DOS application results incomplete in DOSBox relative to full-screen DOS environment \ VOGONS

DOS application results incomplete in DOSBox relative to full-screen DOS environment

Topic actions

Post a reply

First post, by chuck barlow

Posted on 2008-10-03, 18:11

chuck barlow Offline

Rank Newbie

Rank: Newbie
Posts: 7
Joined: 2008-09-26, 23:34

I am running DOSBox version 0.72, in Windows version 5.1 service pack 2, in Windows XP Professional version 2002 service pack 2. I have a Pentium(R) 4 CPU processor at 3.4 GHz with 992 MB of RAM.

I am trying to run an old (1993) full-screen DOS application called CARPET in DOSBox. The application graphics include lots of input menus and output graphs of RADAR performance. I used to be able to do Print Screens in Windows NT and paste into PowerPoint for my presentations. I turned to DOSBox when I found out that Windows XP does not support Print Screen (or Alt-Print Screen) when in a full-screen DOS environment.

I have read the README file for DOSBox.

My problem: DOSBox allows me to do Print Screen and paste into PowerPoint just as I had hoped. However, the output graphs do not show all of the data that should be displayed. E.g., when I run without DOSBox, I get results out to 60 km in range, but when I run with DOSBox and the identical inputs, the results only go out to 10 km. Somehow, DOSBox is interfering with the longer range calculations in the DOS application and/or the longer range portion of the graphs.

I have tried cycles=max and full screen in the config file to no avail (after unloading the application and DOSBox in between config file changes).

Any ideas?

Reply 1 of 10, by wd

Posted on 2008-10-03, 18:24

wd Offline

Rank DOSBox Author

Rank: DOSBox Author
Posts: 10813
Joined: 2003-12-03, 21:23

Use dosbox's screen capturing capabilities (see the readme). This should
result in exactly the same image as what you are seeing.

Reply 2 of 10, by chuck barlow

Posted on 2008-10-03, 19:57

chuck barlow Offline

Rank Newbie

Rank: Newbie
Posts: 7
Joined: 2008-09-26, 23:34

Thanks for your quick reply.

Sorry for not being more clear. It IS the same image as what I am seeing. The problem is that the old DOS application produces an incomplete image when operating within DOSBox, but produces the complete and correct image when not using DOSBox. I know this might sound impossible, but DOSBox causes the CARPET model to display different results than the results obtained when not using DOSBox.

Reply 3 of 10, by MiniMax

Posted on 2008-10-03, 20:06

MiniMax Offline

Rank Moderator

Rank: Moderator
Posts: 5128
Joined: 2004-01-18, 04:15
Location: Stockholm, Sweden, Europe, Earth Interests: Old games & young women

Maybe DOSBox is calculating the results incorrectly? It is after all emulating the CPU. Tried different core-settings?

Reply 4 of 10, by wd

Posted on 2008-10-03, 20:13

wd Offline

Rank DOSBox Author

Rank: DOSBox Author
Posts: 10813
Joined: 2003-12-03, 21:23

but DOSBox causes the CARPET model to display different results than the results obtained when not using DOSBox

Are you talking about what is displayed on the screen? Can you post a screenshot
of dosbox running in windowed mode?

Reply 5 of 10, by chuck barlow

Posted on 2008-10-03, 21:45

chuck barlow Offline

Rank Newbie

Rank: Newbie
Posts: 7
Joined: 2008-09-26, 23:34

Yes, the display on the screen in the full-screen DOS mode has more info (plots and labels beyond 10 km) than the display on the screen inside the DOSBox window in the windowed mode. I will attach a file of the DOSBOX screenshot in a couple of seconds.

Reply 6 of 10, by wd

Posted on 2008-10-04, 06:47

wd Offline

Rank DOSBox Author

Rank: DOSBox Author
Posts: 10813
Joined: 2003-12-03, 21:23

Be sure to check out MiniMax's suggestion, that is use core=dynamic in
your dosbox configuration file.

Reply 7 of 10, by TeaRex

Posted on 2008-10-05, 09:01

TeaRex Offline

Rank Member

Rank: Member
Posts: 398
Joined: 2005-02-06, 07:39

Could this have to do with floating-point precision? IIRC, not all cores in DOSBox support the full 80-bit precision of real Intel x87 coprocessors. Some do all FP calculations in 64 bit.

tearex

Reply 8 of 10, by chuck barlow

Posted on 2008-10-07, 16:18

chuck barlow Offline

Rank Newbie

Rank: Newbie
Posts: 7
Joined: 2008-09-26, 23:34

I won't be back to my office PC until Friday, at which time I will try the suggested core=dynamic setting in my config file. I'll post the results after I try it.

Reply 9 of 10, by chuck barlow

Posted on 2008-10-10, 17:28

chuck barlow Offline

Rank Newbie

Rank: Newbie
Posts: 7
Joined: 2008-09-26, 23:34

It now works! (i.e. with core=dynamic) Thanks wd and minimax! It works with either cycles=max or cycles=auto. Since the readme file recommends cycles=max, I'll leave it there. Attached is a .ppt file with slide 2 showing the new results (slide 1 is the old/bad results, and pay no attention to slide 3).

Attachments

Filename

testing dosbox.ppt

File size

145.5 KiB

Downloads

332 downloads

File license

Fair use/fair dealing exception

Reply 10 of 10, by wd

Posted on 2008-10-10, 19:07

wd Offline

Rank DOSBox Author

Rank: DOSBox Author
Posts: 10813
Joined: 2003-12-03, 21:23

In the next dosbox version it might work with core=normal as well, so the
default configuration should work out of the box. Be sure to try that then
and maybe report back if it works or not.
(the reason why core=normal currently does not work is quite likely what
TeaRex said)

Go to top of page Go to top of page

Back to DOSBox Development

Main menu

Common searches