I tried setting the cycles to fixed 120000, but that just made it even worse.

When you said in other posts you have a very powerful PC, did you mean relative to windows 98 requirements, or relative to modern requirements. As a P4EE or an E2160 is a powerful PC* relative to Win98 requirements on bare metal, but I'm not sure either are up to doing 120,000 cycles reliably on DOSbox. So if it's getting worse, try less cycles, like 60,000.

*Also I believe you need high single core speed, so 8 cores of Atom or Arm while "apes together strong" are kinda weak on a single.

Hey friends, bit of an odd question for the group here. I am trying to use Dosbox-x in a headless VPS via VNC. It launches fine and seems to work aside from the keymap. Every keypress maps to a wrong character. Naturally this is preventing me from actually using Dosbox-x so figuring this out is critical 😀

Any help is appreciated. I found some old posts suggesting turning ‘usescancodes’ to ‘false’ which I have done but there does not appear to be any change.

My PC has a six core Ryzen 5 CPU running at 3.4 GHz with 16 GB of RAM installed.
Also, I tried setting the cycles to 60,000 instead of 120,000 and it also made the sound worse.

Yah, that should get to Pentium 3 type speeds I'd think.

almeath wrote on 2022-10-08, 16:15:

I would recommend trying the regular Sound Blaster 16 (sb16). Windows 98 should already have the files to enable plug-and-play installation.

There is no advantage to using the sb16 over the sb16 vibra in dosbox-x. Rather the contrary, it makes the Win9x installation process slightly more complicated.

See this for more information on the emulated sb16 vibra.
https://dosbox-x.com/wiki/Guide%3AInstalling- … ated_sound_card

BitWrangler wrote on 2022-10-08, 17:27:

When you said in other posts you have a very powerful PC, did you mean relative to windows 98 requirements, or relative to modern requirements. As a P4EE or an E2160 is a powerful PC* relative to Win98 requirements on bare metal, but I'm not sure either are up to doing 120,000 cycles reliably on DOSbox. So if it's getting worse, try less cycles, like 60,000.

*Also I believe you need high single core speed, so 8 cores of Atom or Arm while "apes together strong" are kinda weak on a single.

I have nothing useful to contribute, but that Planet of the Apes quote made my day and I just had to say it . Thank you!

So does this mean that there is nothing I can do about my audio issues then?

Alevam-Inc. wrote on 2022-10-23, 21:59:

So does this mean that there is nothing I can do about my audio issues then?

TheGreatCodeholio mentioned on DOSBox-X's Discord channel a few days ago that he thinks it is due to the way DOSBox (and by extension DOSBox-X) is coded to handle SB emulation in 1ms increments. This works well for most DOS programs, but starts being a problem with Win9x and DMA transfers. Some parts of DOSBox-X have already been cleaned up, but clearly more work is needed. And right now he is overloaded with his professional job.

Windows 10 host
Version 0.84.1 / VS / 64-bit / SDL1
Version 2022.09.0 / MingW / 64-bit / SDL1
Windows 98 SE

I'm having a very odd issue with Terminator Skynet. The mouse is very sluggish in the menus, and basically stops working, however, if I keep a keyboard button pressed, it moves just fine. When I release the key, the mouse keeps moving for a second or two but then stops.
It only happens in menus though; works normally while playing (without having to press a key).

This issue does not happen in Future Shock if launched with its own shortcut (not through Skynet).

Also the mouse works just fine when running the game with PCem.

Link to recent commit "core_dyn_x86: Don't use 64-bit registers for pushl. Found this bug through another patch in the FreeBSD ports tree."
https://github.com/joncampbell123/dosbox-x/co … 575facad29e8bed

The original code had a register constraint of "r" as shown below. It specifies use of a register, but it is now changed to "Q", a more restrictive register constraint for 32-bit x86 code as generated by common versions of the gcc compiler. Will this potentially lead to register spillover into memory?

Also, the original code is not expected to access a 64-bit register on the 32-bit x86.

