just say this.... something is off. No way this is the console devs are saying Nintendo worked with third parties with on developing. We still haven't heard any major negative tone from developers about the specs. Something is missing. No way a stock X1 or even worse is in the switch and it is a developers dream and running full unreal 4 games in a week. Something is missing... what is the most "logical" answer? my personal opinion like I said before is whatever they put in the switch is nothing to write home about... but it does its job. It can get ports of xb1 and ps4 games without much hassle.

Or maybe the X1 was underestimated/underutilized to begin with?

The unreal 4 tech demo for it looked pretty damn good to me when nvidia revealed it way back when
One really needs to consider the context for each and every case.

If game A runs happily on 2-3x Jaguars and is built on top of UE4, then said game will be very quickly to port to the Switch (fp16 built-in support and all).

If game B barely runs on 6x Jaguars and is built on top of UE4, then said game will see a heck of a porting time to the Switch, despite how easy the 'get to run at any fps' phase might be. Actually, the game may never leave that initial phase and get to its 'playabe fps' phase.
 

Hermii

Member
There's also the FP16 performance to take into account. If we assume the maximum amount of FP16 code possible in a game is 70% then that 660GFlops turns into around 1TFlop max. Which you can see in something like Snake Pass on UE4, which looks quite similar to the PS4 version at 1080p 30fps locked, where the PS4 version is 60fps unlocked.

That type of comparison wouldn't make sense if based on a flat flop to flop comparison of 1.8TF to ~400GF.

I think 60fps games would be much easier to port to a weaker cpu or system in general than 30fps games, because you can always half the framerate.
 

TLZ

Banned
Wouldn't there be some sort of NDA on those?

Only use it to play games, don't show anyone footage from a certain part of the game etc.

Oh well. They can NDA all the want. March is soon enough and won't be long before someone opens it up.
 
Wouldn't there be some sort of NDA on those?

Only use it to play games, don't show anyone footage from a certain part of the game etc.

hiphoptherobot isn't under any NDA, so if someone paid him for a replacement (and maybe a bit more for compensation) someone could open his unit up. I don't think there's anything Nintendo could do to prevent that from getting out there.
 

Hermii

Member
hiphoptherobot isn't under any NDA, so if someone paid him for a replacement (and maybe a bit more for compensation) someone could open his unit up. I don't think there's anything Nintendo could do to prevent that from getting out there.
Isn't he obligated to review whatever game they sent him the switch for? Probably Zelda.
 
Isn't he obligated to review whatever game they sent him the switch for? Probably Zelda.

He's not a reviewer and he signed no NDA, he just preordered a Switch and got it extremely early because some unnamed store broke the street date by 2 weeks. He didn't get any games though.

I'm serious, if someone (even like the guys at Digital Foundry) really wanted to pay him some money we'd be able to get a teardown within a couple days probably.
 

Hermii

Member
He's not a reviewer and he signed no NDA, he just preordered a Switch and got it extremely early because some unnamed store broke the street date by 2 weeks. He didn't get any games though.

I'm serious, if someone (even like the guys at Digital Foundry) really wanted to pay him some money we'd be able to get a teardown within a couple days probably.

Maybe he is a huge Nintendo fan, and if he orders a new Switch now he wont actually get one till who know when. Its not a given he wants to sell it. I guess they could probably do a teardown without breaking it and give it back if they really wanted to. Where is this guy from? It could take a while to send it to wherever Eurogamer is based (England?).

Its probably best to just wait 2 more weeks.
 
Maybe he is a huge Nintendo fan, and if he orders a new Switch now he wont actually get one till who know when. Its not a given he wants to sell it. I guess they could probably do a teardown without breaking it and give it back if they really wanted to. Where is this guy from? It could take a while to send it to wherever Eurogamer is based (England?).

Its probably best to just wait 2 more weeks.

Oh I'm not claiming to know if he would or wants to sell it, just that if he is so inclined he could probably make a good bit of money on it.

And it's not like he'll be able to do much with it in the next 2 weeks without any games. If it were me I'd probably sell it to someone like DF. As long as they can guarantee me another one on launch day I guess.
 
Here's a question. How far below 720p can a game's resolution in portable mode go before the IQ starts looking massively shitty?

