How the PlayStation 4 is better than a PC

[BallaTheFeared]

That wasn't what I took from the full interview.

Compute was going to be used to cull unseen portions of the screen before they were rendered (as far as I could tell.) Because you were able to only render what was necessary, the perceived performance would (could) be much higher.

The article mentioned this would likely only be optimized for in exclusive titles and only a year or two into the consoles lifetime, but would keep its perceived level of performance high for a longer period of time.

Because of HSA and the optimizations made to GCN for this application, the PS4 architect was certain this type of application performance optimization would not be available on other platforms for some time.

This seems to dovetail into their desire for longer life cycles (the ability to extract a higher level of perceived performance from current gen hardware vs other competing platforms.)

It seems clever and interesting to me. For folks claiming the PS4 is boring, I think it certainly has some interesting aspects. Like the Cell, there is some exciting possibilities. Unlike the Cell, the PS4 will be a solid piece of hardware straight away regardless.

So a more powerful form of z-culling?


It will be nice to see some actual comparisons once the hardware is ready, the UE4 demo was pretty clear in the fact that the PS4 won't compete with a 680 in raw power, as is known since in raw power the PS4 is the lesser machine.

 

[blckgrffn]

So a more powerful form of z-culling?


It will be nice to see some actual comparisons once the hardware is ready, the UE4 demo was pretty clear in the fact that the PS4 won't compete with a 680 in raw power, as is known since in raw power the PS4 is the lesser machine.

""There are a broad variety of techniques we've come up with to reduce the vertex bottlenecks, in some cases they are enhancements to the hardware," said Cerny. "The most interesting of those is that you can use compute as a frontend for your graphics."

This technique, he said, is "a mix of hardware, firmware inside of the GPU, and compiler technology. What happens is you take your vertex shader, and you compile it twice, once as a compute shader, once as a vertex shader. The compute shader does a triangle sieve -- it just does the position computations from the original vertex shader and sees if the triangle is backfaced, or the like. And it's generating, on the fly, a reduced set of triangles for the vertex shader to use. This compute shader and the vertex shader are very, very tightly linked inside of the hardware."

It's also not a hard solution to implement, Cerny suggested. "From a graphics programmer perspective, using this technique means setting some compiler flags and using a different mode of the graphics API. So this is the kind of thing where you can try it in an afternoon and see if it happens to bump up your performance.""

http://www.gamasutra.com/view/feature/191007/inside_the_playstation_4_with_mark_.php?page=3

On moving the bar on bottlenecks. Like I said, I don't know enough to determine if this is hype-BS or not, but general feedback seemed positive.

 

[Olikan]

It seems clever and interesting to me. For folks claiming the PS4 is boring, I think it certainly has some interesting aspects. Like the Cell, there is some exciting possibilities. Unlike the Cell, the PS4 will be a solid piece of hardware straight away regardless.

it's a clever hardware for sure, very, VERY balanced in any metric that i can imagine.....but very boring too
PS4:
vanilla cpu cores, vanilla GCN 2.0....the only part interesting is hNUMA

xbox 720 at least seems to have a interesting hw blocks...the "move engines"

 

[cmdrdredd]

it's a clever hardware for sure, very, VERY balanced in any metric that i can imagine.....but very boring too
PS4:
vanilla cpu cores, vanilla GCN 2.0....the only part interesting is hNUMA

xbox 720 at least seems to have a interesting hw blocks...the "move engines"

Too bad the software Microsoft pushes out isn't anywhere near the quality. When you get down to it all the beat hardware in the world means very little with no software to make use of it and be quality at the same time.

 

[PrincessFrosty]

Carmack said that because his game Rage sucked as in it wasn't optimized for PC's at all and he is now whining about it. But its his problem as they didn't offer low settings for Rage so even mediocre PC's were forced to play at basically max settings.

Usually I'm a big fan of Carmacks, I can listen to him talk about tech literally all day, but feel he made some big mistakes with RAGE on the PC.

First of all the entire addition of mega texture was really just a way of getting around the consoles inherent lack of memory, due to the large amount of RAM and vRAM present in most AAA capable gaming PCs it's more than sufficient to not need mega texture technology.

