Does anyone know what happens to the stack pointer after making an Int 10h call from a DOS4G app? I've patched ICR2 to make a dummy interrupt during the aspect ratio calculation, but the stack pointer appears to be different to what it is in the execution trace, the other registers seem to be preserved. Some documentation on how this is handled would be useful.

I know Int10h is a real mode call. Maybe I could modify the Int10h vector to directly int10h instead of massaging a protected mode to real mode transition. I don't know if that's possible.

The reason this is important is the game appears to use different aspect ratios: 4:3 for external views and something closer to 1:1 for cockpit views . I could probably create a code cave and add my code in, but I'm not great with dos and don't really know where I can insert stuff, assuming there's unused space in there.

Maybe this could be helpful: https://www.opferman.com/Articles/Text/index. … up=1001&order=4

Makus wrote on 2025-10-23, 21:27:

Maybe this could be helpful: https://www.opferman.com/Articles/Text/index. … up=1001&order=4

Thanks @Makus!

Just a quick update on widescreen support with ICR2, the aspect ratios are correct for 16:9 and 32:9 through modelview matrix manipulation, but the sprites in game are scaled equally in both dimensions so they're out of whack at the moment:
[EDIT] https://youtu.be/broxTVRiUEI (sprite scaling fixed)[/EDIT]
https://youtu.be/jwRRvG4TJkA

(Also available here for with just a sign-in and an extra formula in the description: https://www.patreon.com/posts/indycar-racing- … ntent=join_link)

An earlier attempt is available here with indepdent x and y scaling (of models):
https://youtu.be/gSLTet0VncI
(https://www.patreon.com/posts/indycar-racing- … ntent=join_link)

The game uses a 3x3 matrix with both 16.16 and 2.30 fixed point arithmetic.

My xbox one controller (wired) seems to not be recognized in icr2. Dosbox does recognize it (Verified it by going ctrl+f1 before going into icr2), but inside the game, the joystick calibration doesn't register the movement of the joystick, neither of the triggers. The inputs sort of "glitch" when in the calibration tab. Im using the latest rendition alpha (watemarked) along icr2rend by gplaps.

josemike03 wrote on 2025-11-08, 00:55:

My xbox one controller (wired) seems to not be recognized in icr2. Dosbox does recognize it (Verified it by going ctrl+f1 before going into icr2), but inside the game, the joystick calibration doesn't register the movement of the joystick, neither of the triggers. The inputs sort of "glitch" when in the calibration tab. Im using the latest rendition alpha (watemarked) along icr2rend by gplaps.

I have one a 360 pad, I'll see if I can get it to work. Give me an hour to get back to my desk.

You're right. It doesn't work in dosbox-rendition but does in dosbox-staging-rendition.

Dosbox-rendition does appear to detect it correctly, even though it doesn't actually work. Dosbox uses SDL1 and staging uses SDL2. Let me see if it's fixable or whether it needs a separate xinput codepath.

OK, it looks like I have to port SDL3's XInput support to SDL1. That's currently underway.

@josemike03, This will take quite a while, the XInput code in newer SDLs is C++ based and SDL1 is C based. I already have some XInput code I used for something else. I can try and integrate that directly into dosbox. I'll try doing that once I get widescreen working. It's a bit of a big job, but its next on my list after widescreen.

sharangad wrote on 2025-11-08, 09:59:

@josemike03, This will take quite a while, the XInput code in newer SDLs is C++ based and SDL1 is C based. I already have some XInput code I used for something else. I can try and integrate that directly into dosbox. I'll try doing that once I get widescreen working. It's a bit of a big job, but its next on my list after widescreen.

Thank you very much... Now at least I know it is not possible for now instead of taking hours trying to figure it out myself. I'm no programmer so I'll wait... Hope you can get widescreen soon!

josemike03 wrote on 2025-11-09, 00:07:

sharangad wrote on 2025-11-08, 09:59:

@josemike03, This will take quite a while, the XInput code in newer SDLs is C++ based and SDL1 is C based. I already have some XInput code I used for something else. I can try and integrate that directly into dosbox. I'll try doing that once I get widescreen working. It's a bit of a big job, but its next on my list after widescreen.

