• 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.

Confirmed: The Nintendo Switch is powered by an Nvidia Tegra X1

Status
Not open for further replies.
Yes the Switch pro does seem more like a half generation upgrade like PS4 Pro and Scorprio but we're talking about a handheld here, so its a lot more significant than you think. I'd reckon we'd have to go to 7nm to reach that power with a good battery life and high power efficiency.

Its not a fantasy when you see it in games I am Setsuna, Snake Pass, and Legocity undercover. the fp16 feature narrows the gap quite a bit. This can explain why PS4 pro of Snake Pass has double the framerate AND pushing out +56% more pixels than the OG PS4 version. On paper, PS4 pro only is 2.2x more powerful in GPU power with a minor bump in CPU and bandwith, but we saw more than 3x the power of the OG PS4 displayed on the PRO. Mixed precision mode explains the extra power, as well as on the Switch as well.

Snake Pass is a indie game and the other two games are running on 7th gen hardware as well. It appears like a bad joke that a game like I am Setsuna is a benchmark for anything.

It's also interesting how games like Dragon Quest Heroes just get ignored in the "the gap isn't that big" narrative.
 

Lonely1

Unconfirmed Member
Snake Pass is a indie game and the other two games are running on 7th gen hardware as well. It appears like a bad joke that a game like I am Setsuna is a benchmark for anything.

It's also interesting how games like Dragon Quest Heroes just get ignored in the "the gap isn't that big" narrative.

Apart from Dragon Quest Heroes, which other games are a counter example? And why should indy games get disqualified? And why should DQH be considered while Legocity undercover shouldn't, as both are running in 7th gen hardware and bellow (Vita)?
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
Apart from Dragon Quest Heroes, which other games are a counter example? And why should indy games get disqualified? And why should DQH be considered while Legocity undercover shouldn't, as both are running in 7th gen hardware and bellow (Vita)?
You're talking to a certified drive-by-er. (just read on from that post)
 
Apart from Dragon Quest Heroes, which other games are a counter example? And why should indy games get disqualified? And why should DQH be considered while Legocity undercover shouldn't, as both are running in 7th gen hardware and bellow (Vita)?

You mean games like Zelda getting the "devs were just lazy" treatment in this thread and that was just an upport from a weaker system.

If a game like Setsuna which is the bottom line of games on the PS4 must run with half of the framerate. Then the narrative that the Switch holds up well because various reasons a plain fairy tale.

You're talking to a certified drive-by-er. (just read on from that post)

None, of the made statements wrong. But you are free to read to the projected costs of a single Prometheus liquid propulsion module.

But you are busy with the FP16 nonsense here.
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
None, of the made statements wrong.

But you are busy with the FP16 nonsense here.
And that spacex tread is still expecting backing up of your claims to a $1M per Arianespace launch.. Like a said, a certified drive-by-er.
 
And that spacex tread is still expecting backing up of your claims to a $1M per Arianespace launch.. Like a said, a certified drive-by-er.

Funny how I didn't claim that.

But just for what I claimed:

The target price for a Prometheus engine is 1 million euros, one-tenth the cost of the Ariane 6's liquid-oxygen and liquid-hydrogen Vulcain 2.1 engine. The Prometheus program is making extensive use of new technologies and production methods, including 3-D printing, and a large amount of technical design work already completed in France and Germany, according to an Airbus Safran Launchers presentation.

- See more at: http://spacenews.com/frances-promet...ject-gets-funding-boost/#sthash.V7mBFhaM.dpuf
 
You mean games like Zelda getting the "devs were just lazy" treatment in this thread and that was just an upport from a weaker system.

If a game like Setsuna which is the bottom line of games on the PS4 must run with half of the framerate. Then the narrative that the Switch holds up well because various reasons a plain fairy tale.



None, of the made statements wrong. But you are free to read to the projected costs of a single Prometheus liquid propulsion module.

