Arcade monitor enhance performance?

[RayB]

Here's an interesting question.... If a game is not running at full frame rate when I am using a PC monitor (with no screen effects or anything--just straight up best resolution for PC monitor), can I expect any sort of performance increase when I move to a real CGA arcade monitor?

The reason I ask is that if a PC monitor is > 31 hz (actually, probably over 50-60 hz, right?) then isn't it a little less work for the PC to output an image for 15 hz? Or is it completely unrelated?

I guess I'm trying to be optimistic about some games where I can't quite get maximum frame rates.

[Minwah]

The reason I ask is that if a PC monitor is > 31 hz (actually, probably over 50-60 hz, right?) then isn't it a little less work for the PC to output an image for 15 hz? Or is it completely unrelated?

Actually a PC monitor has a horizontal refresh rate of 31khz (or above).  60+hz is the vertical refresh rate.

You may see a slight performance difference, but I would expect it to be not much different to 640x480.

[elvis]

It depends on a lot of things.  Pure software stretch is fairly CPU intesive on older machines, particularly with PCI cards with poor 2D drivers.  Those sorts of setups are rare, however.  Most systems 2D output are fairly optimised, and have been for some time.

On something modern, say a P4 3GHz with an off-the-shelf Nvidia or ATi card (or even an ArcadeVGA) the performance difference moving from "hi res" (640x480) down to CGA would be almost nothing.  You are certainly NOT going to be changing borderline games magically to full framerate.

The easy way to test or course is to run your games at a 320x240 style mode, without stretching and see what the performance is like with a benchmark.  The system doesn't care what the output device is, only what the software is actually rendering at.

But I'm still sticking with the theory that you won't get more performance out of it.  As always, your best bet is to purchase a faster processor and be done with it.

[JoeB]

In MAME, the most CPU intensive component is the actual processor emulator, not the graphics output.  So you will not notice a difference.

[Minwah]

In MAME, the most CPU intensive component is the actual processor emulator, not the graphics output.  So you will not notice a difference.

Kindof OT...but just worth noting that you can sometimes (typically with a pc monitor) get better performance by using different video modes.  For example 1024x768 will likely be slower than 640x480.  I've used some laptops which have major performance issues with some modes for some reason.  Also, -hswtretch can sometimes improve performance.

[elvis]

Stretch modes are entirely dependant on the blitter being used, and if that blitter can offload to the video card.

Using something like DirectDraw, when doing stretching of an image you can easily offload the stretch function to the video card if you use something modern like a GeForce or Radeon card, and the proper drivers.  That's also the reason why you cannot turn off the bilinear filter on DD stretching with these cards, because it's done at a card level and not at a software level.

Likewise if you are using SVGA or VBE, some cards can happily stretch images using video hardware with the correct VESA extensions in hardware.

If however you're using a particular blitter to do the stretch, and either you have the wrong drivers (or none at all loaded) or a card that doesn't support the particular type of stretch then the stretch will be done by the CPU, and in turn interfere with performance.

The best way to test this is empirically using the benchmarking features built into MAME.  On my main workstation (GeForce4 video card, using SDL under Linux), rendering at 1024x768 compared to 640x480 makes negligable difference (1 or 2 frames per second in games running at 300FPS anyway).  Likewise on my MAME cabinet (GeForce2 running DirectDraw) the different is sweet bugger all.  This is entirely a result of using 2D accellerated drivers on modern cards.

Test for yourself on your own systems, and see what the results are.