What should people use as a baseline for comparison? CVS? 0.63?

cvs.

Haven't tested that much but it seems the same as normal CVS to me.

Do you notice a difference yourself ih8?

Yes. Idle cpu utilization halved, increase cycles from a little to a lot depending on the game. My system being an athlon xp 2700, 256k l2. I watch cpu usage with process explorer http://www.sysinternals.com/Utilities/ProcessExplorer.html running dosbox windowed on 98se.

Odd, I could have swore that DosBox was close to 0% @ the CLI on my 2800 in earlier builds. Now after reading your comment I loaded up the latest CVS and it seems to hang around 32%. Using your build it seems to be about half that much. Interesting....

Quake v1.08
320x200

Your CVS
42000 cycles=95-99%

Ordinary CVS
40000 cycles=95-99%

640x480

Your CVS

68000=95-99%

Ordinary CVS
48000=95-99%

Duke Nukem 3D
320x200

Your CVS

31000=95-99%

Ordinary CVS

32000=95-99%

640x480

Your CVS
42000 cycles= 95-99% CPU

Ordinary CVS
33000 cycles=95-99% CPU

You should probably ignore the 320x scores. I have some background apps running.

This patch is on sourceforge with my other optimization patch:
http://sourceforge.net/tracker/index.php?func … 551&atid=467234

I found that the patch (a part or all of the contents of fast.diff) is now in cvs repository of dosbox.
But it set the value of frameskip to '1' as I tested. (but dosbox says frameskip=0)
Therefore the overall movement of games is not soft(flexible) for me.)

Crap, I think I see what would cause that, a > instead of a >=, which would make the optimization bullshit(the generated code does get rid a jump but still..).

orig:
if (render.frameskip.count<render.frameskip.max) {

modified:
if (render.frameskip.count>render.frameskip.max) {

so should be:
if (render.frameskip.count>=render.frameskip.max) {

🤣. I looked over it as well

Oh boy
I was testing the patch a lot and it kept amazing me 😜
No wonder 😀
hahahhahahahaha