• 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.

Nintendo Switch Dev Kit Stats Leaked? Cortex A57, 4GB RAM, 32GB Storage, Multi-Touch.

Status
Not open for further replies.
GPD uses Intel HD 405 clocked at 200Mhz with a burst speed up to 600Mhz. So about 50Gflops standard and 150Gflops at the top speed of 600Mhz. Not sure if it ever sustains that 600mhz for very long though (?).

That's interesting. Perhaps someone that look that up.

To be fair they didn't expect it to double as their next home console.

The closest I recall were people thinking that the handheld would be a quarter of the power of the console so that the resolution would be 540p for the handheld, and 1080p for the console. At that time, most thought that 720p would be too high for a Nintendo handheld.
 
GPD uses Intel HD 405 clocked at 200Mhz with a burst speed up to 600Mhz. So about 50Gflops standard and 150Gflops at the top speed of 600Mhz. Not sure if it ever sustains that 600mhz for very long though (?).



GPD Win's GPD runs at 400mhz if you limit it. But it can sustain 600mhz if you allow it in the bios. GPD Win may ends up slower or in the same ballpark, but let's be fair: GPD Win's library is far beyond what Switch could ever built in 3 generations.
 
Lower power and different architecture yeah sure. You could say that for xbone compared to PS4 too.

It being arm is not an issue.
ARM requires recompile at least, but the power difference might force adjustment in some of the core code especially anything related to physics. Don't know if cross play would survive in that scenario.
 

Mr Swine

Banned
Isn't there a slight possibility that the custom Tegra X1 is just a stripped down Tegra X1 without the unnecessary stuff that Nintendo will never use?
 

Donnie

Member
GPD Win's GPD runs at 400mhz if you limit it. But it can sustain 600mhz if you allow it in the bios. GPD Win may ends up slower or in the same ballpark, but let's be fair: GPD Win's library is far beyond what Switch could ever built in 3 generations.

Well flop wise it's in the same ballpark as Siwtch in handheld mode (if we assume worst case flops for Switch). But its gaming performance isn't going to be at Switch's level.
 

Easy_D

never left the stone age
Engines will probably have to be heavily modified to run well on the Switch. Engines are programmed with specific things running on each core

I guess PC development must be very hard what with there being so many different CPUs out there that devs have to manually program tasks for their individual cores :p
 

Hermii

Member
Isn't there a slight possibility that the custom Tegra X1 is just a stripped down Tegra X1 without the unnecessary stuff that Nintendo will never use?
Tradiotionally Nintendo hardware got a lot of customizations, altough many of them have been to maintain bc which it's confirmed Switch won't have.

I assume there is some custom hardware involved with the seamless transition between docked and undocked, I don't know if that's something stock hardware can do as elegantly as Switch.

Memory subsystem is also something Nintendo always pays close attention to.
 

Mokujin

Member
I assume there is some custom hardware involved with the seamless transition between docked and undocked, I don't know if that's something stock hardware can do as elegantly as Switch.

While I'm very grumpy in my beliefs that it's very close to a standard X1 (with some CPU tweaking), that is a nice observation.
 
Since this info may be more on-topic in this thread, here are some of Laura Kate Dale's quotes from her AMA:

I initially heard 3 hours max battery, no clue if that's full clock or "portable" clock speed. I've heard longer estimates since which would lead me to suspect 3 hours is if you play a game fully clocked but not connected to power, which is doable at the very least with dev kits.
I suspect we will see similar levels of graphical sacrifices as going from ultra to high to medium on PC. A little less of everything, but the same core experience.


Like pretty much every games console, the switch has gone through multiple dev kit iterations. There was one floating around in July a LOT of info on specs has been based on. Another went out to bigger devs in October. Some bigger indies are still working from the July dev kits. The Oct kit is more powerful overall than the July dev kit.
Neither is the one from this picture. https://i1.wp.com/cartridgegames.net/wp-content/uploads/2016/11/img_1029.jpg
I have seen this image debunked by numerous folk from several different developers.
 

baguetteness

Neo Member
I'm wondering if the final look of the console (both dock and portable) would be like the one we saw already on the teaser. Do you think it's doable for them?

(I'm assuming that the locked parts already in production are the inner hardware bits, being the most problematic thing to change late in development)
 

foltzie1

Member
Do we know for sure they're using DisplayPort alt mode as opposed to HDMI alt mode?

That's what the last parents said.

