The semantic complications of 'demanding' games. *spawn

[Shifty Geezer]

But they do "max out" the GPU and CPU. That's why Snake Pass runs at a lower resolution than most AAA games. If it didn't "max out" the hardware, or in other words, if it wasn't hitting a bottleneck somewhere, it would be running at 1080@30 or even 1080@60.

People are arguing semantics. Both positions are correct in what 'maxxing out' a GPU means as different POV. There's no point arguing the other person is wrong. Anyone wanting to enter into a technical discussion about Switch (or GPUs and console utilisation in general) needs to drop the whole concept of 'maxxing out' and 'demanding games' and use a completely different, unambiguous terminology such as 'saturate the GPU shader units' or 'bottleneck on the GPU somewhere'.

 

[Rootax]

snake pass is not maxing out the hardware. maxing out the hardware is what AAA games do, because they have the budget, resources and skill, and the results show. you think resident evil 5, metal gear rising and dragon quest heroes are maxing out the the tegra x1 gpu. bomberman on switch launched at 30fps, 2 months later it got a patch to be 60fps, its something called optimization.

And the rez drop from 1080 to 720p....

http://www.eurogamer.net/articles/digitalfoundry-2017-super-bomberman-r-60fps-patch-report

 

[Benetanegia]

People are arguing semantics. Both positions are correct in what 'maxxing out' a GPU means as different POV. There's no point arguing the other person is wrong. Anyone wanting to enter into a technical discussion about Switch (or GPUs and console utilisation in general) needs to drop the whole concept of 'maxxing out' and 'demanding games' and use a completely different, unambiguous terminology such as 'saturate the GPU shader units' or 'bottleneck on the GPU somewhere'.

That's why i used quotation marks on "maxing out", and then said in other words...

But I don't agree both positions are correct. His interpretation of "demanding", "maxing out" or whatever you want to call it, is completely out of context here. Of course optimized code will run faster and use the hardware resources better, that's out of the question, but it doesn't mean that the hardware isn't being stressed to its maximum by unoptimized code too. And it's false the premise that optimized code will only run faster on the faster hardware and run even worse on slower hardware.

bomberman on switch launched at 30fps, 2 months later it got a patch to be 60fps, its something called optimization.

And that has nothing to do with "maxing out" the GPU in the patch, while it wasn't before the patch. Optimization does not equal "maxing out" a GPU, optimization is doing the same thing while using less resources or doing more while using the same resources (in the process you may or may not "max out" all of the GPU resources). In both cases before and after the patch, the game was most likely hitting a bottleneck somewhere, hence stressing the hardware to its limits.

 

[Goodtwin]

That's one way of looking at it. Bunnybug appears to be looking at in terms of GPU saturation. Snake Pass is likely stalling a lot compared to something highly optimised and using async compute etc.

There is only one member not accepting what Sebbbi has spent numerous post explaining. Snake Pass is relevant because its the same code running on all pieces of hardware. If a team of programmers were to go in and optimize the code, all platforms would benefit. If you want to believe that only PS4 and X1 would see big improvements from code optimization that is, well, it is just flat out ignorant.

 

[Shifty Geezer]

Optimization does not equal "maxing out" a GPU.

In one interpretation, it does, because it enables the GPU to do more in the same time. A GPU could hit its processing limits on some code, therefore being maxxed out with <100% utilisation. Then it could be optimised to do the same work in 60% of that time, whereupon the devs also add some more graphical prowess to get more functional work. Technically I'd say that's not optimisation, in agreement with you, but as the purpose of optimisation is to get more free time to do more stuff (no-one optimises their current work without making use of the freed up resources; if you're not going to make use of the freed resources, there's no point optimising), it seems reasonable to associate optimisation with getting more work out of a GPU as bunnybug is saying.

You don't have to agree with another perspective for it to exist and be valid. I mean, yes, sometimes points are crazy and illogical and to be contested, but I'm seeing a clash of POV here, certainly in the structure of the discussion. I don't know if bunnybug is saying anything about how much optimisation affects the different platforms or not - I'm not paying that much attention to the conversation. As a moderator, I'm hoping the discussion can be facilitated by opening eyes to the different interpretations and point a way forwards for the discussion, or to show people that they'd be better agreeing to disagree if they can't find a working position for all parties to discuss from.

