videoram=xxxx

E.g. videoram=4096 will yield you a 4MB video card.

uhhh...2mb is plenty of memory.

1024x768x8/8=786,432 bytes.

AFAIK, Magic Carpet 2 uses 256 colors which is 8bit.

@ 16bit color it's 786,432x2=1,572,864....which as you can see is far less than 2mb.

and MG2 only uses 640x480 so memory amount is of course not your problem.

Magic Carpet 2 runs fine @ 640x480 on my XP 2800....slower than 320X200 but perfectly playable.

i see. am running athlon 1.75ghz 512mb of ram played with vesa enabled and tried making the videoram higher but it still reports 2mbs 😢

your right it does run slower but remember it being perfectly playable using a voodoo3 🙄 and 300mhz with 128mbs of ram back in tha day

thanks for the reply though will try few more configs see what i can muster 😎

yeah. now we are speaking......

1.75 GHZ can not emulate a 300 mhz machine......

Dream on 😀

why not? i mean we're looking at not even half the processor speed of my machine surely there must be a way. <--- can imagine this has probably been said a quad billion times and the explaination is probably here in the forums somewhere but isn't it just dealing with interrupts (know a little programming not very much so be nice!) 😀

Because we're not just talking about one CPU emulating another, but rather one CPU emulating another entire computer.

That does bring up a question I've been meaning to ask.....

How far can Dynarec in DosBox go? I've been meaning to perform extensive benchmarking comparing NTVDM to DosBox but I haven't got around to it yet.

Number one CPU killer is scaling, followed by Adlib emulation. Third is DOSBox's memory bookkeeping/protected mode support. (Figures sampled while developing Hq2x, which was at that time as fast as advmame2x. Other scalers and sound emulations will have less impact.)
On at least one of my machines, the cheap onboard sound chip kills 50% CPU just by being used.

So you see, with dynamic CPU, it's not actually the CPU instructions that slow down things. The real problem is that protected mode is far more complex than you'd think, and for correct emulation, a lot of CPU time is spent. Virtualization avoids that by using the host CPU's protected mode (which is why they need a kernel driver). I'm not sure what QEmu does, but I think it's just a little smarter than us, and then again, maybe it suffers the same fate.

I just hang in here, cause this is an issue that intrests me too. I am using an Athlon XP3200+. Due to my TFT (see other post), I have to fullfixed= true to 640x480 with surface, normal2x and scale=true. I am under the impression that i am about as fast as a 486@66 MHz. Is this reasonable?

How much performance would I gain if I could run in fullfixed=false and surface mode, none, no scaling?

Thanks for link or input...

ux-3

You could gain by using opengl/openglnb/overlay or directx output, then normal2x will not matter, and the video card does the scaling. Which output is best depends highly on OS, video card and your preference. Surface mode doesn't differ much with or without normal2x. Your speed seems okayish, around 30.000 cycles should be possible with your box.

So I must be doing something wrong? Using openGLnb, I seem to get audio stutter at 9000 cycles (privateer), while using surface with fullfixed and normal2x runs fine. It seems that I get less performance out of openglnb though I have a GF6800...
Hmm...??? I should double check this

i was just wondering how i can run a game such as "jones in the fast lane" in full screen?

Well yeah, Opengl eats up more cycles than Surface/Overlay/Ddraw by alot. At least it does on my ATI. Simply load up DosBox and Task Manager and switch between different output methods and monitor DosBox processor usage to determine which output method has the least usage of processor cycles.