Also, USB-C certifying HDMI as an alt mode came relatively recently (Sept 2016 maybe) so there is some logic in outputting DisplayPort to the dock and the dock converting to HDMI.
 

antonz

Member
The Jimmy Fallon unit seemed pretty final, so I don't think there will be any significant changes.

I would expect that the Jimmy Fallon Unit was one of the demo units that will be used at the various would events. So pretty much guaranteed its finalized hardware
 
Since this info may be more on-topic in this thread, here are some of Laura Kate Dale's quotes from her AMA:

I don't see how this devkit report makes sense if the July devkits were stock TX1s and the final hardware is supposed to be severely downclocked from those...

Why would they release a devkit more powerful than the previous one which is already far more powerful than the final unit? Maybe the October ones had some of the customizations? Like an extra SM, CPU core, more RAM? Would any of that make sense?
 

Mpl90

Two copies sold? That's not a bomb guys, stop trolling!!!
I would expect that the Jimmy Fallon Unit was one of the demo units that will be used at the various would events. So pretty much guaranteed its finalized hardware

It had a European Certification mark behind it. So, yes, it had to be final.
 

Oregano

Member
I don't see how this devkit report makes sense if the July devkits were stock TX1s and the final hardware is supposed to be severely downclocked from those...

Why would they release a devkit more powerful than the previous one which is already far more powerful than the final unit? Maybe the October ones had some of the customizations? Like an extra SM, CPU core, more RAM? Would any of that make sense?

I'm not sure if there actually is any of that sort of customization. So far no one's come out to say that the X1 stats are wrong or anything, just DF saying it has some Pascal features...
 

KingSnake

The Birthday Skeleton
Since this info may be more on-topic in this thread, here are some of Laura Kate Dale's quotes from her AMA:

And again another piece of the puzzle doesn't fit.

The Oct kit is more powerful overall than the July dev kit

I assume the Oct kit came also with the clocks info that DF posted, so how the hell this fits together?

The July devkits were even more underclocked? Despite the loud fan?

Or are the CPU and GPU clocks already old info and that was revised with the October devkits? But then Laura wouldn't confirm the DF findings.
 
Laurakbuzz said that the demo units have no ethernet.
That's the Switch itself, right? The patent says the dock receives wire internet but can wirelessly transmit to the Switch.

Or, did she say the dock doesn't have it either?

Edit:

The patents talk about ethernet connectivity, not ethernet connection to the dock. The only ports in dock in the patents are 3 USB ports and 1 HDMI.
Oh, gotcha.
 

Jackano

Member
I can't believe it looks as if there's no ethernet.
FFS Nintendo, in 2017, really?

Always bogs me to read that. Wires are so 20th century. Even headphones haven't wires nowadays.

If you're so elitist on speed and response time things, I don't understand why you're playing on consoles, specifically Nintendo ones and on HDTV having frame lag.
 

ecosse_011172

Junior Member
Always bogs me to read that. Wires are so 20th century. Even headphones haven't wires nowadays.

If you're so elitist on speed and response time things, I don't understand why you're playing on consoles, specifically Nintendo ones and on HDTV having frame lag.

Yes, if you give a shit about any type of quality you can only game on an ultimate PC and monitor with the fastest monitor in production.
If you care even remotely about anything to do with cars, why the fuck would you drive a VW, Porsche or STFU.
 

Dakhil

Member
And again another piece of the puzzle doesn't fit.

The Oct kit is more powerful overall than the July dev kit

I assume the Oct kit came also with the clocks info that DF posted, so how the hell this fits together?

The July devkits were even more underclocked? Despite the loud fan?

Or are the CPU and GPU clocks already old info and that was revised with the October devkits? But then Laura wouldn't confirm the DF findings.

I guess it has to do with this tidbit from Eurogamer.

Eurogamer said:
We're also told that Switch has bespoke customisations that may involve pulling in other Pascal optimisations. And it's also worth noting that at the nuts and bolts level, Pascal and Maxwell are already very similar.

I'm guessing that the Pascal features in the customised Tegra chip in the final Nintendo Switch hardware enables the customised Tegra chip to be as powerful (if slightly more powerful) than the Tegra X1 at lower clock speeds. But I could be wrong, considering that the final customised Tegra chip in the final Nintendo Switch hardware may use a 20nm processing technology (though what type of processing technology customised Tegra chip in the final Nintendo Switch hardware uses, if I recall correctly, is unconfirmed).
 

