By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Nintendo Discussion - Nintendo Switch probably more powerful than the Xbox One S, gets PS4 ports

HoloDust said:
Pemalite said:

Sure. That would be entirely possible.

However... SLI isn't efficient.
It also drives up costs.

Memory pools need to be duplicated for each chip.

Well, not saying it would be SLI, more like what Cerny did with PS4 Pro (mirror copy of PS4's GPU, one half disabled in PS4 mode, other running at original PS4 clock).

But, as I said, it's just a thought experiment, I really doubt Nintendo went berserk with tech...one can hope though.

You still need to power gate all that though, which costs transisters and die space and thus costs if you do it at a hardware level.

Sony doesn't need to worry about power consumption to the same degree as a mobile device, so they might have limited it at the Software/OS/Firmware/BIOS/Cats level.

malistix1985 said:
It proberbly won't even have more then 4GB of shared memory, even if it would match the Tflops it would still be worse resolution because of this and I doubt its going to match the Xone TFlops, actually, concidering the nvidia chips used the processor will even be slower and it won't be a octacore either, and the processor in the ps4 and xone is already pretty terrible.

It will be a 720P and under device. Tegra doesn't have the capability to drive high resolutions whilst pushing high levels of fidelity.



--::{PC Gaming Master Race}::--

Around the Network
Xen said:

Carts will almost always be faster than a HDD. No moving parts, just a stream of electrons :)

That's no universal truth. Most SSDs are faster than HDDs, but most memory cards and USB sticks are slower than HDDs. There is a wide range of speed differences for flash memory, depending on the memory itself and the interface.

F.e. some bigger Vita games had some serious loading times. Depending on the data (a few big files or many small files) and the different memory cards, the digital versions of Vita games could transfer 4 - 8 MB/s: http://www.eurogamer.net/articles/df-hardware-how-fast-are-vita-memory-cards

And loading the data from retail cartridges of Vita games was even a bit slower: https://www.reddit.com/r/vita/comments/1wwtq5/ps_vita_game_cartridge_vs_memory_card_speed_test/

The eMMC flash memory of the Wii U also had no speed advantages compared to loading times of an external HDD and there were also WiiU games with annoying loading times.

And the main reason that the loading times of DS games and most 3DS games were no problem is the smaller size of these games... less data to read.

So we will have to wait and see, how fast the loading times of the Switch are and if Nintendo opted for lower costs or for performance.



It's not difficult to see the difference in pricing between bluray and memory. You can find bluray movies on discount labels in the UK for as little as £1 which allowing for VAT, profit, manufacture, box, printing etc clearly indicate bluray's cost pence to produce. Flash memory is considerably dearer and Nintendo use specialist cartridge manufacturer Macronix which isn't going to be that cheap. Probably a few dollars per unit depending on capacity and capacity will be low.

If the development system spec info is correct the switch has 4GB of shared memory and that has to be video memory, main memory and some data storage as cartridges nowadays normally come with heavily compressed code to save on cartridge costs. The Switch only has 32GB of flash memory and cartridge games won't be able to use much of that.

If the cartridges are 32 and 64 gigabit (4 and 8 gigabytes) at launch then you will be able to download a few onto the 32GB of built in memory.

It's not even in the same ballpark as ps4 or xbox one it's still pretty much in the same area of performance as wii u, ps3 and 360. Overall about 2x their performance when you factor in graphics gflops and memory bandwidth.

However it looks like cpu performance is 38,400mips for all 4 cores at 2ghz. Assuming portable mode is lets say 1.2ghz would be 23,040mips which compares to about 9,000 for wii u, 19,200 for 360, ps3 is 35,840. The ps4 is about 38,400 mips and the xbox one slightly more. Neither xbox one or ps4 really pushed the envelope when it came to the cpu. This is just a rough guide to cpu performance not gpu. The x86 processors tend to deliver more performance due to cisc optimisations and 64bit cpu's will perform better with 64bit optimised code.

So looking at that it looks like the switch has reasonable cpu performance so it does have potential to run fairly sophisticated game engines with realistic physics etc. I wasn't expecting the cpu to score so high to be honest. However taking the low memory bandwidth of 25.6GB/s into account with the weak gpu performance your looking at something that may be able to make a stab at ps4 games but with weak horrible visuals in comparison. However those weak visuals will work within the restraints of only 4GB of memory and only 25.6GB/s of memory bandwidth.