 

[Benetanegia]

In one interpretation, it does, because it enables the GPU to do more in the same time. A GPU could hit its processing limits on some code, therefore being maxxed out with <100% utilisation. Then it could be optimised to do the same work in 60% of that time, whereupon the devs also add some more graphical prowess to get more functional work. Technically I'd say that's not optimisation, in agreement with you, but as the purpose of optimisation is to get more free time to do more stuff (no-one optimises their current work without making use of the freed up resources; if you're not going to make use of the freed resources, there's no point optimising), it seems reasonable to associate optimisation with getting more work out of a GPU as bunnybug is saying.

You don't have to agree with another perspective for it to exist and be valid. I mean, yes, sometimes points are crazy and illogical and to be contested, but I'm seeing a clash of POV here, certainly in the structure of the discussion. I don't know if bunnybug is saying anything about how much optimisation affects the different platforms or not - I'm not paying that much attention to the conversation. As a moderator, I'm hoping the discussion can be facilitated by opening eyes to the different interpretations and point a way forwards for the discussion, or to show people that they'd be better agreeing to disagree if they can't find a working position for all parties to discuss from.

I'm not disagreeing with the existence or validation of the other interpretation, I'm just saying that it does not apply to the conversation. in his last post, Goodtwin has explained why that is the case, much better an in fewer words than I would be capable of.

 

[Goodtwin]

n one interpretation, it does, because it enables the GPU to do more in the same time. A GPU could hit its processing limits on some code, therefore being maxxed out with <100% utilisation. Then it could be optimised to do the same work in 60% of that time, whereupon the devs also add some more graphical prowess to get more functional work. Technically I'd say that's not optimisation, in agreement with you, but as the purpose of optimisation is to get more free time to do more stuff (no-one optimises their current work without making use of the freed up resources; if you're not going to make use of the freed resources, there's no point optimising), it seems reasonable to associate optimisation with getting more work out of a GPU as bunnybug is saying.

Optimizing the code isn't actually changing the capabilities of the hardware though. The ceiling for the hardware capabilities didn't change, only the work load changed. The process of optimization is taking various task, and making them lighter requiring less cycles to get the desired result. Real world scenario, lets say you can bench press 200lbs 10 times, but I then lower the weight to 125lbs and you can do it 20 times. Your peak capabilities didn't change, only the effort required to press the weight off your chest changed. I can make the task of pressing the bar off your chest very easy or very challenging depending on the weight, but either way the task of moving the bar off your chest remains the same. By making the code lighter or more efficient makes the task less demanding. Unlike PC where we could just run a bunch of benchmarks and compare, we cant do that with consoles, and thus we must compare multi plat games and extrapolate how things stack up.

 

[dogen]

Been on reading gaming forums for a 15 years now, never seen anyone look at it like that. it's always the technical show cases that get mentioned as the most demanding games on hardware. the fact that anybody can call snake pass the most demanding game on ps4 is mind boggling to me, it looks a gen behind the best looking ps4 games. so games that look like they using every resource they can, to get as much power from the gpu as possible, are not as demanding because they run at 1080p or are well optimized?

Demanding isn't how good the game looks, it's how hard the hardware is working. How demanding = how much work the hardware is doing for each frame. No other definition really makes sense.

Obviously, it doesn't tell you how much bang for buck the game is getting by maxing out the hardware. But if you want to talk about that you'd say "x game isn't getting the most out of the hardware" or something.

TL DR Snake's Pass is demanding (aka a heavy workload), but it doesn't necessarily get the most out of the ps4 or switch hardware.

 

[Deleted member 13524]

Snake Pass is relevant because its the same code running on all pieces of hardware.

It is? I'm pretty much convinced that's not possible.
I'm pretty sure it's not even the same code being compiled to different target platforms, let alone the several platforms running the same code.

Yeah, if we somehow invent some "visuals-per-watt" metric, the Switch wins in Snake Pass against the PS4 and Xbone.
Is this relevant though? If any of the PS4Bone models had been designed to fit into tablets then yes, it would be.
But since the PS4Bone are made to be plugged into the wall at all times, which translates into a thousand decisions that don't take power consumption as priority (e.g. choice of voltage converters, spinning optical/magnetic discs for storage, SoCs using more area-optimized transistors and less power-optimized ones, clocks can reach a bit above their optimal values for power/performance, GDDR/DDR instead of LPDDR, etc.) then no, it doesn't matter.