Thank you very much... Now at least I know it is not possible for now instead of taking hours trying to figure it out myself. I'm no programmer so I'll wait... Hope you can get widescreen soon!

You should be able to get it working with an xinput to directinput converter. You would also have to configure the joystickid in dosbox-rendition.conf to point to the fake directinput instance instead of "Xbox controller"

A bit of progress on CGL, got cglQueryScreen to work, when I get a bit of spare time on to the next function (cglInitScreen):

sharangad wrote on 2025-09-15, 18:22:

@Gona, could you post the resources used in Windows 98 by your 3DBlaster VLB? I'm interested in ports and address ranges. I hope these are listed in device manager.

I hope Windows 95 is also suitable:

1I/O range: 03E0 - 03E3
2IRQ: 05
3Memory Range: C8000000 - C8000FFF
4Memory Range: D9000000 - D9FFFFFF

Gona wrote on 2025-11-16, 21:12:

I hope Windows 95 is also suitable: […]
Show full quote

sharangad wrote on 2025-09-15, 18:22:

@Gona, could you post the resources used in Windows 98 by your 3DBlaster VLB? I'm interested in ports and address ranges. I hope these are listed in device manager.

I hope Windows 95 is also suitable:

Copy code to clipboard

1I/O range: 03E0 - 03E3
2IRQ: 05
3Memory Range: C8000000 - C8000FFF
4Memory Range: D9000000 - D9FFFFFF

Thanks Gona, Win 95 is fine. This is very different to the 3DB PCI and looks *more* straightforward maybe. Only 3 ports (from 03E0), a 16 MB address range from D9xxxxxx (video memory) and a maybe some memory mapped IO at C8xxxxxx. I'll get you a test app next week by next week, bit ill at the moment. Before that I'll need to add these blocks to the known list in Dosbox and see what gets written/read to/from them.

sharangad wrote on 2025-11-16, 22:59:

Thanks Gona, Win 95 is fine. This is very different to the 3DB PCI and looks *more* straightforward maybe. Only 3 ports (from 03E0), a 16 MB address range from D9xxxxxx (video memory) and a maybe some memory mapped IO at C8xxxxxx. I'll get you a test app next week by next week, bit ill at the moment. Before that I'll need to add these blocks to the known list in Dosbox and see what gets written/read to/from them.

OK, get better.
I put here the 3D Blaster VLB default DOS resources if someone seek them:

1Control Region: C1000000
2Memory Region: D0000000
3I/O Port Base: 3E0
4IRQ Requested: 5

Tested Nascar Racing 2: This game shares exactly same issues as Nascar 1999 Edition.
In additional: if you happen to win, the left most person wearing jacket, his jacket have corrupted colors, unconfirmed trough if its same on other versions (seems minor thing)

RaVeN-05 wrote on 2025-11-18, 16:36:

Tested Nascar Racing 2: This game shares exactly same issues as Nascar 1999 Edition.
In additional: if you happen to win, the left most person wearing jacket, his jacket have corrupted colors, unconfirmed trough if its same on other versions (seems minor thing)

Thanks @raven-05! I'll add that to my list of bugs. There was one more reported, flickering in the interstate '76 menu when paused.

Gona wrote on 2025-11-17, 07:25:

OK, get better. I put here the 3D Blaster VLB default DOS resources if someone seek them: […]
Show full quote

sharangad wrote on 2025-11-16, 22:59:

Thanks Gona, Win 95 is fine. This is very different to the 3DB PCI and looks *more* straightforward maybe. Only 3 ports (from 03E0), a 16 MB address range from D9xxxxxx (video memory) and a maybe some memory mapped IO at C8xxxxxx. I'll get you a test app next week by next week, bit ill at the moment. Before that I'll need to add these blocks to the known list in Dosbox and see what gets written/read to/from them.

