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

Forums - Nintendo Discussion - How strong will Nintendo Switch 2 be in 2027?

JRPGfan said:
bowserthedog said:

10 years is a long time.

Xbox one OG is 120 watts and only roughly twice as powerful as Switch.   Don't see how pointing today'd wattage for Xbox One X will have any relevance in a decade.

OG Xbox One is more than twice as powerfull..... its atleast 3.3 times in terms of compute. However look at stuff like Memory Bandwidth.
Switch = 25gb/s   vs  Xbox One = 68.3gb/s + 102 Gb/s (from Esram).
The Xbox also has a stronger CPU and a HDD.

A more heavy compute switch would be memory bandwidth starved, and adding better memory bandwidth would drive up such a Switch 2.0 power consumption.

Also lets keep it fair:

Xbox One X is the cloested xbox released to the Switch. So if you want to talk power effiency lets use that.
Its 6 Tflops and ~180watts.

Switch is 393 Gflops and ~18watts.

Xbox One X is ~15.3 times the power of the Switch, and uses ~10 times the power.

ergo the Xbox One X is more power effecient than the Switch is, dispite the switch going for low power low memory bandwidth ram, not haveing a HDD ect.

 

There is no way a future Switch, in the near future (even by 2027) is able to do what the Xbox One X does now.

On the CPU side its much closer.  But my main point is that 10 years is a long time.   In the past 2 years for example Galaxy phone have gone up in power be 3 times.  That's just in two years.

So by 2019 based on this standard the Switch could be on par with xbox one.  In 4 years. 4.5 terraflops. in 6 years 13.5 terraflos. In 8 years about 40 terraflops and in 10 years about 120 terraflops.

I'm not saying it will grow this fast in power.  I'm just showing you have power in mobile phones and in console grows exponentially. Typically when they come out with the new process they cut the die size in half which doubles the performance per watt plus they add new efficiencies on top of that. 

No doubt the technology will exist in 9 years to come out with a mobile device that beats the xbox one x. It remains to be seen if Nintendo creates one.



Around the Network
JRPGfan said:
bowserthedog said:

10 years is a long time.

Xbox one OG is 120 watts and only roughly twice as powerful as Switch.   Don't see how pointing today'd wattage for Xbox One X will have any relevance in a decade.

OG Xbox One is more than twice as powerfull..... its atleast 3.3 times in terms of compute. However look at stuff like Memory Bandwidth.
Switch = 25gb/s   vs  Xbox One = 68.3gb/s + 102 Gb/s (from Esram).
The Xbox also has a stronger CPU and a HDD.

A more heavy compute switch would be memory bandwidth starved, and adding better memory bandwidth would drive up such a Switch 2.0 power consumption.

Also lets keep it fair:

Xbox One X is the cloested xbox released to the Switch. So if you want to talk power effiency lets use that.
Its 6 Tflops and ~180watts.

Switch is 393 Gflops and ~18watts.

Xbox One X is ~15.3 times the power of the Switch, and uses ~10 times the power.

ergo the Xbox One X is more power effecient than the Switch is, dispite the switch going for low power low memory bandwidth ram, not haveing a HDD ect.

 

There is no way a future Switch, in the near future (even by 2027) is able to do what the Xbox One X does now.

Where did you get the 393 figure from?  Never heard it being so low.  You gotta keep in mind that in the Switch Nintendo was using 2 year old Nvidia Tegra technology because they didn't have anything newer ready at the time. Nintendo was rushed to get the system out. One would only hope that future revisions will be timed with new technology that Nvidia comes out with.  



https://browser.geekbench.com/ios-benchmarks

Take a look at how mobile processing power grows over time. In 7 years iphone went up in performance by about 18 times.



bowserthedog said:
JRPGfan said:

Where did you get the 393 figure from?  Never heard it being so low.  You gotta keep in mind that in the Switch Nintendo was using 2 year old Nvidia Tegra technology because they didn't have anything newer ready at the time. Nintendo was rushed to get the system out. One would only hope that future revisions will be timed with new technology that Nvidia comes out with.  

1) its confirmed via Xray picture analyse that Switch is a standard Tegra X1 chip.
2) We know the Tegra X1 has a configuration of 256 : 16 : 16
3) We know the Gpu Clock speeds (read from a switch) and its between 768 Mhz & 307.2 mhz (depending on docked or not)
4) We know math, how to calculate Gflops.

peak floating point throughput = shaders * 2 * clock frequency

256 shaders * 2 * 0.768  billion cycles per second (768mhz) = 393.2  Gflops

You can look this up on Wikipedia, under say Tegra, and look for the X1.
You could look it up from the Nintendo Switch Wiki, which will also reconfirm all the above.
You could read the Neogaf thread where they work there way towards the right numbers ect, and see it confirmed like this.


This isnt up for debate, its a fact.
Switch is 393 Gflops (when docked).



bowserthedog said:
https://browser.geekbench.com/ios-benchmarks

Take a look at how mobile processing power grows over time. In 7 years iphone went up in performance by about 18 times.

Its a synthetic benchmark, thats reguarly reset for a baseline against a product they pick.
Thus I doubt you could actually say its improved by a factor of 18 (in terms of performance) let alone by score.

