MAMEWorld >> EmuChat
View all threads Index   Threaded Mode Threaded  

Pages: 1

Elapido
Lurker
Reged: 04/10/04
Posts: 225
Send PM


Some doubts (nothing really important) about mame.ini and two setups
#309658 - 06/01/13 09:14 PM


In one of my computers, a core2duo with TFT, Nvidia card, etc. I have MAME setup with stretch in the monitor's native res and with an rgb effecto. This is part of my mame.ini

throttle 1
syncrefresh 1
refreshspeed 0
video d3d
prescale 1
waitvsync 0
hwstretch 1
filter 1
effect 1.png
resolution 1920x1080@60
triplebuffer 1
switchres 0


I've read it's better to use tripple buffer rather than vsync. Is this right? Or should I set both on?

Contrary to syncrefresh, which synchronizes all games to the monitor refresh rate, I don't know exactly what's the result of activating refreshspeed. Performance is silk-smooth as I have it right now, so it's off.

I don't know what's switchres either. If I set it to on, the games run at maximum speed.

Another doubt I have is about multithreading. When enabled, the sound has issues. Is there any performance gain when activated? Oh, and do you recommend priority set to 1?

By the way, do you have hlsl working? 144 goes fine with hlsl and with the same ini, but 148 reports a direct3d error here.




And this is my setup in a different computer, a core2duo, also with XP, with a CRT low res monitor, ArcadeVGA and games running in native resolutions.

throttle 1
syncrefresh 1
refreshspeed 0
video ddraw
prescale 0
waitvsync 0
hwstretch 0
filter 0
effect none
resolution auto
triplebuffer 1
switchres 0


Games run look and run perfectly smooth in low res, but I still have the same doubt about using vsync, triple buffer or both.



Marcelo_20XX
MAME Fan
Reged: 04/20/13
Posts: 43
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: Elapido]
#309661 - 06/01/13 10:25 PM


I always disable whenever possible multithreading, because it messes the audio from time to time somehow, the obvious benefit of having it on is the speed gain in very demanding games.

As for vsync over triple buffering, its better triple buffering because vsync adds input lag.



DaRayu
MAME Fan
Reged: 02/05/13
Posts: 162
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: Marcelo_20XX]
#309747 - 06/03/13 11:20 AM


> As for vsync over triple buffering, its better triple buffering because vsync adds
> input lag.

Nope. I checked it and it's exactly the other way around: Triple buffering adds lag.
Have a look at this thread:
www.mameworld.info/ubbthreads/showflat.php?Cat=&Number=307756



Marcelo_20XX
MAME Fan
Reged: 04/20/13
Posts: 43
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: DaRayu]
#309765 - 06/03/13 06:19 PM


Sad that your post got into a flame war, I was too interested about the input lag. Anyway if I want to use Direct3D, which is the best option?

I ran without noticeably input lag in Direct3D by enabling "force vsync" and disabling "triple buffering" in the Nvidia control panel and disabling waitvsync and triple buffering on MAME, that way I get a smooth scrolling with almost no input lag.



Elapido
Lurker
Reged: 04/10/04
Posts: 225
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: Marcelo_20XX]
#309776 - 06/04/13 12:34 AM


I have no idea. I've been testing and I don't notice any input lag nor any difference between using vsync or tripple buffer.



DaRayu
MAME Fan
Reged: 02/05/13
Posts: 162
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: Marcelo_20XX]
#309824 - 06/05/13 12:17 AM


> Anyway if I want to use Direct3D, which is the best option?

I noticed that the only option that produces real input lag with vsync enabled is sadly the one option that most people would probably want to use: Fullscreen with Direct3D.
If you use DirectDraw or if you use Direct3D in a window, vsync won't add any noticable input lag. (On an average, it's less than 0.5 frames of lag. So, if a button needs three frames to react without vsync, then with vsync it will mostly need three frames as well, and sometimes four.)
But fullscreen and Direct3D will have the same input lag as if you enabled triple buffering. We still haven't found a reliable way to circumvent this. It must be a DirectX issue, not a driver issue, since I had the same results with ATI and nVidia.


