Jump to content

Recommended Posts

Posted

Sorry I can't reproduce it, what I can tell though is that it's starting to play a new track just after these words, so if you customised your tracks it might be related.

The track starting after this sentence is 43 (from the debug messages), see if you did something special with this track.

Eventually just move your games.cfg somewhere else and load the game without one, which will make it run with completely default settings here.

I used the emubios 3.1 + japan + console, so these are exactly the same conditions (except it's in linux and in debug mode), so it's very likely related to the tracks.

Yes, the first time it happened to me I had all the tracks associated to external mp3 files. Then, to confirm it didn't have to do with the associations, I unzipped a fresh installation and played all the way up to this part with all default settings (except the unibios setup) and it happened again.

 

The savestate that I provided is from the this new unzip, without any associations and customs settings. Please. see if you can test it in Windows and with a fresh install. Just use the Unibios because with the default Europe bios (the first from the list) it is curiously not crashing here.

 

By the way, does the savestate store the bios selection information and its setup?

 

 

Also, see if you can take a look at this later:

 

I've also noticed that if you visit the Neo-Geo bios select menu in the NG/NGCD options without a NG game loaded (that is, in a fresh run of the emulator with no bios set at all), the emu will hang if you use the up and down arrow keys to try to select a bios. Is this right?

 

 

Thank you in advance for your help! :praiseyou:

 

PS: I need to say that to use the Kagura character in the ROM cartridge version of KOF 96 I used the Unibios cheat database.

Posted (edited)

I don't know what you did with your setup, but I even tried in windows just now (using wine in linux, but it's the raine32.exe executable, the last one released).

No problem at all using unibios 3.1 japan console + your save, can't do anything more here !

video setup, I guess you are using opengl right ?

Edited by Tux
  • Like 1
Posted

I don't know what you did with your setup, but I even tried in windows just now (using wine in linux, but it's the raine32.exe executable, the last one released).

No problem at all using unibios 3.1 japan console + your save, can't do anything more here !

video setup, I guess you are using opengl right ?

I guess I've found out what's happening. In order to trigger the bug you have to trigger a specific winning quote from Kyo in the winning/score count screen.

 

The quote is "Stop moping around and butch up, you weenie!"

b1d1cd499791477.jpg

 

How to trigger this quote? I don't know exactly. What I am doing is waiting the introduction movie play until you can see "Violent fighting to come again!" in the screen, then load the savestate.

2f39c1499791478.jpg

 

I did this for at least 3 times and could trigger the Kyo quote needed to reproduce the crash.

 

If even with this you can't reproduce it, please see if you can have a quick try in a Windows installation because Wine may not be perfect for this specific testing. I have tried this in another machine with another intel processor and with integrated graphics and could easily reproduce it there (as long as I can trigger that winning quote).

 

Thank you once again for you patience. :praiseyou:

Posted

Ok, reproduced and fixed. It's a very special place to find a speed hack, the pc gets negative if you read it on 4 bytes here (the 68000 reads it on 3 bytes), this is what creates the crash. You can work around it by simply disabling speed hacks in the neo options.

 

I had a look at your freeze while selecting a bios without any neogeo.zip file in your rom directories, it's pretty obvious, but anyway I added some message for this kind of case.

I think you'll agree it's not worth releasing a binary for that !

Anyway it's pushed to git.

I might eventually have a look a this libretro one day, but weather is too warm for that now, so it might take some time !

  • Like 1
Posted

Ok, reproduced and fixed. It's a very special place to find a speed hack, the pc gets negative if you read it on 4 bytes here (the 68000 reads it on 3 bytes), this is what creates the crash. You can work around it by simply disabling speed hacks in the neo options.

I'm glad you could reproduce it. Thank you for fixing this one!

 

 

I had a look at your freeze while selecting a bios without any neogeo.zip file in your rom directories, it's pretty obvious, but anyway I added some message for this kind of case.

Thanks for this too!

 

 

I might eventually have a look a this libretro one day, but weather is too warm for that now, so it might take some time !

Ok, no problem. If you could just find a way to convert these shaders to the format used in Raine it would be great already.