This is all based on the so called leaked spec but it does seem likely to me. If Nintendo were buying the most capable hardware on the cheap this seems close to the best option.



So people are speculating that it will be more powerful than the Xbox One S. So with that theory it will be more expensive than the Xbox One S. It doesn't have a blu-ray drive so you can shave off $30 there but it does have a nice big 720P portable screen which you can bet costs way more than $30 to produce. I for one strongly doubt it will be close to the Xbox One in power. Just look at the size of the two consoles(even factoring in extras in the Xbox like the disk drive, internal power supply etc). Hell even comparing it to the PS4 Slim, it looks tiny compared to the Slim.



One guy who's a known console developer on Anandtech's board's says Switch is "1/3 as fast as an XBox One" (not close) and has no where near the memory bandwidth as XBox One (kinda expected since XBox One has ESRAM that wouldn't be feasible in a portable most likely).

So probably something equivalent to 433 GFLOPS for NS? That would be a little more than double a Wii U. Which sounds about right.

It's powerful for a handheld but for TV play, NS ports are going to look noticably worse than the PS4/XB1 provided the developer is interested in making the effort. 



Around the Network
Soundwave said:

One guy who's a known console developer on Anandtech's board's says Switch is "1/3 as fast as an XBox One" (not close) and has no where near the memory bandwidth as XBox One (kinda expected since XBox One has ESRAM that wouldn't be feasible in a portable most likely).

So probably something equivalent to 433 GFLOPS for NS? That would be a little more than double a Wii U. Which sounds about right.

It's powerful for a handheld but for TV play, NS ports are going to look noticably worse than the PS4/XB1 provided the developer is interested in making the effort. 

iAs you may know, the X2 pascal chip is touted at 1.5 tf, fp16 (half precision) so 750 gf ,fp32 (single prec.) is the max.  Considering Nvidia >>>>> AMD flops, performance wise brings it closer to XB1.  This lines up exactly with what  Emily declared the entire time , about NX performing similar to XB1 but power comparison being a stretch, etc.

Of course, that could be the dev kit simulated performance, so the final NS product could be clocked higher or even lower.  With the ventilation grill on NS, this thing has some mobile ower, but still hard to predict.   Again, maybe in portable mode, the thing runs on a "power saving" setting  350-500 gflops & while docked, "performance setting",  750 gflops more/less.

But I think the worst case scenario, regarding 3rd parties, NS will be a 540port system, displacing Vita. 



se7en7thre3 said:
Soundwave said:

One guy who's a known console developer on Anandtech's board's says Switch is "1/3 as fast as an XBox One" (not close) and has no where near the memory bandwidth as XBox One (kinda expected since XBox One has ESRAM that wouldn't be feasible in a portable most likely).

So probably something equivalent to 433 GFLOPS for NS? That would be a little more than double a Wii U. Which sounds about right.

It's powerful for a handheld but for TV play, NS ports are going to look noticably worse than the PS4/XB1 provided the developer is interested in making the effort. 

iAs you may know, the X2 pascal chip is touted at 1.5 tf, fp16 (half precision) so 750 gf ,fp32 (single prec.) is the max.  Considering Nvidia >>>>> AMD flops, performance wise brings it closer to XB1.  This lines up exactly with what  Emily declared the entire time , about NX performing similar to XB1 but power comparison being a stretch, etc.

Of course, that could be the dev kit simulated performance, so the final NS product could be clocked higher or even lower.  With the ventilation grill on NS, this thing has some mobile ower, but still hard to predict.   Again, maybe in portable mode, the thing runs on a "power saving" setting  350-500 gflops & while docked, "performance setting",  750 gflops more/less.

But I think the worst case scenario, regarding 3rd parties, NS will be a 540port system, displacing Vita. 

I think the developer compared the systems (XBox One and NS) directly though and he was pretty clear ... NS is not close to XBOne in real world performance.

1/3 of XBox One maybe with a lot less memory bandwidth (this is just as important as gigaflops guys). 

