Dolphin, the GameCube and Wii emulator - Forums

Holy crap, apparently the back end is working wonders for the ps3 emulator. o.o 
Of course this guy is showing off only 2d games that wouldn't eat up much power anyway, but still, impressive none the less.
Thoughts?https://www.youtube.com/watch?v=3KeZkYHiBcs

rlaugh0095 Wrote:Holy crap, apparently the back end is working wonders for the ps3 emulator. o.o

And how did you reach that conclusion?

rlaugh0095 Wrote:Thoughts?

There is nothing to talk about without any details.

(09-19-2015, 01:51 PM)NaturalViolence Wrote: [ -> ]

rlaugh0095 Wrote:Holy crap, apparently the back end is working wonders for the ps3 emulator. o.o

And how did you reach that conclusion?

rlaugh0095 Wrote:Thoughts?

There is nothing to talk about without any details.

Before this, weren't even 2d games, running very slowly at best? The last video I saw prior was of Disgaea running very slowly.

(09-19-2015, 01:51 PM)NaturalViolence Wrote: [ -> ]

rlaugh0095 Wrote:Holy crap, apparently the back end is working wonders for the ps3 emulator. o.o

And how did you reach that conclusion?

rlaugh0095 Wrote:Thoughts?

There is nothing to talk about without any details.

The first part is true. The DX12 backend is pretty much without a doubt much faster then the OGL backend. 

Do we have any comparisons between the two backends? Were the two backends made by different authors? Is there a d3d11 backend against which it can be compared?

Let me clarify my point a bit. I remember in the early days of dolphin openGL was much faster than d3d9 (we're talking several times faster). Then d3d9 got optimized by tino and neobrain and was much faster. Then openGL was optimized later on by various parties and became roughly the same speed. Most emulators I've seen have followed a similar pattern in the early days of their graphics backends where performance varies wildly as they get optimized. This often has nothing to do with how intrinsically "good" an API is or even what features of that API are utilized. That becomes more clear later on. So until that happens I simply don't see how I can discuss this any further or form any kind of conclusion about anything.

(09-20-2015, 01:53 AM)NaturalViolence Wrote: [ -> ]Do we have any comparisons between the two backends?  Were the two backends made by different authors?  Is there a d3d11 backend against which it can be compared?

Let me clarify my point a bit.  I remember in the early days of dolphin openGL was much faster than d3d9 (we're talking several times faster).  Then d3d9 got optimized by tino and neobrain and was much faster.  Then openGL was optimized later on by various parties and became roughly the same speed.  Most emulators I've seen have followed a similar pattern in the early days of their graphics backends where performance varies wildly as they get optimized.  This often has nothing to do with how intrinsically "good" an API is or even what features of that API are utilized.  That becomes more clear later on.  So until that happens I simply don't see how I can discuss this any further or form any kind of conclusion about anything.

DX12 is an unofficial branch from what I know. You can go to the RPCS3 Forums and ask there. The biggest improvement I've seen is a few times faster but the game was in the single digits anyway because the game won't work on the JIT

Quote:And how did you reach that conclusion?

Jumped to it.

Nah if he has performance results showing it performs much better then he's right to reach that conclusion. However whether or not that increase in performance is due to advantages of d3d12 or just poor optimization of the openGL backend at this early stage is another story.

I will say there is a bit of conclusion jumping, but I have seen pre dx12 footage and it was not full speed, so I guess you could call it an educated jump. I suppose I should provide evidence, but as NaturalViolence has said it could have been poor optimization of the opengl backend which would still make my post "wrong" regardless.

Also wanted to point out that I wasn't sharing because I think it would speed up Dolphin or anything. I realize that if it was indeed the dx12 backend that gave the boost, then it's possible that the difference in the hardware emulated could result in different results per emulator.

Oh I just thought about it, and wouldn't there be a higher chance of the dx12 api being poorly optimized since it is only a few months old, compared to a backend that has been around for years? o.o

If someone's been dabbling with the other backend for a while, they may have started with misconceptions about how the real hardware worked, so there could be design flaws that mean it can't achieve its potential even after it's known what the hardware is doing. When another one is written from scratch later on, more is known about what's supposed to be going on, so it's easier to write an efficient backend.