He admitted in several interviews that he regrets targeting the technology at the console generation and neglecting PCs. Lets face it the resources and architectures of both platforms are different and tailoring tech to one and porting it to another is a recipe for disaster on the ported platform.

It resulted in RAGE looking no better than most other recent game engines which are more than capable of streaming in level content (unreal engine for example) but also having all the drawbacks of megatexture technology.

Funny enough that today the situation is much better, with SSDs we lack the seek time penalty of pulling data from disc/disk, and we also have extremely high transfer rates especially when coupled with fast CPUs and high bandwidth PCI-E bus and large memory caches.

If you look at the internals of the RAGE engine and the kind of settings you can tweak in order to force the engine to use more memory when you have it, you realise the engine (and the game) simply weren't made to work well on quality hardware out of the box.

I'm glad this is a problem that hardware vendors are fixing lower level, the upcoming introduction of unified vRAM and system RAM should go a long way to knocking crap like this on the head, when I have a game with a 20gb texture dump and a system with 16Gb of largely unused system RAM and 2Gb of vRAM we should NEVER EVER see any kind of bottleneck loading resources. Hopefully that'll get better over the next few years, there's no reason why we shouldn't just be dumping huge swaths of resources to system RAM and then streaming that quickly to vRAM as/when needed, we've had a gross excess of system RAM for many years now.

 

[futurefields]

Galego are you sure the PC UE4 demo is running lower res than PS4 version?

 

[galego]

I'm not ignoring anything, I'm stating the gpu in the 360 and in PS4 are what they are, there is no magic performance gain. I can take out my 9800 GT and easily beat anything consoles can put out. That's not even a doubling in performance, and there is already hardware out now on the PC that is faster vs the 360 than what G80 was vs the 360.

PS3 GPU: 170 GFLOP

9800 GT: 504 GFLOP

Your card has triple performance than a PS3. Moreover your card has between 2 and 4 times more VRAM than the PS3, which generates extra gain in performance. Add that your PC has much much more RAM as well (PS3 has only 256 MB RAM). Finally consider that the PS3 was incredibly difficult to program (only a fraction of its peak power was used by 'average' programmers) and your attempt to argue is gone.

Try to run the same games using only 256 RAM and 256 VRAM in your PC and say me about how you beat nothing.

 

[galego]

Perhaps some things get lost in translation but following the thread from the first page what I see is that the PS4 will deliver 2~3x performance without the OS overhead that PC's have to suffer, in other words for all the raw performance metrics being equal a PS4 should theoretically deliver much better results than an equivalent PC based gaming system !

Yes, as a leading game developer wrote in twitter:

Consoles run 2x or so better than equal PC hardware, but it isn’t just API in the way, focus a single spec also matters.

https://twitter.com/ID_AA_Carmack/status/50277106856370176

 

[Pottuvoi]

On moving the bar on bottlenecks. Like I said, I don't know enough to determine if this is hype-BS or not, but general feedback seemed positive.

Sounds a lot what developers did on ps3, SPUs were used to prune all vertex data that was sent to RSX. (if polygon didn't contribute to image it was culled. (including a polygon which was visible, but didn't hit any MSAA sample points..))

 

[BallaTheFeared]

PS3 GPU: 170 GFLOP

9800 GT: 504 GFLOP

Your card has triple performance than a PS3. Moreover your card has between 2 and 4 times more VRAM than the PS3, which generates extra gain in performance. Add that your PC has much much more RAM as well (PS3 has only 256 MB RAM). Finally consider that the PS3 was incredibly difficult to program (only a fraction of its peak power was used by 'average' programmers) and your attempt to argue is gone.

Try to run the same games using only 256 RAM and 256 VRAM in your PC and say me about how you beat nothing.

Luckily I was speaking on the 360, which means I can totally ignore everything you said as you seem to do.

Equal playing field created.

 

[galego]

Galego are you sure the PC UE4 demo is running lower res than PS4 version?