> I ran without noticeably input lag in Direct3D by enabling "force vsync" and
> disabling "triple buffering" in the Nvidia control panel and disabling waitvsync and
> triple buffering on MAME, that way I get a smooth scrolling with almost no input lag.

I guess you would get the same result if you just enabled vsync in MAME. No need to use the control panel.

But what do you mean with "noticeably" input lag? Did you ever notice input lag at all? Because if you didn't, how do you make a difference between laggy and not laggy?
I guess I'm pretty reliable in detecting input lag in "Super Mario Bros.", so all I have to do is play "Vs. Super Mario Bros." to see if the jump button reacts in time. But I checked all of this with the 60 fps camera, as I described in the other thread. So, it's not just a subjective feeling, but can be proven.


By the way, triple buffering lags by design:

DirectX renders one frame and puts it in memory.
Then it renders the second frame and puts it in memory.
Then it renders the third frame and puts it in memory.
Then it displays the first frame.
Then it renders the fourth frame and puts it in memory.
Then it displays the second frame.
Then it renders the fifth frame and puts it in memory.
Then it displays the third frame.
And so on.

So, as you see, the scene that you are seeing on the screen is always two frames behind the current status of the game. (If you press the jump button when seeing Mario on the last possible spot in front of an abyss where jumping is possible, he will fall because Mario's actual position is two frames ahead of the scene that you see on-screen.)
That's why never enable triple buffering.

About vsync: In general, it's the best option, but there is that stupid DirectX behavior where Direct3D in fullscreen will produce the lag. (Calamity can tell you more about the technical reasons for it.)



DaRayu
MAME Fan
Reged: 02/05/13
Posts: 162
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: Elapido]
#309825 - 06/05/13 12:19 AM


> I have no idea. I've been testing and I don't notice any input lag nor any difference
> between using vsync or tripple buffer.

Do you notice any input lag between using vsync or triple buffering and between not using any of it at all?



Marcelo_20XX
MAME Fan
Reged: 04/20/13
Posts: 43
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: DaRayu]
#309835 - 06/05/13 08:07 AM


I tested using Super Mario Bros too, and I noticed input lag with the options you described in your thread about this issue.

And as far as I remember the vsync option is a little different in MAME that in the Control Panel but both achieve the same thing although doing things differently. So I prefer to be enabled by the Nvidia driver.

The only reason I use Direct3D over Direct Draw is because HQ2XS only works in Direct3D.



DaRayu
MAME Fan
Reged: 02/05/13
Posts: 162
Send PM


Re: Some doubts (nothing really important) about mame.ini and two setups new [Re: Marcelo_20XX]
#309837 - 06/05/13 09:41 AM


> And as far as I remember the vsync option is a little different in MAME that in the
> Control Panel but both achieve the same thing although doing things differently. So I
> prefer to be enabled by the Nvidia driver.

Does the vsync option in the control panel produce less input lag than using vsync in MAME?


> The only reason I use Direct3D over Direct Draw is because HQ2XS only works in
> Direct3D.

Yeah, there are many reasons to use Direct3D. For example if you want to use bezel artwork, then the DirectDraw function will first resize the artwork how it would look like on a screen with minimum resolution (i.e. where the game image wouldn't need to be stretched) and then it enlarges this small image to the actual size. Which results in a low quality artwork even though the actual artwork had the correct resolution. Also, scanlines are one line per pixel row of the screen in Direct3D and one line per pixel row of the game's raw image in DirectDraw.

Using Direct3D in windowed mode worked fine on my system. Of course, you have the window border and the visible desktop with that.


Pages: 1

MAMEWorld >> EmuChat
View all threads Index   Threaded Mode Threaded  

Extra information Permissions
Moderator:  Robbbert, Tafoid 
0 registered and 250 anonymous users are browsing this forum.
You cannot start new topics
You cannot reply to topics
HTML is enabled
UBBCode is enabled
Thread views: 2830