Jump to content

Tux

Ultra Members
  • Posts

    1,056
  • Joined

  • Last visited

  • Days Won

    219

Everything posted by Tux

  1. I've been thinking about the sh2 in raine for quite some years now, but I finally got a binary which starts to work. The motivation is simple : there are quite a few modern arcade games which use this cpu, and it's an interesting one, what the 68k cpus could look like if they had evolved towards the risc family of cpus. For now my choice was to use the gens sh2 emulator from there : http://www.gens.me/downloads.shtml It's a pure x86 asm emulator like in the old raine days, but the criteria of choice was that it's the only one which is open source and usable directly with some documentation and no need to extract the emulator from a big project, it's really the only one of its kind in this category. After a few expected problems I was able to add "gunbird 2" and "strikers 1945 iii", which are fully playable, but in 32 bits only then. Now the surprise is that I have some serious bugs with some more sh2 games, like tetris grand master 2, no sound at all, and the game hangs when the music is supposed to start. At first I thought it was a problem with the sound chip emulation, but after a long investigation, it's not. In this case, it's a problem in the sh2 emulation. Fixing it would probably be impossible, or I'd have to be very lucky, the 1st big problem is to find the instruction causing this, and since it doesn't crash but just has a bad behavior it's not going to be easy at all. So now I am wondering : should I release this 32 bits binary anyway now ? I think it's not a bad idea anyway, even if I later change the sh2 emulation used... There is a big heat wave here anyway for now, so I'll wait until the temperature cools down to decide, but comments welcome anyway ! If I change the sh2 emulation, my next choice would either be kronos, yabause (these 2 are actually related), or mame. 1st choice would probably be kronos, there is a "simple" sh2 emulator in it, with a more advanced one, the simple one would allow to at least test the games with problems. The version in mame might not be so hard to extract, not sure about it really, things can get really messy in mame.
  2. Errr, maybe for Italians, but for them only !
  3. Oh there were tools for that, html2pdf but it was made before the web became crazy and it was mainly for simple web design. Might work with zzap though, after all it's still some quite old html too. But I am sure you prefer to revise it at the same time and change a few things manually instead of having something automatic do it !
  4. There is no pc driver, the auto detection problem is in linux only since kernel 5.16 or so, so if you boot only windows it will always be recognized as an xbox controller. And even for me, if I want the xbox controller the only reliable method is to boot windows 1st, then reboot in linux, too bad they didn't include something to force this auto detection to something... ! It's annoying, but it's just that, annoying, you can work around it.
  5. Yeah that's the way fb is working, I stay away from them now too... !
  6. While on the subject of controllers, I tested this cheap controller on my part : https://www.amazon.com/gp/product/B07WBYFX24/ref=ppx_yo_dt_b_asin_title_o05_s00?ie=UTF8&psc=1 The plus : it's solid. The minus : the d-pad is really not precise, not advised if you want to use this for fighting games ! Also from kernel 5.16 in linux if i remember correctly, this thing is recognized as a nintendo switch controller instead of an xbox one, which means inversion of buttons and the sticks feel slightly different (if you reboot in windows after that, the sticks are unusable, windows can't adapt to this config ! you need to turn the computer off to reset the autodetection so that it can recognize it as an xbox controler). Ah, and a last thing, it has some long usb cable to charge the battery but oddly it can't be used as an usb controller, you must use it in wireless mode, eventually charging its battery with the usb cable at the same time ! But I have it since 2021, and it still does its job, so it's not bad. You can turn off the annoying lights of course. The battery is charged quite quickly and lasts for long sessions after that so it's ok.
  7. He added a page for the famous issue 12 which is italian only, and paper only, well at least you get a page now... ! Except that fixed some links mainly, a few typos, this kind of thing...
  8. ... and way too small pictures to read anything !
  9. I guess you don't even read the story in these games now so you don't care about the translation, but except some sprites modifications there is also some translation for the story mode in the perfect edition... ! It seems to me something like a final++ version, not equivalent to the perfect one on other platforms, but still a little better than the basic final version.
  10. Thanks for all the info, and your english is improving ! I guess if pf has some translations included it can be better than nothing then, even if it's not as good as the pc version. It also means these 2 games can probably accept most of the cheats coming from the special edition, but that's a lot of cheats to test... Maybe later...!
  11. Oh finally for perfect the roms look mostly different so it's probably a worthy clone after all : most sprites are in common (except the 2 last roms), the sound is identical to what's used by fe (for the samples but not for the z80 code !), except that the 68k & z80 have totally different roms, plus the text layer has a different rom too. So it really doesn't look like a tiny clone, now you'll tell me if it's worth playing it ! Neogeo id : $273, not used until now, that's why it isn't recognized by unibios for the cheats. Added to git too (samsh5pf).