But you are busy with the FP16 nonsense here.
What is up with people downplaying fp16? You don't think devs can use this to get more performance out of the Switch and PS4 Pro? Why do you feel that blu and other posters are overstating the possible gains?
 
What is up with people downplaying fp16? You don't think devs can use this to get more performance out of the Switch and PS4 Pro? Why do you feel that blu and other posters are overstating the possible gains?

Well, the kind of users never provided any proofs in which parts of game engines fp16 could massively improve performance.
FP16 doesn't suddenly increase total performance with the factor 2.

Mass Effect Andromeda devs claimed to gain 30% more performance with using fp 16 while rendingering the game in checkerboard mode.
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
Snake Pass is a indie game and the other two games are running on 7th gen hardware as well. It appears like a bad joke that a game like I am Setsuna is a benchmark for anything.

It's also interesting how games like Dragon Quest Heroes just get ignored in the "the gap isn't that big" narrative.



Tbh, your reasoning about "Snake Pass =Indie game" is silly, even though I agree with what you may have meant: Snake Pass is a fine looking game but one with limited area, slow pacing and smaller scale.

As for Dragon Quest Heroes, despite framerate issues, it's still interesting to see it running at 1080p docked as opposed to 720p on PS3. But it's also easy to see why it could be dismissed.
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
Do you suffer a stroke right now? What do you think what a booster is?
An Ariane 6-class rocket will deliver its payload from launch to target orbit on a single engine? Are you high?
 
Tbh, your reasoning about "Snake Pass =Indie game" is silly, even though I agree with what you may have meant: Snake Pass is a fine looking game but one with limited area, slow pacing and smaller scale.

As for Dragon Quest Heroes, despite framerate issues, it's still interesting to see it running at 1080p docked as opposed to 720p on PS3. But it's also easy to see why it could be dismissed.

Snake Pass is still a bottom line PS4 game. And even many of such ports still pay it with half the framerate and missing effects.

There is no fp16 magic which will let the Switch punches over its own weight, at least not in the way some people are expecting it.

An Ariane 6-class rocket will deliver its payload from launch to target orbit on a single engine? Are you high?

You are trying too hard to construct a gatcha moment. But just for you. Target price for one booster -> 1 million, something that will reduce the possible advantage of resuable boosters quite a lot. Never talked about the total launch costs of anything (which the boosters are just a part of them).
 
Well, the kind of users never provided any proofs in which parts of game engines fp16 could massively improve performance.
FP16 doesn't suddenly increase total performance with the factor 2.

Mass Effect Andromeda devs claimed to gain 30% more performance with using fp 16 while rendingering the game in checkerboard mode.

I see where you're coming from, but Shadowfow8 was just theorizing why we are seeing some good ports on the Switch. At this time, it is not even confirmed if those games even uses fp16. Either way, I don't think it should be completely discredited due to the little we know.

A 30% gain is actually pretty good. In the case of the PS4Pro, that pushed the system performance from 70% to 91% of Scropio's raw power.
 
Snake Pass is still a bottom line PS4 game. And even many of such ports still pay it with half the framerate and missing effects.

There is no fp16 magic which will let the Switch punches over its own weight, at least not in the way some people are expecting it.



Right but as I said, it's the shortcut "indie game = dismissed" that I adressed. Being an indie game doesnt mean a title cant push the hardware. Snake Pass definitelt isnt pushing for reasons I adressed earlier, being a smaller scale game, with fancy effects but that's all.

I also dont believe in the so called FP16 magic. Right, I can see some elements being lighter on calculation thanks to FP16, but that's all.

Snake Pass is although a good exemple on what people can expect from UE4 smaller scale titles being a non issue on Switch compared to PS4.
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
Snake Pass is still a bottom line PS4 game. And even many of such ports still pay it with half the framerate and missing effects.

There is no fp16 magic which will let the Switch punches over its own weight, at least not in the way some people are expecting it.
Ah, the 'some people' argument. 'Some people' expect 'fp16 magic' exactly to the effect of small improvements in performance - quite a few of them post in this thread.