I wanted to sort of test it on my current phone, which has a 5" 720p screen. But the pixel density of that is much higher than the Switch's screen. 480p wouldn't look great, but I wouldn't be surprised if someone attempted it.

If Switch get's a few of the bigger name, more hardware demanding third party games I could see it running them at native 540p in handheld mode and native 720p in docked mode. Those resolutions vs 720p and 900p would take a lot of strain off the limited hardware.
 

Thraktor

Member
Wouldn't download play (multiple local Switches playing a game with only one copy of the game being bought) have some pretty intense wifi bandwidth requirements? I'm fairly sure we've had confirmation that some Switch games will use download play. I know the 3DS had this but Switch game files are obviously much larger.

Yeah, that's a fair point. It would speed up single-card multiplayer downloads quite a bit. If you were actually to fill up ~3GB of RAM with assets it would potentially be quite slow otherwise.
 
If Switch get's a few of the bigger name, more hardware demanding third party games I could see it running them at native 540p in handheld mode and native 720p in docked mode. Those resolutions vs 720p and 900p would take a lot of strain off the limited hardware.
What's the processing power jump from 540p to 720p? Its not as big as 720p to 1080p, right?

If Nintendo really ends up locking that 2.5x power from portable to dock to a resolution bump only for devs, it would be even more of a waste of untapped power.
 
So apparently Digital Foundry is playing around with their Switch already, so maybe we'll hear some impressions on the tech soon. I have no idea if they could discern anything about clock speeds or even relative power (as I don't know what games they might have to compare with) but we'll hear something soon I guess.

I'm sure the NDA/embargo they signed forbids them from taking the unit apart before launch so I'm sure we won't hear anything like that.

What's the processing power jump from 540p to 720p? Its not as big as 720p to 1080p, right?

If Nintendo really ends up locking that 2.5x power from portable to dock to a resolution bump only for devs, it would be even more of a waste of untapped power.

But we established this on the last page, we know they aren't mandating that simply due to Zelda BotW not going from 720p to 1080p docked. So why even entertain that idea?
 
But we don't know why its 900p instead of 1080p, and from what we've seen, there's no indication that they are using any of that extra power in docked mode for switch whatsoever on anything outside resolution.

What we do know is that switch version isn't exactly optimized to the fullest extent, but even then, the switch should be more than enough to handle 1080p when docked. We don't even have to use mario kart's 720p to 1080p example. Everything on paper says from X1 specs and using digital foundary or foxconn specs says it all. The switch version could literally be an exact port of the Wii u version with 900p when docked, as Nintendo has claimed is the only difference outside of controls. Wouldn't be too surprised if frame rate was the same.

I feel like they are pulling the same thing they did with Zelda TP on Wii. Not pushing the switch version to its fullest capabilities either because they are rushing the game and/or maybe they want to maintain as close parity to last Gen as possible to not alienate fans who who the last Gen port/original version of the game.
 
^ That makes no sense to me. If BotW was being held back to match the Wii U version then why is it at 900p in the first place? Of course it should theoretically be able to get to 1080p from 720p, but that's exactly what I'm trying to tell you.

Since Zelda is 900p instead of 1080p, then we have undeniable proof that developers don't have to use the 2.5x GPU increase to render at 1080p. Which is what you seem to be worried about.

As for Zelda, it's very likely that they also increased draw distance, texture resolution, foliage distance/detail, LoD or other similar effects when docked because those things matter a lot more on a big screen than on a small screen. So it's likely they decided it was worth sacrificing a bit of resolution for some increased effects.
 
I think we have a misunderstanding here.
Yes of course they don't have to go full 1080p from handheld to dock. I just listed some reasons why Botw isn't fully optimized on switch based on what we've seen so far.
I'm arguing about using docked power for graphical fidelity outside of resolution. I'm hoping that the docked mode upclock isn't restricted to just a resolution boost. Because as we both know handheld clockspeeds is the bottleneck and main deciding factor that decides how close switch's games will look compared to xbone and ps4 in docked mode in graphical fidelity for the console experience, if up locking is restricted to just resolution bumps.
 

Zil33184

Member
What's the processing power jump from 540p to 720p? Its not as big as 720p to 1080p, right?

