Fullscreen does run faster. As for showing FPS in fullscreen....nope, none that I know of. I remember once running FRAPS and then running DosBox in D3D/OGL and seeing FPS in the DosBox window but I don't know how accurate it was.

As for the people who say dosbox is too slow or sucks.....avoid them....like the plague. (We don't need them or want them here.) I've wasted waaaayy too much time on people who don't bother checking the facts. Those people who actually do read the readme and bother to learn the program are few and far between.

I used to be one of them until I sat down one day and toyed with dosbox getting my old games working. I'm really glad I did! Now I can't live without it 😁

I do feel obligated to prove these guys wrong in an attempt to bring them over to the dosbox side. Dos support will eventually be dead (as if xp didn't kill it enough) and its good that we have an almost perfect solution.

As I explained....as CPUs get faster so will dosbox. I wonder how an AMD FX-57 perfoms...anyone? I own an FX-53 and dosbox is like a dream (well build games aside) and I would love to hear stories from the newest processors.

As for showing FPS in fullscreen...

RivaTuner has a utility entitled RivaTuner Statistics Server which can display FPS for D3D/OGL/DirectDraw apps, windowed and fullscreen. AFAIK it measures the frequency of Front->Back buffer flips.

FPS will only show emulated graphics card's refresh rate (unless you're using smartupdate patch) not game fps. But it useful to see, since if it drops it's a sign of overcycling. I also believe a cycles/fps/whatever overlay display is somewhere on the dosbox wishlist 😁

I only ask because I wanted to prove a point. I guess the knowledge that fullscreen is faster than windowed mode will do.

DosFreak: What are the technical reasons for DOSBox running faster in fullscreen than in a window of the same resolution and with the same scaler and output settings? I wasn't aware of this.

avatar_58 wrote:

I only ask because I wanted to prove a point. I guess the knowledge that fullscreen is faster than windowed mode will do.

It depends. It is so fo "surface", but the opposite is true for "opengl"...

Mirek

HunterZ wrote:

DosFreak: What are the technical reasons for DOSBox running faster in fullscreen than in a window of the same resolution and with the same scaler and output settings? I wasn't aware of this.

I am not a programmer but I suppose it's because the difference in resolution and that Fullscreen may get full attention (foreground) from the Operating System. E.g. My windows desktop is at 1280x1024x32bpp resolution (I guess that when you are running something windowed, you are still using the desktop resolution). When I use DOSBox fullscreen I run it at 640x480 screen resolution. Some maths ...

1280x1024x32 /8 (bytes) / 1024 (KB) /1024 (MB) = 5 MB

640x480x32 /8 (bytes) / 1024 (KB) /1024 (MB) = 1,17 MB

So, working with a 1280x1024 screen (same colour depth) compared to a 640x480 one needs at least 4 times more video memory and processing power. Performance and resources consumption would be a half if running 16bpp (65535 colours).

...which is why I mentioned same-resolution comparison between fullscreen and windowed mode. It's misleading to say that fullscreen is faster than windowed if you're not trying to make all controllable factors equal before doing a comparison.

IIRC, SDL doesn't double-buffer in a Window. Only full-screen. (Unless your using OGL/D3D?)

Also depends on hardware/OS/video drivers. So it's easier to say DosBox will perform better in Full-Screen than Windowed mode since certain hardware/situations DosBox will not perform as well in a Window due to SDL crossplatform restrictions.

Normally, a full-screen application has exclusive access to the video buffer. Windowed apps all go through a "window manager" (GDI), since they all share the same screen space. The manager is responsible for z-ordering, clipping, alphablending, v-synch:ing et al. Although most of these ops are hardware accelerated nowadays, they still cost performance. Generally windowed apps are not as efficient in comparison to fullscreen apps, however the difference in most cases is marginal since the app itself is usually the computational pig, not the video driver.

Cool, that's what I was looking for. Thanks to both of you for the info - this gives me something to think about, as I usually run in a window.

One thing to note is that I use two monitors, and when I run full-screen apps I can still see Windows desktop applications on the second monitor (although the ATI drivers tend to be a bit obtuse and do things like pushing the second monitor's contents off to the side a ways).