Sample block of code from the commit:

1--- a/src/cpu/core_dyn_x86/risc_x86.h
2+++ b/src/cpu/core_dyn_x86/risc_x86.h
3@@ -123,7 +123,7 @@ static BlockReturnDynX86 gen_runcode(uint8_t * code) {
4 		"popl %%ebp							\n"
5 		"popl %%ebx							\n"
6 		:"=a" (retval), "=c" (tempflags)
7-		:"r" (tempflags),"r" (code)
8+		:"Q" (tempflags),"r" (code)
9 		:"%edx","%edi","%esi","cc","memory"
10 	);
11 	reg_flags=(reg_flags & ~FMASK_TEST) | (tempflags & FMASK_TEST);

That's dumb. That code is never compiled/used for 64-bit x86, so the situation that it's trying to prevent can never occur.

Is the main DOSBox-X web page going to have a new release soon? I didn't read anywhere that the release schedule changed. thanks.

jmarsh wrote on 2022-12-19, 08:13:

That's dumb. That code is never compiled/used for 64-bit x86, so the situation that it's trying to prevent can never occur.

Also, in FreeBSD, from where the patch was taken, as per the pull request description, it uses "a" (the eax register). I cannot find any discussion about this. This seems to be how it entered the tree: https://reviews.freebsd.org/R11:eefcfdabff914 … c554405b4ca9c4f

Well, I found something: The same patch is used against plain dosbox, apparently added as a lazy workaround for building with an old version of clang 8 years ago: https://github.com/freebsd/freebsd-ports/comm … 5de8e767e87862c

EDIT: https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=191630

EDIT2: Also the commit description is completely wrong, as x86 does not even have 64 bit general registers.

valnar wrote on 2022-12-19, 16:52:

Is the main DOSBox-X web page going to have a new release soon? I didn't read anywhere that the release schedule changed. thanks.

The main developer has been busy with his actual work, and one of the other developers stepped down, so that leaves a bit of a gap.

But there are nightly releases you can download.

_Rob wrote on 2022-12-20, 08:36:

valnar wrote on 2022-12-19, 16:52:

Is the main DOSBox-X web page going to have a new release soon? I didn't read anywhere that the release schedule changed. thanks.

The main developer has been busy with his actual work, and one of the other developers stepped down, so that leaves a bit of a gap.

But there are nightly releases you can download.

I got a release out the day after Christmas, and I can work on it a bit more now.

TheGreatCodeholio wrote on 2023-01-04, 05:06:

_Rob wrote on 2022-12-20, 08:36:

valnar wrote on 2022-12-19, 16:52:

Is the main DOSBox-X web page going to have a new release soon? I didn't read anywhere that the release schedule changed. thanks.

The main developer has been busy with his actual work, and one of the other developers stepped down, so that leaves a bit of a gap.

But there are nightly releases you can download.

I got a release out the day after Christmas, and I can work on it a bit more now.

is it possible or you to check the dgvoodoo2 thread
i've reported an issue with screen scaling when using dgvoodoo2 and X with Archimedean Dynasty 3dfx patch

DosBox fullscreen problems with dgVoodoo 3Dfx

TheGreatCodeholio wrote on 2023-01-04, 05:06:

_Rob wrote on 2022-12-20, 08:36:

valnar wrote on 2022-12-19, 16:52:

Is the main DOSBox-X web page going to have a new release soon? I didn't read anywhere that the release schedule changed. thanks.

The main developer has been busy with his actual work, and one of the other developers stepped down, so that leaves a bit of a gap.

But there are nightly releases you can download.

I got a release out the day after Christmas, and I can work on it a bit more now.

Can you explain how to install the latest release over a current install?

buckeye wrote on 2023-01-13, 17:52:

TheGreatCodeholio wrote on 2023-01-04, 05:06:

_Rob wrote on 2022-12-20, 08:36:

The main developer has been busy with his actual work, and one of the other developers stepped down, so that leaves a bit of a gap.

But there are nightly releases you can download.

I got a release out the day after Christmas, and I can work on it a bit more now.

Can you explain how to install the latest release over a current install?

You should be able to manually download the build, unzip it, and copy it over the existing install.

The contributor that normally ran the installer builder has gone off to another project and so I haven't run it yet.

There is a pull request "Sync dh fpu state with normal fpu":
https://github.com/joncampbell123/dosbox-x/pull/3965

"...fixes the fpu state out of sync problem when dynamic core calls normal core to handle self modification code."

On testing the latest commits in the above patch with core=dynamic, there is a significant performance penalty, such as for win32 games as available to a Win95 guest OS. It may be worth verifying in other environments. Otherwise, a better option is to test performance (and the above fpu state issue) while running core=dynamic with a build based on the non-x86 fpu (instead of the x86 fpu emulation).