Duke Nukem 3D - New version work on

[torsti76]

Wow, THAT's what I call a README!
Others should follow this shining example of elaborate documentation.

Thank you very much for the release, will try it soon!

[BSzili]

Haha! To be fair this readme has been accumulating through multiple games, and there are about 40 lines from the original readme describing the multiplayer. I did my best to describe everything you can tinker with though. Anyway let me know how it runs.

[paraj]

Good job

Getting around 10-16 FPS on my 50Mhz 1260/16MB (320x200 AGA), but it feels pretty responsive and plays just fine.

However it seems to guru most of the time on startup (sometimes just after opening the screen, sometimes just before the main main) and every time on exit with either an address error or illegal instruction (80000003/80000004).

Running w/o startup-sequence, but with setpatch + idefix. I see the same behavior in winuae with the 040 version (040 + both with and without JIT).

Tried with both the full and shareware versions. Also tried deleting the grpfiles.cache file and doing a clean install.

Not something you need to spend time on, as I won't be playing it much with my shitty amiga mouse, just figured I'd report it.

[Aardvark]

Working flawlessly with TF1260, very impressed by the FPS [ Show youtube player ]

[BSzili]

Originally Posted by paraj

Good job

Getting around 10-16 FPS on my 50Mhz 1260/16MB (320x200 AGA), but it feels pretty responsive and plays just fine.

However it seems to guru most of the time on startup (sometimes just after opening the screen, sometimes just before the main main) and every time on exit with either an address error or illegal instruction (80000003/80000004).

Running w/o startup-sequence, but with setpatch + idefix. I see the same behavior in winuae with the 040 version (040 + both with and without JIT).

Tried with both the full and shareware versions. Also tried deleting the grpfiles.cache file and doing a clean install.

Not something you need to spend time on, as I won't be playing it much with my shitty amiga mouse, just figured I'd report it.

That doesn't sound very good. Does the WinUAE guru happen without startup-sequence as well? If you could send me a config file I'd like to look into this, there's always room to improve compatibility.

Originally Posted by Aardvark

Working flawlessly with TF1260, very impressed by the FPS [ Show youtube player ]

Using a PC as a MIDI synth, that's pretty funny How was the framerate in RTG mode?

[Aardvark]

Originally Posted by BSzili

How was the framerate in RTG mode?

Around 25-40 in 320x200, 8-10 in 640x400.

[paraj]

Originally Posted by BSzili

That doesn't sound very good. Does the WinUAE guru happen without startup-sequence as well? If you could send me a config file I'd like to look into this, there's always room to improve compatibility.

Config file attached. Tested in 4.9.0 and 4.9.1b4.



I made a fresh install of WB3.1 (reported as: WB version 40.42, KS 39.106). Boot with no startup-sequence, run setpatch, run duke3d.040 => guru 8000000B (line F).


Installed latest mmulib (I have this on my real amiga as well) and guru changes to 80000004 (iillegal instruction).


I start the game and press enter a couple of times to skip the intros. Gurus just after showing the "alien" background (see screenshot).


Same result if I change CPU to 060 (and run duke3d.060).


I can zone the HD-directory if that makes it easier.

[BSzili]

Just gave it a quick test with an existing WB3.1 install, and it didn't quite work out. If you can put the HD dir in the Zone it'd be much appreciated.

[mateusz_s]

Originally Posted by BSzili

After being on hold for a while, the first version it's finally available on Aminet:
https://aminet.net/package/game/shoot/jfduke3d

Great thanks!!

[pcotter]

Tested on PiStorm/Emu68 and Aros :-)
[ Show youtube player ]

[paraj]

Originally Posted by BSzili

Just gave it a quick test with an existing WB3.1 install, and it didn't quite work out. If you can put the HD dir in the Zone it'd be much appreciated.

In the zone as duke_test.zip. I've modified startup-sequence to boot straight to it (after setpatch).


The crash seems to happen just after 4001ae74 (exception 3 now, but I've also seen the instruction following negx.b d6 be illegal (7716).

[paraj]

Ahhh, that address is suspiciously close to the stack pointer.

Increasing the stacksize seems to fix all issues

[mateusz_s]

Originally Posted by BSzili

Just gave it a quick test with an existing WB3.1 install, and it didn't quite work out. If you can put the HD dir in the Zone it'd be much appreciated.