If you have to show actual improvement, pick a benchmark that isnt synthetic.



Around the Network
JRPGfan said:
bowserthedog said:
https://browser.geekbench.com/ios-benchmarks

Take a look at how mobile processing power grows over time. In 7 years iphone went up in performance by about 18 times.

Its a synthetic benchmark, thats reguarly reset for a baseline against a product they pick.
Thus I doubt you could actually say its improved by a factor of 18 (in terms of performance) let alone by score.

If you have to show actual improvement, pick a benchmark that isnt synthetic.

I'd like you to provide me with a benchmark or something more accurate that shows how much mobile increases in power over a 10 year period of time that shows a much more modest increase. I would like you to state clearly and defend your position on how much realistically mobile goes up in performance over 10 years.



Here's another comparison for you.

The Vita was 28.4 Gflops. As you say the Switch is 393 Gflops.

So in 7 years there was an increase in gflops here of 13.83 times. If you extrapolate that over 10 years what you're looking at is an advancement of 19.76 times every 10 years. Apply that to Switch in in 2027 it puts the system at about 7.7 terraflops in 2027.

Again... I'm still waiting for you to provide some sort of fair example of how much mobile performance/power increases typically in 10 years.



bowserthedog said:
Here's another comparison for you.

The Vita was 28.4 Gflops. As you say the Switch is 393 Gflops.

So in 7 years there was an increase in gflops here of 13.83 times. If you extrapolate that over 10 years what you're looking at is an advancement of 19.76 times every 10 years. Apply that to Switch in in 2027 it puts the system at about 7.7 terraflops in 2027.

Again... I'm still waiting for you to provide some sort of fair example of how much mobile performance/power increases typically in 10 years.

Thats actually not right.... >_<

The Vita is 51 Gflops.

So in actuality it should say "in 7 years there was a increase in gflops here of 7.7 times".

 

"Apply that to Switch in in 2027 it puts the system at about 7.7 terraflops in 2027."

We went over this... thats crazy talk.
But feel free to quote this if by then, VGC is still a thing, and it turns out your right.

 

bowserthedog said:
JRPGfan said:

Its a synthetic benchmark, thats reguarly reset for a baseline against a product they pick.
Thus I doubt you could actually say its improved by a factor of 18 (in terms of performance) let alone by score.

If you have to show actual improvement, pick a benchmark that isnt synthetic.

I'd like you to provide me with a benchmark or something more accurate that shows how much mobile increases in power over a 10 year period of time that shows a much more modest increase. I would like you to state clearly and defend your position on how much realistically mobile goes up in performance over 10 years.

I dont really care enough about mobile phone performances, to know these things, can cant be arsed to go look them up.



JRPGfan said:
bowserthedog said:
Here's another comparison for you.

The Vita was 28.4 Gflops. As you say the Switch is 393 Gflops.

So in 7 years there was an increase in gflops here of 13.83 times. If you extrapolate that over 10 years what you're looking at is an advancement of 19.76 times every 10 years. Apply that to Switch in in 2027 it puts the system at about 7.7 terraflops in 2027.

Again... I'm still waiting for you to provide some sort of fair example of how much mobile performance/power increases typically in 10 years.

Thats actually not right.... >_<

The Vita is 51 Gflops.

So in actuality it should say "in 7 years there was a increase in gflops here of 7.7 times".

 

"Apply that to Switch in in 2027 it puts the system at about 7.7 terraflops in 2027."

We went over this... thats crazy talk.
But feel free to quote this if by then, VGC is still a thing, and it turns out your right.

 

bowserthedog said:

I'd like you to provide me with a benchmark or something more accurate that shows how much mobile increases in power over a 10 year period of time that shows a much more modest increase. I would like you to state clearly and defend your position on how much realistically mobile goes up in performance over 10 years.

I dont really care enough about mobile phone performances, to know these things, can cant be arsed to go look them up.

 

Its just that i'd love for you to provide anything that is based on historic performance growth in the mobile space to back up your claims. 



JRPGfan said:
curl-6 said:

By 2023 it should be well and truly possible to achieve base PS4 performance when portable, and a boost to around Xbox One X level when docked.

That doesnt sound likely.... your way too optimistic about these things.

from 150/190 Gflops (2017) -> 1840 Gflops (2023) doesnt sound likely.

Graphics effeciency wont improve x12 over the next 5-6years.

 

Also theres no way your fitting something like a Xbox One X into a small portable handheld.
Xbox One X can use upwards of 180 watts of power.
Switch when in handheld mode, uses like 16-17watts at most when in handheld mode.

You think in 5years time, we will have improve power effeciency by a factor of 12 or so too?

Get real.
Look how long it takes to go to smaller and smaller processing nodes.
How long will we be stuck at 7nm? will 5nm be soon there after? will it be enough to do what you think possible in 5-6 years? no.

I never said it would achieve Xbox One X tier power in handheld mode.

Also, FLOPS are not an accurate measurement of relative power when comparing 2 disparate chipsets with totally different architectures, released many years apart.

Last edited by curl-6 - on 20 February 2018