Therefore, it's an irrelevant comparison.
Want to make a (perhaps less ir)relevant comparison? Have someone compare the docked/undocked Switch versions against a Surface Pro 4 with the 5W SDP Core m3, using the same resolution and settings, for example.
I happen to own that tablet, but I couldn't care less about Snake Pass to be honest.

 

[AlNom]

Need more nop.

 

[Goodtwin]

It is? I'm pretty much convinced that's not possible.
I'm pretty sure it's not even the same code being compiled to different target platforms, let alone the several platforms running the same code.

It uses Unreal 4, so even though its not identical code, it is still the same engine across multiple platforms. Snake Pass on Switch also seems to be using one step lower across the board on the settings. Pretty much every multi plat has some unique code with variances in settings on each individual platform, there is still a lot of information you can extrapolate when comparing hardware. No conclusion can come from a single game, but over time we will have enough direct comparisons to draw more accurate conclusions.

 

[iroboto]

wow wtf. @bunnybug how the hell did you manage to get sebbbi to respond to so many of your posts? I might get like 3-5 responses to my posts from him a year, and I would consider that a good year.

damn... I go to remove stain from my crusty deck for 4 hours and this happens.

 

[Deleted member 13524]

No conclusion can come from a single game, but over time we will have enough direct comparisons to draw more accurate conclusions.

That would require the existence of several of multiplatform ports of games that aren't 2D or simplistic 3D indies on both the PS4Bone and the Switch. Which is very unlikely.

 

[bunnybug]

In one interpretation, it does, because it enables the GPU to do more in the same time. A GPU could hit its processing limits on some code, therefore being maxxed out with <100% utilisation. Then it could be optimised to do the same work in 60% of that time, whereupon the devs also add some more graphical prowess to get more functional work. Technically I'd say that's not optimisation, in agreement with you, but as the purpose of optimisation is to get more free time to do more stuff (no-one optimises their current work without making use of the freed up resources; if you're not going to make use of the freed resources, there's no point optimising), it seems reasonable to associate optimisation with getting more work out of a GPU as bunnybug is saying.

Thank you.

And the rez drop from 1080 to 720p....

http://www.eurogamer.net/articles/digitalfoundry-2017-super-bomberman-r-60fps-patch-report

so now according to sebbi, bomberman is more demanding then every switch game that runs 60fps at higher resolution then 720p on switch, how ridiculous does that sound.

 

[zed]

damn... I go to remove stain from my crusty deck for 4 hours and this happens.

TMI

 

[bunnybug]

And the rez drop from 1080 to 720p....

http://www.eurogamer.net/articles/digitalfoundry-2017-super-bomberman-r-60fps-patch-report

And the rez drop from 1080 to 720p....

http://www.eurogamer.net/articles/digitalfoundry-2017-super-bomberman-r-60fps-patch-report

you guys wanna see something interesting, here is a zone of enders hd remastered on ps3, look what a little more optimization, time and budget, did for the game. it was originally only running 720p/20-30fps when released , later patched 1080p/60fps. now what's more demanding using the gpu to it's potential or running a code that bottlenecks the gpu, and doesn't really use the gpu properly. to me 1080p/60fps is more demanding cause it using way more of the gpu power. i realize were both not wrong, and it's different pov, but damn his pov just feels wrong to me lol.

 

[tongue_of_colicab]

you guys wanna see something interesting, here is a zone of enders hd remastered on ps3, look what a little more optimization, time and budget, did for the game.

Isn't that what everybody has been trying to tell you? Running optimized code will get better results than running optimized code even though resource utilization might be similar.

It's the same when you compare it with using 50% of your energy while riding a bike. On a normal bike you might do 15 ~ 20kmph doing so but on a racing bike (skinny wheels and more aerodynamic seating position resulting in less drag) you will go much faster using the same amount of energy (better optimization).

It's not about what you see on screen.

 

[bunnybug]

Isn't that what everybody has been trying to tell you? Running optimized code will get better results than running optimized code even though resource utilization might be similar.