You are trying too hard to construct a gatcha moment. But just for you. Target price for one booster -> 1 million, something that will reduce the possible advantage of resuable boosters quite a lot. Never talked about the total launch costs of anything (which the boosters are just a part of them).
I didn't have to construct anything as you've constructed it all yourself - your posts in that spacex thread are easy to follow. Now that we have put your '$1M per rocket' claim to rest, you can explain why reusable engines are not eligible to any of the price-reduction techniques projected for Arianespace's Prometheus (preferably back in the original thread for better effect).

You seem to have a habit of rushing in a thread, accuse posters of arguments clouded by fanboyism and when asked to back up claims demonstrate fundamental flows in understanding of the subject. Ironical to the brink.
 
Snake Pass is a indie game and the other two games are running on 7th gen hardware as well. It appears like a bad joke that a game like I am Setsuna is a benchmark for anything.

It's also interesting how games like Dragon Quest Heroes just get ignored in the "the gap isn't that big" narrative.

Calling snake pass just an indie game doesnt seem quite right to me, yes they are self publishing it but they arent what you'd really consider an indie developer, they normally do big games for big publishers
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
Question. How would a game look and run if coded entirely in FP16?
fp16 biggest disadvantage to fp32 in the context of game graphics is not so much the reduced precision as the reduced range. But for some titles that reduced range would not be an issue at all. And for some precision was never an issue, to boot. For instance, some 2D games would look exactly the same in entirely-fp16 shader arithmetics.
 

mrklaw

MrArseFace
To my limited understanding, it would seem FP16 could be useful for open world games to support greater draw distance, or reduce performance requirements for items further away from the player, where lower detail wouldn't be as noticeable. Would that be correct?
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
To my limited understanding, it would seem FP16 could be useful for open world games to support greater draw distance, or reduce performance requirements for items further away from the player, where lower detail wouldn't be as noticeable. Would that be correct?
No, fp16 is not a good candidate for 3d positional arithmetics due to the possible 'shimmering' and/or 'snapping' effect of vertices. Also, fp16 generally lacks range, so 'far away geometry' might be entirely out of range for an fp16, depending on content scale. Then would come z-fighting issues, which could be a can of their own worms.

fp16's forte is in the surface shading.
 

mrklaw

MrArseFace
No, fp16 is not a good candidate for 3d positional arithmetics due to the possible 'shimmering' and/or 'snapping' effect of vertices. Also, fp16 generally lacks range, so 'far away geometry' might be entirely out of range for an fp16, depending on content scale. Then would come z-fighting issues, which could be a can of their own worms.

fp16's forte is in the surface shading.

Ok interesting, thanks.
 

Panajev2001a

GAF's Pleasant Genius
fp16 biggest disadvantage to fp32 in the context of game graphics is not so much the reduced precision as the reduced range. But for some titles that reduced range would not be an issue at all. And for some precision was never an issue, to boot. For instance, some 2D games would look exactly the same in entirely-fp16 shader arithmetics.

Possibly blending could hurt if abused, but the internal 32 precision in the multiplier would still suffice (blending was not horrible on SNES).
 

Pasedo

Member
fp16 biggest disadvantage to fp32 in the context of game graphics is not so much the reduced precision as the reduced range. But for some titles that reduced range would not be an issue at all. And for some precision was never an issue, to boot. For instance, some 2D games would look exactly the same in entirely-fp16 shader arithmetics.

I was reading this somewhere as well in regards to reduced range but didn't understand what they meant by range. Is this range referring to draw distance? If so at what range would FP16 coding suffice? First person shooters for example where range isn't too far away at any point?
 
That's quite the small step compared to people who are already talking about moving to the next architecture because reasons.

Also no comment about that FP16 fantasy.

Well, the kind of users never provided any proofs in which parts of game engines fp16 could massively improve performance.
FP16 doesn't suddenly increase total performance with the factor 2.

Mass Effect Andromeda devs claimed to gain 30% more performance with using fp 16 while rendingering the game in checkerboard mode.