Schnozberry

Member
I'm guessing that the Pascal features in the customised Tegra chip in the final Nintendo Switch hardware enables the customised Tegra chip to be as powerful (if slightly more powerful) than the Tegra X1 at lower clock speeds. But I could be wrong, considering that the final customised Tegra chip in the final Nintendo Switch hardware may use a 20nm processing technology (though what type of processing technology customised Tegra chip in the final Nintendo Switch hardware uses, if I recall correctly, is unconfirmed).

There is no way whatever pascal features would make the Tegra X1 faster than Pascal at lower clocks. Nvidia's biggest advantage in going to Pascal was the ability to run faster clocks due to the change in lithography.

If customizations were made that made it run faster at lower clocks, it would be adding additional CUDA and/or ARM Cores.
 
If I got the infos right somebody posted on a microblogsite some supposed hardware specs and we have no clue who that person is and if it was reliable in the past? Way to shady to talk about it yet.


This does go in line with Nate Drake's info about the battery being 5-8 hours long and it being pascal(both separate rumors)

Is a 4219 mAh battery on Switch even possible or feasible though?
 
This does go in line with Nate Drake's info about the battery being 5-8 hours long and it being pascal(both separate rumors)

Is a 4219 mAh battery on Switch even possible or feasible though?

My cellphone is 5inches, slim and it has a 3500mah battery. It's more than feasible. I think Nintendo should have gone with a bigger battery.
 

ggx2ac

Member
Did anyone mention this already? : https://mynintendonews.com/2016/12/26/rumour-another-set-of-potential-nintendo-switch-specs-leaked/

Both of their sources are from China, but the one from Baidu was removed.

Don't fall for it.

Remember this thread?

http://m.neogaf.com/showthread.php?t=1283573&page=1

Well, as a Chinese, I found some one in that a9vg thread just points out this image is from somebody's yy(imagination) in tieba.baidu.com(another chinese form like reddit)

It was later pointed out in another NX thread that it was fake with the intent to fool people worldwide since it was doing something similar to the PS4 Slim and Pro leaks.
 
I was expecting a 4000mAh battery minimum. Otherwise you're not going to be able to power Wii U level stuff on the go for at least 3 hours seeing as phones have slightly smaller batteries than that nowadays and die within two to three hours when running less demanding games.
 
Sorry if this has been discussed already, but random thought: since Switch games will presumably use SD cards, how should load times generally compare to, say, PS4 games loading off a stock 5400RPM HDD? Like for multiplat games.

Idea came up because I'm save scumming Dishonored 2 on PS4 and it's taking forever to save/load. Faster load times could potentially sway me to third party games on Switch even if the graphics take a bit of a hit.
 
Sorry if this has been discussed already, but random thought: since Switch games will presumably use SD cards, how should load times generally compare to, say, PS4 games loading off a stock 5400RPM HDD? Like for multiplat games.

Compared to your average 2.5" 5400 rpm hdd it should be about a wash for sequential and anywhere from much worse to much better in randoms. I think they probably will make sure that fully random 4k reads can't go slower than a 5400 rpm hdd though, since seek times usually aren't a problem
 
Compared to your average 2.5" 5400 rpm hdd it should be about a wash for sequential and anywhere from much worse to much better in randoms. I think they probably will make sure that fully random 4k reads can't go slower than a 5400 rpm hdd though, since seek times usually aren't a problem

Can't developers intentionally place more critical parts of data on certain sectors to increase performance like how you can put certain parts of data sequentially on a CD/DVD for improved performance?
 
I was expecting a 4000mAh battery minimum. Otherwise you're not going to be able to power Wii U level stuff on the go for at least 3 hours seeing as phones have slightly smaller batteries than that nowadays and die within two to three hours when running less demanding games.

Phones are also pushing 2k resolution instead of 720p tho.
 

FyreWulff

Member
Can't developers intentionally place more critical parts of data on certain sectors to increase performance like how you can put certain parts of data sequentially on a CD/DVD for improved performance?

No. Mostly because the hard drive controller has to account for bad sectors (and blocks on SSD), so even if you de-fragged the entire drive in software, the drive controller is actually hiding the underlying fragmentation from you, and still has the same seek costs.

Normally to nicely put a game's data all in order you'd need to run a very long defragmentation process to re-arrange every single game on the drive every time you installed a new one.
 
Status
Not open for further replies.
Top Bottom