If Nintendo is willing to offer a supplemental dock with actual hardware in it, that could change the story, but then developers would likely have to make two versions/settings for each game. Mind you, this is now becoming the case for Sony/MS developers too, but I can see Nintendo as always being stubborn about that. 

Tegra X1/X2 are very good *mobile* chips, but mobile is worlds different from desktop/home console cards, the electrical envelopes are miles apart, NS probably at maximum can only run at 9-12 watts, and not all that can be for the GPU either, the LCD/memory/WiFi/CPU will also chew up some of that electrical budget too. An XBox One S even with a large die shrink (16nm) and more efficient design still consumes 50+ watts (it's a monster reduction from the OG XBox One that consumes 110 watts). 



Soundwave said:
se7en7thre3 said:

iAs you may know, the X2 pascal chip is touted at 1.5 tf, fp16 (half precision) so 750 gf ,fp32 (single prec.) is the max.  Considering Nvidia >>>>> AMD flops, performance wise brings it closer to XB1.  This lines up exactly with what  Emily declared the entire time , about NX performing similar to XB1 but power comparison being a stretch, etc.

Of course, that could be the dev kit simulated performance, so the final NS product could be clocked higher or even lower.  With the ventilation grill on NS, this thing has some mobile ower, but still hard to predict.   Again, maybe in portable mode, the thing runs on a "power saving" setting  350-500 gflops & while docked, "performance setting",  750 gflops more/less.

But I think the worst case scenario, regarding 3rd parties, NS will be a 540port system, displacing Vita. 

I think the developer compared the systems (XBox One and NS) directly though and he was pretty clear ... NS is not close to XBOne in real world performance.

1/3 of XBox One maybe with a lot less memory bandwidth (this is just as important as gigaflops guys). 

If Nintendo is willing to offer a supplemental dock with actual hardware in it, that could change the story, but then developers would likely have to make two versions/settings for each game. Mind you, this is now becoming the case for Sony/MS developers too, but I can see Nintendo as always being stubborn about that. 

Tegra X1/X2 are very good *mobile* chips, but mobile is worlds different from desktop/home console cards, the electrical envelopes are miles apart, NS probably at maximum can only run at 9-11 watts, and not all that can be for the GPU either, the LCD/memory/WiFi/CPU will also chew up some of that electrical budget too. An XBox One S even with a large die shrink (16nm) and more efficient design still consumes 50+ watts. 

Can you link if possible? 

But yeah, its hard to expect miracles from mobile tech, esp. when you factor in battery life, cooling & Nintendo lol.



its almost like people didnt learn their lesson from Wii U



se7en7thre3 said:
Soundwave said:

I think the developer compared the systems (XBox One and NS) directly though and he was pretty clear ... NS is not close to XBOne in real world performance.

1/3 of XBox One maybe with a lot less memory bandwidth (this is just as important as gigaflops guys). 

If Nintendo is willing to offer a supplemental dock with actual hardware in it, that could change the story, but then developers would likely have to make two versions/settings for each game. Mind you, this is now becoming the case for Sony/MS developers too, but I can see Nintendo as always being stubborn about that. 

Tegra X1/X2 are very good *mobile* chips, but mobile is worlds different from desktop/home console cards, the electrical envelopes are miles apart, NS probably at maximum can only run at 9-11 watts, and not all that can be for the GPU either, the LCD/memory/WiFi/CPU will also chew up some of that electrical budget too. An XBox One S even with a large die shrink (16nm) and more efficient design still consumes 50+ watts. 

Can you link if possible? 

But yeah, its hard to expect miracles from mobile tech, esp. when you factor in battery life, cooling & Nintendo lol.

https://forums.anandtech.com/threads/nintendo-switch-is-powered-by-nvidia.2481278/page-21

Devs name is Zlatan, he has the last two posts on the page, and the next page has reaction to his comments. He has been vetted as a developer in the past, for what that's worth, but obviously I can't know for 100% sure if this guy is legit. 

It does make sense though it's simply thermal dynamics ... you can't have a 50 watt chip comparable to a XBox One inside of a form factor that is only 1/10th the size. Even with a fan your internals will melt after like 2 hours of continous gameplay.