Nobody said the fp16 here would increase performance by a factor of two. 100% double fp16 isn't possible, because there are some things fp16 can't run that fp32 can. Its more like closer to a 50% with mixed precision(combination of fp16 and fp32). I alluded that with PS4 Pro's performance over regular PS4 for Snake Pass. I believe 56% more pixels and double the resolution isn't something PS4 Pro could do with only 2.2x the GPU power of the PS4.

Also so what if Snake Pass is an indie game. You're saying that indie devs can't push a system to make it look good? lol
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
Possibly blending could hurt if abused, but the internal 32 precision in the multiplier would still suffice (blending was not horrible on SNES).
True. But as you say, reasonable amounts of blending was something used back in the SNES days, and fp16's 11-bit mantissa is way more than what 8-bit machines had back then per channel. Actually, fp16's precision is very close to what dx8 hw used for pixels (though that was fixed-point), and that hw could do proper amounts of blending at no issue.

I was reading this somewhere as well in regards to reduced range but didn't understand what they meant by range. Is this range referring to draw distance? If so at what range would FP16 coding suffice? First person shooters for example where range isn't too far away at any point?
Range affects much more than draw distance (and as mentioned earlier, fp16 is traditionally not suited for vertex arithmetics, to boot). Things like full-frame post-processing effects like HDR tone mapping are heavily reliant on range (that's why fp16 is not a good candidate for tone mapping).

For positive-example reference (and at the risk of repeating myself), here's a video of game footage that runs _entirely_ in fp16 in the pixel shaders: https://www.youtube.com/watch?v=3OX9cLKDEdo
 
Nobody said the fp16 here would increase performance by a factor of two. 100% double fp16 isn't possible, because there are some things fp16 can't run that fp32 can. Its more like closer to a 50% with mixed precision(combination of fp16 and fp32). I alluded that with PS4 Pro's performance over regular PS4 for Snake Pass. I believe 56% more pixels and double the resolution isn't something PS4 Pro could do with only 2.2x the GPU power of the PS4.

Also so what if Snake Pass is an indie game. You're saying that indie devs can't push a system to make it look good? lol

Indie games like Snake Pass aren't pushing hardware like high budget AAA titles, yes. There is nothing wrong with it.

Also lol... that doubling the gflops numbers in fp 16 mode to compare the number with fp 32 numbers don't happen here, yeah.
 

JaffeLion

Banned
Well, it's an optimisation means - if used properly it can surely bring the pro closer to scorpio.

well, when you look at how many devs not even bring a pro patch at all, or where even new games look and run the same on PS4 and PS4 PRO, e.g.: http://www.neogaf.com/forum/showthread.php?t=1359414

I don't think devs will then go ahead and optimise this single detail for a SINGLE platform that does not even have this huge install base. DX12 on the other hand for xbox scorpio is prominent on xbox one, where scorpio even supports even more DX12 features AND PC also supports DX12... so this is much more like to be used then FP16 or whatever. but we will see!
 

wildfire

Banned
I don't mean to say that the PS4Pro will perform better overall, that's not at all what I meant. I just meant it could be a significant advantage in this area, namely the double speed FP16 processing. The same way that this FP16 advantage exists for the Switch over the XB1 but doesn't bring it to the same overall level.

Sorry if that was confusing.

Developers will go many different ways. MS stated they analyzed carefully what devs were doing with code and finely tuned their Scorpio to match those tendencies.

Also the Xbox GPU will have primitive shaders which if leveraged will drastically reduce polygon usage by as much as 80%.
 

blu

Wants the largest console games publisher to avoid Nintendo's platforms.
well, when you look at how many devs not even bring a pro patch at all, or where even new games look and run the same on PS4 and PS4 PRO, e.g.: http://www.neogaf.com/forum/showthread.php?t=1359414

I don't think devs will then go ahead and optimise this single detail for a SINGLE platform that does not even have this huge install base. DX12 on the other hand for xbox scorpio is prominent on xbox one, where scorpio even supports even more DX12 features AND PC also supports DX12... so this is much more like to be used then FP16 or whatever. but we will see!
A fair prognosis. For the record, I'm far from the thought every developer short on flops would try using fp16 on the pro.
 