  12. Well thanks for the info. I am afraid I won't be able to tell much about the interesting points of this rom, but I can at least tell after looking that for fe (final edition) it's an entirely new game, no rom is in common with any other neogeo game contrary to most clones usually. And it's not encrypted, logical if it was made by the community. For the cheats, unibios recognizes the game because it has exactly the same rom id as samsh5sp (special edition), that is $272. The ram usage might be the same, but the rom is very probably different, anyway there are probably only ram cheats in unibios because it deals with real roms most of the time so they are really read only for it. And by the way fbneo has some source on github too, and it's quite different from mame source actually, which is a surprise because originally it was made for afterburner before it was integrated to mame, and it was something like a showcase, just to demonstrate the game worked well. So I assumed it was based on mame source, I was wrong apparently ! But it's very readable, and this part is easy to understand since there is no crazy encryption like what the special edition has. It seems to be Japan only, it ignores the region. I'll check this perfect thing later, it sounds less interesting, this one seems to be just a hack on top of this final edition. But if it has no cheats in unibios it means they changed the id of the rom, so maybe there are some heavy changes there too... Anyway samsh5fe is added in git.
  13. For the 1, no having a setting for default region would be overkill, the default region is set to Europe for neocd games because raine is made in Europe. I guess I could detect the current locale and use japan instead if the locale is japenese, but even that would be quite overkill. It's not a problem of correct or not, it's just the game is not recognized before the 1st part is loaded. Once it's loaded if you reset the name doesn't change, so it knows how to take the right settings. So if you really want to see this animation, there's just a rest for you. It's not an emulation bug neither, the neocd console has a fixed region so it always plays all games using the same region. Here we choose to allow the player to set the region independently for each game. So that produces a difference, but my goal is once again not to emulate precisely the console itself, but just to support neocd games as well as we can, that's all. The games are good, but the console itself is not interesting. For the 2, I'll try later without any configuration file although I doubt it will make any difference, but it's annoying, it's always longer to retrieve the files needed in this case. I reproduce it very easily with last blade 2, but not samsho3, go figure... ! Anyway, it was probably here since the beginning of the sdl2 version, so quite some time ago already, but it's right we played less neocd games with that version, and it was the kind of thing which could happen. Anyway the fix is just 1 line, very easy to fix. It affects the games with a width of 320 (although for some reason it's not so obvious for me in samsho3, there might be some other parameters there). Anyway it's fixed... thanks for the report !
  14. For your 1 you could have guessed the reason : when you load a neocd game the emulator doesn't know which game it is before it loads at least the 1st part, and so it does not know which region to apply 1st, so you have the default region 1st, that's all. Then once the game is identified, it gets the last region used for that game or its default region. For your 2 I can't reproduce it for now, launched smasho3 with the us region in windowed mode (in linux), started the game, the picture is not cut on its right side, and displaying the gui doesn't change anything to it. If the bug is just because you are using a 640x480 window, just forget it ! I am not sure for now. edit : and after testing in wine, I still can't reproduce this. No idea how you did that. edit2 : and even in win10, and even with a 640x480 window, no cut on the right. Oh well... !
  15. Actually this version number was displayed for neocd in the title bar until may 29th 2021, commit 90761c9f7afc679394d550c4eb4f5e6ef3b42358. The commit was not specifically about that but to fix an incompatibility between 2 games, I guess I probably took it off because I found it made too long titles which is probably true for a lot of games there. Oh well since you are the most efficient bug finder around here I can put it back... !
  16. Ok it's fixed, ironic find, it's a specific cue file which creates the crash, and from internet archive, I hadn't even noticed they had neocd files. I won't put auto-download from there though because such a format is really a waste of space ! Anyway the indexes are taken into account that's why there was a crash here, that's indeed the index which crashed simply because it tried to seek in the file before loading it... !!! At least the fix was short. I don't plan to release a binary just now for that, this bug has been here for years ! For info to reproduce it, I had an iso + wav files dump and I hadn't even converted the wav to mp3 (lazy... !). So I just adapted my cue file to look like yours. Then I made a savegame during the demo just before the highscore screen appears. For me it crashed just after the hiscore screen. This way it was very fast to get to the crash, these neocd saves are super efficient ! And finally converting my cue file was not even necessary it would have crashed even with the one I had originally, it's just that nobody left the demo running until now, and for some reason this kind of crash doesn't seem to happen during normal play... !!! Quite crazy ! But at least it was easy to fix. It's a super old bug here again, from neoraine time, you are super efficient to find bugs currently !
