could you post a screenshot of it ?

I use default settings except for setting the cpu to dynamic
seems to work fine,
but then my cvs has some changes as well...

Perhaps I haven't made this entirely clear: the problem arises *only* with dynamic core.

I attached an image with hq2x enabled (sorry forgot to disable it but it's not easy to get a shot of the problem and it seems to be independent of hq2x). You can see the line in the lower left part of the image which should not be there.

The game basically freezes in this screen with the music playing on.

I'll attach some more screenshots (this time w/o hq2x) displaying the problem. Somehow the graphics errors seem to distinguish between background and sprites...

If you skip the intro you can actually make it into the save/load screen but you get an exit to error pretty soon afterwards.

Seems to be caused by the mpu401 changes.
It only occurs if you have wing commander set up with
roland, and dosbox with dynamic core and intelligent=true.

So if you really want to use the roland setting, it
should work with intelligent=false.

wd

Hmm, very strange. What does the MPU device have to do with the graphics? Or does Wing Commander trigger a bug of the special combination dynamic core and intelligent mode which leads to unforeseeable Dosbox behaviour (i.e. random error symptoms)?

I see. I always use soundblaster with those games.

Hence it was allright at my place.

> What does the MPU device have to do with the graphics?

Well memory corruption can have various effects.
No idea what really goes wrong, but might be something
with the thrown int, which does only occur in intelli mode.

wd

Uhh, a new mpu issue..I too take my bets on interrupt.
(It calls ISR which sends some data to PIC as well)
Wing commander I think uses 0xD0-7 commands to send data and interrupt is triggered on every one.
Similar windows 3.1enh bug was fixed with introducing, IIRC, io-virtualisation to all but dynamic core.

I've seen lots of games having problems while accessing mpu401 in dynamic mode. Many dos extenders seem to interact strangely with mpu+dynamic using latest CVS. I have no example at hand, but the majority of dos-extender games in my retro collection does have that problem, often leading to an outright crash (which vanishes on core=full).