Epic claims that PC version is sub-1080p @ 30 fps

http://www.unrealengine.com/files/misc/The_Technology_Behind_the_Elemental_Demo_16x9_(2).pdf

 

[galego]

It will be nice to see some actual comparisons once the hardware is ready, the UE4 demo was pretty clear in the fact that the PS4 won't compete with a 680 in raw power, as is known since in raw power the PS4 is the lesser machine.

Not even close:

But it is important to put all of this into context. The DirectX 11 API is very mature while the PS4 tools and APIs are still in their initial stages of development - it's going to take time for devs to fully get to grips with the new hardware. Over and above that, assuming this is the same demo that was shown at the PlayStation 4 reveal, we know for a fact that most studios only received final dev kits in the weeks beforehand, the suggestion being that most of the UE4 work will have been produced on unfinished hardware.

With all of this in mind, the fact that PS4 is within striking distance at all is a fairly substantial achievement.


[...]


Update: Brian Karis, senior graphics programmer at Epic Games adds some more insight in the comments below, explaining some of the more obvious differences - particularly in terms of the very different lighting schemes. At the technical level, the two demos are closer than it seems:

"The biggest changes actually came from the merging of two separate cinematics, the original Elemental and the extended Elemental we showed at PS4's launch event. Each had different sun directions and required some compromises to join them. This resulted in some major lighting differences that aren't platform related but were due to it being a joined cinematic. Another effect, in the original you could see the mountains through the door where in the merged one we made the view through the door white since the mountains outside were no longer the same. Same deal with the mountain fly by. The old mountain range doesn't exist in the new one. These changes from the merge make direct comparisons somewhat inaccurate.


"Feature wise most everything is the same, AA resolution, meshes, textures (PS4 has tons of memory), DOF (I assure you both use the same Bokeh DOF, not sure why that one shot has different focal range), motion blur.


"Biggest differences are SVOGI has been replaced with a more efficient GI solution, a slight scale down in the number of particles for some FX, and tessellation is broken on ps4 in the current build which the lava used for displacement. We will fix the tessellation in the future."

Moreover, the PS4 demo run limited to 1.5 GB and one third of the AH resources

http://www.psu.com/forums/showthrea...ummies/page3?p=6023661&viewfull=1#post6023661

 

[BallaTheFeared]

That explains why they toned it way down for the PS4 version, and removed global illumination all together... Because it's slower.

 

[galego]

Luckily I was speaking on the 360, which means I can totally ignore everything you said as you seem to do.

360 GPU: 240 GFLOP

9800 GT: 504 GFLOP

Your card has more than twice the performance of a Xbox 360. Moreover your card has between 1 and 2 times more VRAM than the whole memory (RAM+VRAM) of the Xbox, which generates extra gain in performance for your side. Add that your PC has much much more RAM as well (The Xbox has only 512 MB for both CPU and GPU) and your attempt to argue is gone.

Try to run the same games using only 512 MB or total memory (e.g, using 256 RAM and 256 VRAM) in your 2x more powerful PC and say me about how you beat nothing.

 

[galego]

That explains why they toned it way down for the PS4 version, and removed global illumination all together... Because it's slower.

Yes the dev.kit they used for the demo was much slower than the real PS4 and the kit was limited to only 1.5 GB VRAM and your point was?

 

[BallaTheFeared]

360 GPU: 240 GFLOP

9800 GT: 504 GFLOP

Your card has more than twice the performance of a Xbox 360. Moreover your card has between 1 and 2 times more VRAM than the whole memory (RAM+VRAM) of the Xbox, which generates extra gain in performance for your side. Add that your PC has much much more RAM as well (The Xbox has only 512 MB for both CPU and GPU) and your attempt to argue is gone.

Try to run the same games using only 512 MB or total memory (e.g, using 256 RAM and 256 VRAM) in your 2x more powerful PC and say me about how you beat nothing.

Why would I limit myself to console specs? I thought we were just talking about the gpu? I thought we were comparing consoles to PCs?