OK, get better.
I put here the 3D Blaster VLB default DOS resources if someone seek them:

Copy code to clipboard

1Control Region: C1000000
2Memory Region: D0000000
3I/O Port Base: 3E0
4IRQ Requested: 5

I think port 3E0 is used to work out the other settings. We'll find out soon enough.

Gona wrote on 2025-11-17, 07:25:

OK, get better. I put here the 3D Blaster VLB default DOS resources if someone seek them: […]
Show full quote

sharangad wrote on 2025-11-16, 22:59:

Thanks Gona, Win 95 is fine. This is very different to the 3DB PCI and looks *more* straightforward maybe. Only 3 ports (from 03E0), a 16 MB address range from D9xxxxxx (video memory) and a maybe some memory mapped IO at C8xxxxxx. I'll get you a test app next week by next week, bit ill at the moment. Before that I'll need to add these blocks to the known list in Dosbox and see what gets written/read to/from them.

OK, get better.
I put here the 3D Blaster VLB default DOS resources if someone seek them:

Copy code to clipboard

1Control Region: C1000000
2Memory Region: D0000000
3I/O Port Base: 3E0
4IRQ Requested: 5

Gona, would it be possible to share the driver CD for the 3d blaster vlb? I don't know if it's available already.

[edit] they're already available on vogon's drivers:
https://www.vogonsdrivers.com/getfile.php?fileid=824

josemike03 wrote on 2025-11-08, 00:55:

My xbox one controller (wired) seems to not be recognized in icr2. Dosbox does recognize it (Verified it by going ctrl+f1 before going into icr2), but inside the game, the joystick calibration doesn't register the movement of the joystick, neither of the triggers. The inputs sort of "glitch" when in the calibration tab. Im using the latest rendition alpha (watemarked) along icr2rend by gplaps.

https://nirvtek.com/downloads/RReady.Alpha.20251120.001.7z
MD5: a24d11a3bbfb96882f8089d973c9ed3c

https://nirvtek.com/downloads/RReady.Alpha.20251120.002.7z
MD5: 60eb37963add010606826a144d004737

- joystick multithreading can be moved to its own thread:
[joystick]
multithreaded=true
(This is the default).
This reduces the input latency and guarantees 200 joystick polls per second at the cost of an extra CPU thread. Previously the joystick was polled at best once every 20 ms provided no other SDL events (like keyboard processing) were present. If there were any events, they all needed to be dealt with before joystick processing, adding to latency.
- Preliminary support for XInput joysticks
- Run dosbox-rendition and close it. Click view log. It should list the available joysticks like this:
2025/11/20|7:25:11.757 Available Joysticks
2025/11/20|7:25:11.757 0 is Controller (Gamepad F310)
2025/11/20|7:25:11.757 1 is Controller (XBOX 360 For Windows)
2025/11/20|7:25:11.757 2 is XInput Controller (player #1)
2025/11/20|7:25:11.757 3 is XInput Controller (player #2)

To use an Xbox controller:
joysticktype = 4axis
usejoystickids = true
joystickids = 3 (to use player#2's controller) OR
joystickids = 2 (to use player#1's controller)

For two player
joysticktype = 2axis
usejoystickids = true
joystickids = 3 ,2 (to use player#2's controller as stick one and player#1's controller as stick two)

There's also :
axesmapping=0,1,2,3
for a 4-axis controller this maps stick 1 and stick 2 as x1, y1 and x2, y2. To use the triggers as axis 3 this should be set to:
axesmapping=0,1,4,3
This makes stick axis 4 (triggers) function as the third axis.

For an xbox controller, the mmsystem/directinput nodes such as sticks #0 and #1 can be used, but axismapping would have to be adjusted. By default for a 360 controller, axis #2 is mapped to the triggers.
axesmapping=0,1,4,3 OR axesmapping=0,1,3,4 should work.