 

 

By the way, I"ve turned off Double buffer in the OpenGL renderer options and now the GUI shows again in fullscreen! :-) Why is double buffering causing this now? :wacko:

 

 

Finally, I'll leave here again some information about that Marvel vs Capcom graphics glitch which I had reported in the old forum. Perhaps you can check it some day. :sweaty:

You can see the glitches mainly in the screens below:

 

character select screen (little black squares)

a77697500112644.jpg

 

the floor in the Onslaught stage (black)

a18745500112640.jpg

 

Onslaught's head (missing)

702fc5500112642.jpg

 

You can find a savestate for the Onslaught stage here:

 

 

Thank you so much for your continuing work on Raine. :praiseyou:

Posted

Double buffer : it shouldn't happen, but it happens sometimes, a real bug in windows (but there is no such double buffer in linux), I never call the function to swap the buffers but sometimes when returning the gui it doesn't display the right buffer. To work around that in sdl2 they removed the double buffer in windows too by getting entirely rid of the 2d functions, everything is 3d now, with some hidden complexity, quite extreme, but it works.

Anyway there is a fix for that for next version, it simply disables this double buffer when using opengl.

 

The shaders : actually I knew about them but I had forgotten, I had a copy of this on my disk from 2013, 3 years already ! The main problem with this is they are using Cg which was a very good toolkit from nvidia (to unify opengl and direct3d), but it's discontinued for I don't know which reason, since something like 2013... The other problem is the multi pass shaders would probably require to use frame buffers which I never used, it would be tricky. This or trying to blindly use their api, don't know yet how easy it is to do... I am going to try to have another look at that later though...

 

And the famous cps2 game with display bugs : afaik it's the only cps2 game of this kind, it uses display interrupts to change the rendering and they are not emulated by raine (and it's the only game using this - with its clones). Add to that that I almost never played this game, maybe once or twice only, so motivation to do that is very low (it's complex to do, synchronize cycles from the cpu with refresh rate on screen... !). So it's unlikely for now that I'll ever fix this.

  • Like 1
  • 3 weeks later...
Posted

hum, should I mirror this ? What are the changes from the last "official" build ?

  • 3 months later...
Posted

Gotta be honest here im not fully up on what CPU cores Raine has, But just a small heads up to mention when it comes to the recent MCU dumps that add full sound to Fire Shark

Teki Paki and Vimana and if you were interested in hooking up the sound for these games in Raine if you dont have the Z180 CPU a Z80 CPU will do the job just fine confirmed to

work in my olde MAME Src and Currant Final Burn Alpha builds.

Posted

Gotta be honest here im not fully up on what CPU cores Raine has, But just a small heads up to mention when it comes to the recent MCU dumps that add full sound to Fire Shark

Teki Paki and Vimana and if you were interested in hooking up the sound for these games in Raine if you dont have the Z180 CPU a Z80 CPU will do the job just fine confirmed to

work in my olde MAME Src and Currant Final Burn Alpha builds.

 

Not sure I have the motivation to do that now, mcu stuff is pretty tricky usually, but thanks for the info anyway !

  • 1 month later...
Posted

Hi, it's dink from the fbalpha emulator project. we found a way to get perfect sound with f3 games (dariusg etc) with avery simple algo. If interested I can share the code (es5506.cpp)? I think this will be a nice addition to Raine.

Of course you can always get the code from www.fbalpha.com after the next release with added f3 support, but the next release might be a couple months away...

Posted

Well there is no issue with the f3 sound afaik, but ok send the file, I'll take a look (we shared the one in raine with mame but it hasn't been changed for a very long time and I doubt the interface is still compatible, but anyway, send it !).

Posted (edited)

Some f3 games sound OK in Raine, but some are completely wrong like Darius G - especially the intro/attract mode sequence. This is due to the es550x chip playing sounds at different samplerates, but the linear resampler is not right. It's an easy fix. porting will be simple, just a small amount of code needs to be ported from our ES5506Update to your es5506_update.

Code & info: http://pastebin.com/4m8JKkNK

 

best regards,

- dink / fbalpha, resampling code by iq_132 of fbalpha project.

Edited by dink
Posted (edited)

The videos on youtube don't tell if they were recorded from the arcade or from mame, so it's hard to tell but assuming this one is from the arcade : ( video sjv7lpp5nTU in case it doesn't show in this quite broken forum !)

https://www.youtube.com/watch?v=sjv7lpp5nTU

I guess the difference you are talking about is the sound which can be heard while seeing the astronaut in the 1st images just before zooming on the planet, right ?

Anyway the intro is the place where we see the most the bad zooming emulation in raine, I had hoped Antiriad would finish this one day but he lost all interest in this long ago (about 17 years ago actually !).

 

Well the code is quite different if you had a look at the code for raine in the git section, we still don't use the c++ interface for one, and the update function uses 2 levels of while loops, I'd admit your code is clearer, there was some hasty testing in this area in raine and nobody never tried to clean things up... I am not sure I want to risk a rewrite just for this missing sound, I'll think about it...

Edited by Tux

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...