Works on original Duke Nukem 3D or Atomic Edition only?

[BSzili]

Originally Posted by paraj

Ahhh, that address is suspiciously close to the stack pointer.

Increasing the stacksize seems to fix all issues

Hmm, interesting! How much stack did it need? Currently the hardcoded minimum is 32KB

Originally Posted by mateusz_s

Works on original Duke Nukem 3D or Atomic Edition only?

Atomic Edition is not required, it should work with all versions listed at the bottom of this page:
https://www.jonof.id.au/jfduke3d/index.html

[paraj]

Originally Posted by BSzili

Hmm, interesting! How much stack did it need? Currently the hardcoded minimum is 32KB


Atomic Edition is not required, it should work with all versions listed at the bottom of this page:
https://www.jonof.id.au/jfduke3d/index.html

Checking with the same setup I zoned, it crashes on exit with less than 32600. 32768 seems to work, but is probably too close for comfort.

[BSzili]

That is bonkers because that's the exact amount I set as the minimum:
https://github.com/BSzili/jfbuild/co...beba3244a7b141
I'll have to check with Scout what's going on, thanks for the test archive.

[paraj]

Originally Posted by BSzili

That is bonkers because that's the exact amount I set as the minimum:
https://github.com/BSzili/jfbuild/co...beba3244a7b141
I'll have to check with Scout what's going on, thanks for the test archive.

NP. My best bet would be that the __stkinit stuff simply doesn't work/isn't called (at least in this setup). Just for the record I'm testing with calling "stack <number>" before starting the exe and FWIW starting from the icon in WB gives the same result.
I tried putting a breakpoint on StackSwap but the results seem a bit odd (maybe I'm doing it wrong).

Also tried building your amiga branch (to just put in some printf statements), but there was too much friction. Managed to download what I think are the correct dependencies and get a working Makefile.user, but had to make some code changes and the resulting executable fails when loading table.dat. It's not something I think you should spend much time (if any) on, but you could consider extending the README with your environment settings.

[manossg]

Just tried it on my V4 and it plays fantastically! What a great job!

[mateusz_s]

Originally Posted by BSzili

Just gave it a quick test with an existing WB3.1 install, and it didn't quite work out. If you can put the HD dir in the Zone it'd be much appreciated.

Can I ask a bit technical question?

Do you manage screen-tearing some how?
I was looking thru Blood src code once and noticed
ChangeScreenBuffer() I guess (but not sure if it was for AGA or RTG)

I am working on my raycasting engine and I used system Triple Buffering
to avoid screen tearing - which works... but not under Aros

Do you provide maybe an Amiga OS framework sources somewhere to see?
Or maybe a little tutorial?

[BSzili]

Originally Posted by paraj

NP. My best bet would be that the __stkinit stuff simply doesn't work/isn't called (at least in this setup). Just for the record I'm testing with calling "stack <number>" before starting the exe and FWIW starting from the icon in WB gives the same result.
I tried putting a breakpoint on StackSwap but the results seem a bit odd (maybe I'm doing it wrong).

Also tried building your amiga branch (to just put in some printf statements), but there was too much friction. Managed to download what I think are the correct dependencies and get a working Makefile.user, but had to make some code changes and the resulting executable fails when loading table.dat. It's not something I think you should spend much time (if any) on, but you could consider extending the README with your environment settings.

That's my hunch too, but I'm not sure how that could happen. I already spent plenty of time getting stack swap to work, so I'm absolutely determined to figure this out
If you had to create a makefile.user file then something probably went wrong. In theory all you need of bebbo's toolchain, and the cross compile scripts should build the game out of the box.

Originally Posted by manossg

Just tried it on my V4 and it plays fantastically! What a great job!

Nice, thanks!

Originally Posted by mateusz_s

Can I ask a bit technical question?

Do you manage screen-tearing some how?
I was looking thru Blood src code once and noticed
ChangeScreenBuffer() I guess (but not sure if it was for AGA or RTG)

I am working on my raycasting engine and I used system Triple Buffering
to avoid screen tearing - which works... but not under Aros

Do you provide maybe an Amiga OS framework sources somewhere to see?
Or maybe a little tutorial?

For the platform specific code minus the sounds look for amilayer.c in jfbuild. Thomas Rapp has an excellent example for synchronized double buffering:
http://thomas-rapp.homepage.t-online.de/examples/dbuf.c