It's the same when you compare it with using 50% of your energy while riding a bike. On a normal bike you might do 15 ~ 20kmph doing so but on a racing bike (skinny wheels and more aerodynamic seating position resulting in less drag) you will go much faster using the same amount of energy (better optimization).

It's not about what you see on screen.

not really, you have 3 members here actually agreeing with me or saying both pov are not wrong, all of them moderators. Shifty Geezer explained it best

People are arguing semantics. Both positions are correct in what 'maxxing out' a GPU means as different POV. There's no point arguing the other person is wrong. Anyone wanting to enter into a technical discussion about Switch (or GPUs and console utilisation in general) needs to drop the whole concept of 'maxxing out' and 'demanding games' and use a completely different, unambiguous terminology such as 'saturate the GPU shader units' or 'bottleneck on the GPU somewhere'.

That's one way of looking at it. Bunnybug appears to be looking at in terms of GPU saturation. Snake Pass is likely stalling a lot compared to something highly optimised and using async compute etc. To go with your real life example:

Two burly guys are digging up the road. One is digging into soft dirt, while the other is hacking at dense stony ground. The second guy is progressing slower than the first because his work is more demanding. We can conclude that the guy working slower has the more demanding work.

Two other burly guys are digging up a different road. One is slower than the other. We can conclude that the guy working slower has the more demanding work which is why he's slower, but in reality he's also digging soft dirt and is just not working as hard.

A burly guy and a scrawny guy are digging up the road. The scrawny guy's progressing slower than the burly guy, so we can conclude that the scrawny guy's work is more demanding as he's progressing slower. But in this case, he's digging the soft dirt, so in absolute terms it's less demanding, and he's just slower. But in relative terms, the work is far slower for the scrawny guy because it's harder for him to dig soft dirt than it is for the burly guy to dig stony ground.

This whole 'demanding' term has generated a protracted, nonsense discussion as no-one's clear what exactly is being talked about.

If nothing else I've come up with a non-car analogy that should get London-Boy interested.

 

[tongue_of_colicab]

Okay have it your way.

 

[HTupolev]

how ridiculous does that sound.

Depending on what you're talking about, not very. Context.

You're not making a very meaningful argument here, just mocking someone's clearly-stated and coherent semantics.

It's the same when you compare it with using 50% of your energy while riding a bike. On a normal bike you might do 15 ~ 20kmph doing so but on a racing bike (skinny wheels and more aerodynamic seating position resulting in less drag) you will go much faster using the same amount of energy (better optimization).

"Skinny wheels" actually has fairly little direct benefit to bicycle performance.

There are two major factors to tire rolling resistance:
1-As tires roll along the ground, energy is lost as they deform and rebound. Wider tires "deform" less for a given amount of vertical deflection, so if a wide and a narrow tire are set to be about equally squishy, the wider tire actually does better in this regard.
2-Suspension. When a tire rolls over surface irregularities, it can deform around them and behave as suspension for the rest of the bike, or it can propagate the deflection upward to the rest of the bicycle (and deflecting a bicycle upward over irregularities is a big waste of energy). The squishier the tire, the better it tends to perform as suspension.
Since wider tires can do (1) as well as narrower tires when squishier, rolling resistance tends to be overall more favorable for them, assuming a wide and a narrow tire are otherwise constructed similarly. (On the other hand, width tends to make tires heavier and less aerodynamic. These counterbalancing effects mean that - within something vaguely resembling reason - tire width tends to have relatively little direct effect on paved performance, much less than other aspects of a tire's construction.)

There are two main issues holding back the performance of wide road tires, and neither are a direct result of them being wide:
1-There are very few tire brands that actually make high-performance road tires in wide sizes. The vast majority of wide tires are either MTB tires, or are sluggish bombproof rubber bricks intended for long wear life and puncture-free touring or whatever.
2-Aero road racing rims are designed to form an aerodynamic shape when a tire of specific width is mounted on them, and they don't make such rims for tires more than 30mm wide.

The good news is that, in recent years, high-performance wide road tires are becoming slightly available. Road-ish bikes equipped with super-wide tires are a blast: they can be pretty comparable to a "normal" road bike on pavement, while still being perfectly happy to veer off onto gnarly gravel roads.