  17. Well I'll look into this eventually when the temperature will go down a little, not for now. Yes you can simplify a lot this cue file, I am not sure but I think the index lines are simply ignored for the mp3 files I'll have to double check that, and the flags lines are ignored for sure. Version is not really important in the window title, there is the nice about dialog for that with its superb color cycling despite what sdl-image did lately ! Yeah I am not sure either it's really the windows version, it's you who proposed the idea and until now I just thought it had been fixed in sdl, not even in the graphics driver. Whatever the issue, it's on your side though. You can forget it for now if you want since we can live with a white cross in windows, but it's definitely a problem on your setup.
  18. I haven't got the latest one, but still more recent than that, I had to allow an update specifically to be able to update the nvidia driver, I couldn't install the latest version without that, so I guess you don't have the latest nvidia version neither ? It must be the 22h2 if I remember correctly, I disabled windows update after that and I hope to keep it at this level for long. Well you are lucky it's minor, just a white cross instead of a green one, because supporting such a specific windows version would seem quite crazy. Anyway... ! edit : confirmed after verification, 22h2, installed late in march 23 because I blocked windows update before, and I still blocked an update on this version (I can now update the nvidia driver so it's enough for me).
  19. Ok 0.95.5e posted, no dll inside, you'll have to download the dlls32-0.95 or dlls64-0.95 package again if you install in an empty dir, sdl2 got updated to 2.26.5. The changes are : a lot of neogeo games which had an horizontal resolution of 304 pixels are back to the normal 320 value, that's the biggest fix in this one. sfz3j which had been removed in 0.95.5d is back since sfa3 doesn't contain the text from japan ! Also removed the japan region for all the sfa3 games and clones. And re-enabled the workaround for some green colors in windows when displaying the green cross in the gui, I can't reproduce anymore this one here, mer-curious is the only one who can for now, more feedback welcome. I really hope it's the last 0.95.5 version, all this is starting to be annoying !
  20. It's not such an old bug after verification, it was fixed on feb 2022, you can see the commit here : (I thought it was older) https://github.com/zelurker/raine/commit/e5ed04b19393f70144d9cf19f7280eb25072a9a8 Then when I noticed I couldn't reproduce it anymore on my win10 here, I assumed it had been fixed on sdl. It seems there was something else, but I can't guess what. The workaround was disabled 1 year later, in feb 2023, here : https://github.com/zelurker/raine/commit/d3ae87763 I assure you I absolutely can't reproduce this anymore at all here, the nvidia driver is 531.61 from april 13th. Well I would have liked to be able to reproduce this, I would have tried to find something better, but I can always re-enable the old workaround, you'll get white crosses instead of green ones in the gui in windows from now on... !
  21. It's all in italian, but there are nice pictures, you can see the page of the kof98 test, which also talks about ips patches. And you can translate the text if you want, by using google lens on a cell phone for example ! Here is the link : https://zzapmagazine.blogspot.com/2023/05/zzapraine-numero-12-in-arrivo.html?m=1
  22. It's an old windows bug, but it was fixed in some sdl version, make sure you are using a recent sdl (latest SDL2.dll date is april 2023, version 2.26.4 in the about dialog). If you still have the bug with the latest sdl, then maybe the bug is fixed only for some graphic drivers and in this case I'll have to take my workaround back. For now I can't reproduce this at all, with wine in linux or directly in win10.
  23. Not for me, I tried the same game, samsho2, I don't have any sound association for it but opened "manage associations" anyway, which was empty, absolutely no effect on the game (except of course it stops the currently playing track if there is one).
  24. sfz3a is using english text so for this one it's sure sfa3 is totally equivalent is setting the region switch to asia. For sfz3j, it's Japanese text which is obviously not in the rom, so I'll take just this one back for now, but I'll have to check those I removed earlier... progearj seems ok when using progear with japan region at least, so it's probably something specific to sfa... ! 1944 with japan region has almost no text from japan, it even writes USA in its 1st screen but not the copyright screen. So I'll assume the only problem was with sfz3j, if anyone finds another problem, let me know !
  25. I was mistaken, I thought there was a new decryption code finally allowing to totally decrypt the rom, but no !!! In 2023 we are still stuck with opcodes / operands, and by the way raine was already using the same algorithm as xcopy when saving a cps2 rom in debug mode, but what you get are the decrypted opcodes. Some areas of the rom are totally readable directly, like the 8 1st bytes : $ff8000 this is the initial value of the stack pointer at offset 0, but if you go only to offset $60, same file, it should be the vector for irq0, and this vector is obviously unreadable, it's $14e for sfa3/sfz3j but it must be read from the encrypted area. What this means is the roms can't be compared directly, there's still no reliable way to do this ! Ton of crap... ! When you think that the rom encryption keys were in some volatile ram using a battery and so when the battery died you got a useless machine, it would probably have been declared illegal in this time, that's obvious planned obsolescence... ! And so no surprise that you are still unable to get a fully decrypted rom from that. Well, maybe I should just take back all the removed sets, or at least sfz3j then, I'll see... !
×
×
  • Create New...