If Nintendo really ends up locking that 2.5x power from portable to dock to a resolution bump only for devs, it would be even more of a waste of untapped power.

The increase in clocks is just one part of it though, for a game like BoTW it's likely that it's already using the full memory bandwidth in portable mode and going docked won't make any difference. A resolution increase may not even be on the cards if a game is bandwidth constrained, unless we're talking about checkerboard 1080p.
 
I think we have a misunderstanding here.
Yes of course they don't have to go full 1080p from handheld to dock. I just listed some reasons why Botw isn't fully optimized on switch based on what we've seen so far.
I'm arguing about using docked power for graphical fidelity outside of resolution. I'm hoping that the docked mode upclock isn't restricted to just a resolution boost. Because as we both know handheld clockspeeds is the bottleneck and main deciding factor that decides how close switch's games will look compared to xbone and ps4 in docked mode in graphical fidelity for the console experience, if up locking is restricted to just resolution bumps.

Well, theoretically docked mode should be enough of a jump in order to get 1080p with Zelda from 720p if everything else remains constant. I don't know if RAM bandwidth would make a difference here because the textures should be the same resolution between 720p and 1080p/900p. So theoretically, the only possible reason, lack of optimization included*, that Zelda isn't 1080p is that there are other graphical fidelity boosts between handheld mode and docked mode, like increased draw distance or LoD.

That's basically what I'm trying to say- the only way to explain Zelda being 900p- in theory- is that there are other graphical boosts outside of resolution.

Again, this is all theoretical as GPU requirements don't scale perfectly with resolution, so while in theory you need 2.25x the GPU performance it may vary in practice. Also we still have no official confirmation that docked mode clocks being higher actually is a thing.

So basically, Zelda doesn't confirm this, but it's very likely that Zelda is using other graphical boosts (specifically for things like draw distance, as that is much more important on a bigger screen) meaning that games can indeed use the dock boost for other graphical boosts.

* I don't see why lack of optimization would prevent it from reaching 1080p if all of the graphics settings are identical between docked and undocked mode. You would still gain 2.5x the GPU power regardless of the optimization of the port.
 

Pokemaniac

Member
Yeah, that's a fair point. It would speed up single-card multiplayer downloads quite a bit. If you were actually to fill up ~3GB of RAM with assets it would potentially be quite slow otherwise.

Just FYI, the 3DS version of download play doesn't actually run from RAM. The download play title actually gets temporarily installed to system memory (source). I suspect Switch will be similar if it supports the feature.
 

TLZ

Banned
So apparently Digital Foundry is playing around with their Switch already, so maybe we'll hear some impressions on the tech soon. I have no idea if they could discern anything about clock speeds or even relative power (as I don't know what games they might have to compare with) but we'll hear something soon I guess.

I'm sure the NDA/embargo they signed forbids them from taking the unit apart before launch so I'm sure we won't hear anything like that.

Nice! DF on it already! Most likely they've teared it down already like starving vultures lol. Like you said though, most probably info won't come out til NDA lifted. Hopefully these 2 weeks go by quick.
 
Not sure if this has been asked already, but what can we discern about the Switch's GPU/CPU etc from a tear down?

Amount of RAM, type of CPU(A57 vs 72), how many cores the CPU has, number of how many SMs the GPU is comprised of.. The node size (16nm vs 20nm).

Someone said clockspeeds earlier could actually be found out by those above. Not sure how exactly, but I think the size of the node is a big factor. 16nm, then 921mhz from the foxconn leak is possible, though not guaranteed. I think DF can test the clockspeeds themselves through a program.
 

BuggyMike

Member
Amount of RAM, type of CPU(A57 vs 72(, how many cores the CPU has, how many SMs the GPU has.. The node size (16nm vs 20nm).

Someone said clockspeeds earlier could actually be found out by those above.

Wouldn't that detailed info be found through die shots and not only a tear down tho?
 

BuggyMike

Member
You need to tear the switch down to take die shots.. Some things require sanding off and using an electron microscope as well.

Right, I know that part, was just asking what kind of info we could get from the teardown itself since the die shots will probably take some time to come out.
 
Top Bottom