Nobody said the fp16 here would increase performance by a factor of two. 100% double fp16 isn't possible, because there are some things fp16 can't run that fp32 can. Its more like closer to a 50% with mixed precision(combination of fp16 and fp32). I alluded that with PS4 Pro's performance over regular PS4 for Snake Pass. I believe 56% more pixels and double the resolution isn't something PS4 Pro could do with only 2.2x the GPU power of the PS4.

Also so what if Snake Pass is an indie game. You're saying that indie devs can't push a system to make it look good? lol

I think based on what Sebbbi (Ubisoft developer) said on the Beyond3D forums (no idea where that link is now) about ~70% of his game's code being perfectly doable in FP16, we can expect that 50% boost due to FP16 double processing to likely be a best case scenario. For Switch and PS4Pro. We may get a few outliers like some 2D exclusive games utilizing more (maybe even 100% according to blu?) but we shouldn't expect more than 50% for most games, especially multiplats.

Which isn't anything to sneeze at, but not at all anything like magic or secret sauce.

Developers will go many different ways. MS stated they analyzed carefully what devs were doing with code and finely tuned their Scorpio to match those tendencies.

Also the Xbox GPU will have primitive shaders which if leveraged will drastically reduce polygon usage by as much as 80%.

Right, all of this will depend totally on what developers choose to do, what to support, types of games, etc. And like I was saying this won't give it an overall advantage, just the FP16 advantage which could be notable in some titles.
 
I think based on what Sebbbi (Ubisoft developer) said on the Beyond3D forums (no idea where that link is now) about ~70% of his game's code being perfectly doable in FP16, we can expect that 50% boost due to FP16 double processing to likely be a best case scenario. For Switch and PS4Pro. We may get a few outliers like some 2D exclusive games utilizing more (maybe even 100% according to blu?) but we shouldn't expect more than 50% for most games, especially multiplats.

Which isn't anything to sneeze at, but not at all anything like magic or secret sauce.



Right, all of this will depend totally on what developers choose to do, what to support, types of games, etc. And like I was saying this won't give it an overall advantage, just the FP16 advantage which could be notable in some titles.

Like a few people have been explained in this thread, not every effect can be properly converted to FP16 so a game where 70% of the code can be converted doesn't tell us much if we don't actually know what the game is. Bthe examples blu posted showed had a 2D game being able to be fully FP16. And I think it was beril who said that FP16 is supported for all effects. I don't think you can even really expect 50% increase.

I think it's more realistic that if a game is going to get a switch version a dev may look at what crucial effects they wanna maintain can go FP16 with little issue. But if you had to quantify it with a number I dont see 50% increase being realistic tbh.
 

Lonely1

Unconfirmed Member
You mean games like Zelda getting the "devs were just lazy" treatment in this thread and that was just an upport from a weaker system.

If a game like Setsuna which is the bottom line of games on the PS4 must run with half of the framerate. Then the narrative that the Switch holds up well because various reasons a plain fairy tale.

Huh? You answer to my post makes no sense. And how is Setsuna bottom of the line?
 
Like a few people have been explained in this thread, not every effect can be properly converted to FP16 so a game where 70% of the code can be converted doesn't tell us much if we don't actually know what the game is. Bthe examples blu posted showed had a 2D game being able to be fully FP16. And I think it was beril who said that FP16 is supported for all effects. I don't think you can even really expect 50% increase.

I think it's more realistic that if a game is going to get a switch version a dev may look at what crucial effects they wanna maintain can go FP16 with little issue. But if you had to quantify it with a number I dont see 50% increase being realistic tbh.

I see 50% increase (AKA ~400GFLOPS to ~600GFLOPS) as the general best case scenario, not something to be expected for most or even many games. I thought that's how I presented it too.
 
