The race itself still can't start neither from demo, quick race, new game.
From quick race and new game I see level descriptions being corrupted palette mismatch and everything have symbols or numbers.
Credits fully played without any issues at fullspeed,..

All other rhings works exactly as before.
Game uses 75hz,if you can enforce it to 60hz my capture card will capable to doit, and I can show you video in case you need it.

Thanks Raven. I can cap the refresh to 60 Hz, but that won't be necessary. The race still won't start.

That's the thing I was trying to fix.

Multiple command lists have been enqueued and can't tell which command is causing the problem.

I'll get you a refresh rate capped version in a bit.

Refresh.txt should contain a single number the refresh rate. The supplied file contains 60.

Usually V_SetDisplayMode's refresh rate parameter is set to 0: maximum refresh. If this file exists, the refresh rate will always be forced.

(Broken paletted textures with refresh rate adjustment)
https://1drv.ms/u/s!As-dKk-N73dSl04Mo-Zu9Ad0nTvO
MD5: b9cf75fddcccfe99c508362202116739

(Black screen version with refresh rate cap)
https://1drv.ms/u/s!As-dKk-N73dSl08Mo-Zu9Ad0nTvO
MD5: 417520fde50647c2d7906bcb744f0516

Videos from both would be nice. If that's too much work then just the second. Maybe there's something that can be seen.

I'll add more logging and get you a build tomorrow nightish.

I need to focus on Speedy.

https://www.twitch.tv/videos/2063411074?t=00h53m30s

Thanks Raven. It seems kind of painful to test.

I'm not putting enough effort into this at the moment and I'm still convinced it's my rapper's fault.

I'll get back to this when I can put all my time into it.

OK anytime will be glad to test anything)

This will still lockup, but it should produce logs that I can compare to regular RReady:

https://1drv.ms/u/s!As-dKk-N73dSl1AMo-Zu9Ad0nTvO
MD5: e23f4219d630a3fb4a88a445f59d872a

It'll be ridiculously slow. That's if it doesn't crash in other ways.

done

Thanks!

Could you try this? It has changes to both logging and behaviour.

https://1drv.ms/u/s!As-dKk-N73dSl1cMo-Zu9Ad0nTvO
MD5: 3e216ebb330d137e3ba6aceeeb861bdb

https://youtu.be/pB_MDPdZAao

https://youtu.be/LlpHYlIuQr0

I actually messed up my logging, but it doesn't matter. The problem doesn't appear to be where I thought it was.

I actually think a protection fault (memory access violation) is bringing the game down, but because it's crashing the game isn't exiting full screen mode.

The fault is almost certainly mine.

Last one for today. A video won't be necessary:

[Skip This]
https://1drv.ms/u/s!As-dKk-N73dSl1oMo-Zu9Ad0nTvO
MD5: 85cd4601bc3678081acabe351c6511a0[
[/Skip]

Try this instead:
https://1drv.ms/u/s!As-dKk-N73dSl1wMo-Zu9Ad0nTvO
MD5: d9666a9253b623b49933a68fec58e0a7

I've exported a full set of Verite functions from the v2k's verite.dll. As far as I can tell system verite and redline statically link to a subset of them. But I've exp0rted the whole lot just in case something's being dynamically loaded.

skipped, tried next

Thanks.

Did it return to the desktop?

https://1drv.ms/u/s!As-dKk-N73dSl18Mo-Zu9Ad0nTvO
MD5: d8f65beab8bdabddb29139db505bc5b9

I found a possible issue with paletted textures. The stride value is 64 for a 256x256 pel 16 colour texture.

I think the stride should be 128 (half a byte per pixel, so the stride, the number of bytes between rows should be 256/2=128). I don't know if strides are based on 16 bit words. If they are then 64 x 2 word (x 2)is correct.

What I've done now is set the texture to match to match the stride when selecting. I've also added extra logging for VL_LoadBuffer for paletted textures. If the paletted textures are the problem, then this is where it'll probably go wrong. The block of memory passed to VL_LoadBuffer might be too small for the board to DMA across or something. It might cross over to uninitialised memory.

sharangad wrote on 2024-02-21, 10:51:

Thanks.

Did it return to the desktop?

no