As I stated Titan has a larger gap than the 9800 has, and the 9800 easily beats the 360 at a higher res with higher iq, not even considering the texture quality but simply res/lighting/shadows/post processing, everything, a more powerful PC gpu = a higher fidelity experience despite the 100 times a billion trillion overhead of the cpu.

Yes the dev.kit they used for the demo was much slower than the real PS4 and the kit was limited to only 1.5 GB VRAM and your point was?

Let's stick to the facts, PS4 demo was lower quality than the PC and completely lacked a high powered lighting effect at the same time.

 

[Enigmoid]

Well the 8800gt according to the anandtech bench can get around 30 fps at 1050p high 4x AA on skyrim. Despite having significantly more power as galego claims you also get significantly better visuals out of it.

 

[Erenhardt]

Let's stick to the facts, PS4 demo was lower quality than the PC and completely lacked a high powered lighting effect at the same time.

Totaly agree. SVOGI was a hog in PC demo.
http://playstationgang.com/epic-games-tim-sweeney-explains-lack-of-global-illumination-in-unreal-engine-4/

Sweeney told GameTrailers that SVOGI was extremely expensive. Therefore, Epic Games decided to create a series of graphical effects that achieve the same image fidelity as SVOGI with far better performance.

SVOGI was just a bad design and hardware has nothing to it.

Comparing two demos so much different is pointless. Who compares 1 GPU in Tomb raider v1.0 to other in v1.0.722.3?

Ohhh...And a question:
Is there a difference between porting a game from PC to console and other way around?

 

[BallaTheFeared]

Totaly agree. SVOGI was a hog in PC demo.
http://playstationgang.com/epic-games-tim-sweeney-explains-lack-of-global-illumination-in-unreal-engine-4/


SVOGI was just a bad design and hardware has nothing to it.

Perhaps, all lighting effects are hard on gpus.

 

[BTRY B 529th FA BN]

Looks to me like the PC version of the Unreal 4 Engine Demo has a slight edge over the PS4..

Link

Not by much though.

 

[BTRY B 529th FA BN]

Also looks like Epic is trying to stay in favor with Play station because I can't imagine this is true:

"...because most PCs are running a 32-bit version of Windows." - Mark Rein, Vice President of Unreal Engine.

Go where the money is, Mark.

 

[toyota]

Also looks like Epic is trying to stay in favor with Play station because I can't imagine this is true:

"...because most PCs are running a 32-bit version of Windows." - Mark Rein, Vice President of Unreal Engine.

Go where the money is, Mark.

there is an edit button.

on lensoftruth it looked as if the pc version was more blurry in spots

 

[BallaTheFeared]

Probably because the PC version has a properly working DOF filter, motion blur, real lighting not pre-rendered, and of course tessellation.

 

[futurefields]

Which cpu do you guys think has more power per core?

The Xbox 360 cpu or the PS4 cpu?

Xbox 360 cpu has twice the clock speed (3.2 ghz per core vs 1.6 on the PS4)

I bet its pretty close, PS4 obviously being more powerful overall due to core count. That 1.6ghz is dissapointingly low.

 

[galego]

Why would I limit myself to console specs? I thought we were just talking about the gpu? I thought we were comparing consoles to PCs?

It is simple to understand. The point is/was:

Consoles run 2x or so better than equal PC hardware, but it isn’t just API in the way, focus a single spec also matters.

You pretended (and still pretend) to invalidate this fact by comparing an ancient console to a PC with much much more memory, among other improvements.

Run the game using only 512MB of total memory in your PC and says us how your GPU beats that in a Xbox 360...

Unlike you I am comparing console to "equal PC hardware". When I used the 2x factor to compare the PS4 to a GTX-680 I assumed games limited to 2 GB VRAM, albeit I know that the PS4 has much more memory.

Future games will be using more memory and then a GTX-680 will be outperformed by the PS4. That is evident.

Let's stick to the facts, PS4 demo was lower quality than the PC and completely lacked a high powered lighting effect at the same time.

The fact is, as everyone is saying (including the eurogamer article) that that demo was not running on the PS4 but in an early dev. kit with unfinished APIs, which does not reflect the real performance of the PS4.