Indie games like Snake Pass aren't pushing hardware like high budget AAA titles, yes. There is nothing wrong with it.

Also lol... that doubling the gflops numbers in fp 16 mode to compare the number with fp 32 numbers don't happen here, yeah.

Why aren't developers of snake pass pushing hardware? I want you to tell me where the hell you came up with that fallacious logic. You don't need to make an open world GTA/Zelda or call of duty blockbuster game hat renders a bunch of objects on the screen within a 50 yard radius in your virtual screen to push hardware.There's plenty of ways to do so. Just having a complex physics engine for movement takes up a lot of processing power.
We don't know how much exactly Snake Pass is pushing hardware, but it looks fantastic and it has a complex physics system for moving the snake. But again, there are many ways to push a console to its limits, and they certainly don't need to be made from a AAA dev team. Saying that only AAA devs can optimize consoles is fallacious. Sure, they can add more content because $$$, but it takes talent to optimize hardware, and that's everywhere.
http://www.neogaf.com/forum/showthread.php?t=1358237&highlight=snake+pass+xbox

I think based on what Sebbbi (Ubisoft developer) said on the Beyond3D forums (no idea where that link is now) about ~70% of his game's code being perfectly doable in FP16, we can expect that 50% boost due to FP16 double processing to likely be a best case scenario. For Switch and PS4Pro. We may get a few outliers like some 2D exclusive games utilizing more (maybe even 100% according to blu?) but we shouldn't expect more than 50% for most games, especially multiplats.

Which isn't anything to sneeze at, but not at all anything like magic or secret sauce.<


That theoretical increase is actually very significant. Whether it goes to resolution, framerate, textures, lighting, etc.. Every bit helps to help Switch narrow the power the gap with the twins. And that boost for the ps4 pro shows as well as I've said like 3 times already
 
Yes, Snake Pass is pushing hardware.
Can't wait to see the next bunch of pushing multiplatform titles.

That there is a lack of them in the near future is just a mere detail.
 
Yes, Snake Pass is pushing hardware.
Can't wait to see the next bunch of pushing multiplatform titles.

That there is a lack of them in the near future is just a mere detail.

I've seen you put higher-effort trolling attempts than this.

Every game, indie or otherwise, is capable of utilizing the hardware available. You're not a Developer and you don't know what's going on under the hood.
 

Lonely1

Unconfirmed Member
Yes, Snake Pass is pushing hardware.
Can't wait to see the next bunch of pushing multiplatform titles.

That there is a lack of them in the near future is just a mere detail.

Since the game has to run at sub 900p on the PS4, it seems like a very resource intensive game indeed.
 
I've seen you put higher-effort trolling attempts than this.

Every game, indie or otherwise, is capable of utilizing the hardware available. You're not a Developer and you don't know what's going on under the hood.

I don't care about Snake Pass. If you want to see Snake Pass as legimate benchmarck for future ports of multiplatform titles, okay, go for it!

The future is bright. There is nothing to argue until we see the next bunch of multiplatform titles on Switch and PS4.
 
I don't care about Snake Pass. If you want to see Snake Pass as legimate benchmarck for future ports of multiplatform titles, okay, go for it!

The future is bright. There is nothing to argue until we see the next bunch of multiplatform titles on Switch and PS4.

The only problem with this is most people seem to want to only look at the ports that fit their narrative (on both sides). Snake Pass, I am Setsuna, and Lego City U don't count...for reasons and on the other side Dragon Quest Heroes I and II don't count...again for reasons.

Edit: I'm no expert but to me it seems the truth is somewhere in the middle.
 

ModBot

Not a mod, just a bot.
We're not interested in a fan fiction thread where people try to figure out how based Sony-sama's PS4 Pro is actually more powerful than the Scorpio, especially while hijacking a Switch thread to do so. Seek a therapist if you find yourself hugging your pillow and screaming all night over hardware anxiety.
 
Status
Not open for further replies.
Top Bottom