Forums - Nintendo Discussion - Rumor/Leak: Switch Hardware Specs, 1024 FLOP (Possibly 1TF) Device, Maxwell Architecture, Chat w/out Smartphone, Bluetooth Enabled...

m_csquare said:
Things that matter:

G: Is there any possibility for Final Fantasy XV to be released on Nintendo Switch?

HT: There are no Plans, it wouldn’t run…

Well that would require Switch to run an engine made by Square which would take months of development time overall. I'm sure it could run, it would just be missing key effects, frame rate issues, etc. in other words it wouldn't look as good or play as well. I didn't make this thread to have hope that Switch can run huge, graphic intensive AAA games, it was more of an overall dump based off of the documents procured. 



Around the Network
Alkibiádēs said:
m_csquare said:
Things that matter:

G: Is there any possibility for Final Fantasy XV to be released on Nintendo Switch?

HT: There are no Plans, it wouldn’t run…

Funny how you leave out the next sentence where he said he didn't even try it and just assumes it. 

You're just in for another disappointment



m_csquare said:
Alkibiádēs said:

Funny how you leave out the next sentence where he said he didn't even try it and just assumes it. 

You're just in for another disappointment

How? I have no desire to play a game like FF XV. 

Just calling you out on your nonsense. 



"The strong do what they can and the weak suffer what they must" - Thoukydides

Lets compare 2 scenarios:
-16FP 1 teraflop
-32FP 0.5 teraFLOP

Which would perform better for gaming graphics?
Would we use a mixed, with teraflops contage in middleway?

curl-6 said:

Yeah, I was wondering what the go was there. With so many PS2 games (including GT4 itself) failing to hit even true 480p, 1080i seemed a bridge too far.

For years people keep bringing up GT4's supposed 1080i resolution to try to make PS2 look better compared to Gamecube and Wii, but it always felt fishy.

How was GT4 failing to hit true 480p?

As far as I know, it was native 480p. 



PSN: Hynad --- NNID: Hynad81 --- Gamertag: Hynad --- Steam: Hynad81

FunFan said:
JRPGfan said:

xbox one is like 80watts? or so to reach 1.4 teraflops.

Switch is like 20? watts.

Its just not possible.

True. The Maxwell based GeForce GT 945A TDP is rated at 33w but can only reach 1024 gflops FP32. Note that actual power comsumption tends to be a bit less than the TDP rating.

There aren't any Pascal desktop cards under 75 watts yet, tho.

*edited.

NVM I always get them confused.



Predictions (from jan 2017):
This year PS4slims hit 199$ during the holidays.
This year PS4 sales go over 18,5m.

Necro-bump this 2020: http://gamrconnect.vgchartz.com/thread.php?id=229249

jonathanalis said:
Lets compare 2 scenarios:
-16FP 1 teraflop
-32FP 0.5 teraFLOP

Which would perform better for gaming graphics?
Would we use a mixed, with teraflops contage in middleway?

It depends on the programming, if you program everything in FP16..... the code will run just was well as a 1 teraflop FP32 machine.

The problem is often they cant get by with FP16, so almost everything is just FP32.

The PS4pro can do FP16, and its not really done much for it, it doesnt run like a 8 teraflop machine, compaired to the normal PS4 of 1,84.

The same will probably be true for the switch, so for all intense and purposes it ll probably run like a FP32 0.5 Teraflop machine.



Predictions (from jan 2017):
This year PS4slims hit 199$ during the holidays.
This year PS4 sales go over 18,5m.

Necro-bump this 2020: http://gamrconnect.vgchartz.com/thread.php?id=229249

Hynad said:
curl-6 said:

Yeah, I was wondering what the go was there. With so many PS2 games (including GT4 itself) failing to hit even true 480p, 1080i seemed a bridge too far.

For years people keep bringing up GT4's supposed 1080i resolution to try to make PS2 look better compared to Gamecube and Wii, but it always felt fishy.

How was GT4 failing to hit true 480p?

As far as I know, it was native 480p. 

In their analysis of Gran Turismo 5, Digital Foundry gave the native resolution of GT4 as 632x448.



curl-6 said:
Hynad said:

How was GT4 failing to hit true 480p?

As far as I know, it was native 480p. 

In their analysis of Gran Turismo 5, Digital Foundry gave the native resolution of GT4 as 632x448.

First time I've heard of that. Consider me surprised. Because the game looked crisp and ran very smoothly.



PSN: Hynad --- NNID: Hynad81 --- Gamertag: Hynad --- Steam: Hynad81

Around the Network
curl-6 said:

Yeah, I was wondering what the go was there. With so many PS2 games (including GT4 itself) failing to hit even true 480p, 1080i seemed a bridge too far.

For years people keep bringing up GT4's supposed 1080i resolution to try to make PS2 look better compared to Gamecube and Wii, but it always felt fishy.

Because people don't make the distinction between rendered resolution and output resolution.

jonathanalis said:
Lets compare 2 scenarios:
-16FP 1 teraflop
-32FP 0.5 teraFLOP

Which would perform better for gaming graphics?
Would we use a mixed, with teraflops contage in middleway?

FP16 (Half Precision) gives you better performance on nVidia's Tegra due to Packed Math. (I.E. Two FP16 operations are done on a FP32 unit.)

The caveat is a graphics quality hit. I have already been over this and demonstrated the difference betweem FP16 and FP32 in regards to image quality, so would rather not reiterate upon that again... But the difference can be massive.

The other bonus to using FP16 is battery life, FP16 rules the mobile world because of it, whilst more PC-centric platforms tend to use FP32 exclusively.

FP8 or Quarter-Precision seems to be gaining traction as well, interesting to see how that plays out in the future, should offer double the performance of FP16 if it's packed properly.

What will games use? Well. For tasks that don't need precision, FP16. For tasks that do. FP32. So there will be a mixed use.
Also depends on the developer and their vision as well.

JRPGfan said:

It depends on the programming, if you program everything in FP16..... the code will run just was well as a 1 teraflop FP32 machine.

The problem is often they cant get by with FP16, so almost everything is just FP32.

The PS4pro can do FP16, and its not really done much for it, it doesnt run like a 8 teraflop machine, compaired to the normal PS4 of 1,84.

The same will probably be true for the switch, so for all intense and purposes it ll probably run like a FP32 0.5 Teraflop machine.

Like hell it does.

You need to understand how flops, regardless of precision relates to how a game is rendered to make such claims.



---:::}}} Part of the PC Gaming Master Race. {{{:::---