VOGONS


First post, by vintagefugg

User metadata
Rank Newbie
Rank
Newbie

Hi folks.

This post is forking off from the tradition UNISOUND thread after my comments and the follow up discussion here Re: UNISOUND - Universal ISA PnP Sound Card Driver for DOS v0.82a
Once again, all praise to JazeFox for building UNISOUND. If they come back from their unannounced hiatus and are unhappy about this I'll hand all my stuff over to them and delete this.
I appreciate the PMs but I am still too new here to reply so please put your bug reports, dump output links, and feature requests below where I can reply. Hopefully one day I can PM people.

I've been working on some fixes and features requests. I've also tested with a decent chunk of my sound cards here and I'm pleased to announce the latest release:
Unisound v0.82a-unofficial-fix-4

I've put the release zip up on github along with the decomp. Since the last release (fix-3 in the linked post above) I've made the following changes:

New features

  • UNIDUMP.EXE: UNIDUMP > DUMP.TXT writes one report of the machine's sound hardware and the system around it (DOS setup, BIOS, PCI, IRQs, the PnP BIOS's devices, every ISA PnP card's ID and resource data, what answers at the usual SB, OPL, MPU-401, WSS and PSS ports). Attach it to problem reports, especially for a card UNISOUND doesn't find or support. It changes no card's settings.
  • OPTi 82C929 (MAD16 Pro): /NOPNP sets it up in SB Pro mode, with the MPU-401 from BLASTER P and U.
  • WSS mode on the OPTi 82C928, 82C929 and 82C930 (/NOPNP /XWSS): the codec gets its own IRQ and DMA from ULTRA16, with a SET ULTRA16 line to add when it's missing. On the 82C930 only the WSS settings are shown, since its SB part has no IRQ in WSS mode.
  • OPTi 82C930 IDE CD-ROM port: C170 turns it on, C0 off, and without C it's left as it was. Fixed at 170, IRQ 15. (Darmok)
  • More cards recognised (experimental, not yet tried on a real card): ESS0100, ESS0102, ESS1878, ESS1879, any CSC??25 or CDC??25 (CS4235), AZT1021, RTL3000, DXP3201 (MED3210) and CDC1111 (Dynasonic Pro). If you have one, please send logs! I've tested some of these on 86box and ordered more sound cards on ebay to test them myself but it's getting a little expensive.
  • UNISOUND_CD: BLASTER's IDE fields C and Y can go in their own variable, for programs that reject unknown BLASTER fields.
  • BLASTER suggestion adds P for the MPU-401 and E for an AWE32/AWE64's EMU8000.
  • /LOG prints the resource data of a PnP card skipped as unknown, and more OPTi registers.
  • UNISOUND.COM is now a small-model EXE under the same name; AUTOEXEC.BAT lines keep working. This gave me the space to implement new features without constantly fighting to maintain stack space. I've tested this on MSDOS 3.30, MSDOS 5.00, MSDOS 6.22, and FreeDOS 1.4 and it seems fine.

Bug fixes

  • AZT1022: Aztech's AZT2320 card with this ID is set up as an AZT2320 again, not as an AD1816 (experimental).
  • /NOPNP no longer reports an OAK OTI601 / OPTi 82C928 that isn't there. (analog_programmer)
  • /XESSB without BLASTER R tries control port 660h first, then 800h.
  • Non-PnP SB16 check: each DSP is reset before its version is read, and an older DSP (1.xx-3.xx) is named before the /XSLAM and /XESSB hint.
  • Non-PnP SB16: the IRQ is shown in decimal (IRQ 10 was IRQ:A).
  • OPTi 82C930: the CD-ROM interface setting is kept, so OPTi's CDSETUP.SYS /T:I in CONFIG.SYS followed by UNISOUND without C now works without DEVLOAD. v0.82a and fix-3 turned the IDE port off.
  • OPTi 82C931: the IDE port is set up in the order its resource data gives, and /LOG reads the codec at its own port.
  • CD-ROM drivers: UNISOUND runs after CONFIG.SYS, so load the CD-ROM driver after it with DEVLOAD, as UNISOUND.TXT says (or, on the 82C930, use CDSETUP.SYS as above). A warm reboot doesn't help on the boards we tried: they reset the card.

If you have a request for a new card, a feature request, or a bug report please attach the following output to a new comment here: UNISOUND /LOG > LOG.TXT and UNIDUMP > DUMP.TXT.
I'd prefer bug reports here for vogons users because this is the home of unisound. If you have to use github issues do please still post here indicating that you did so and link to your issue.
Be aware that UNIDUMP will dump your config.sys and autoexec.bat by default for now. I'll add an option to skip just them later on. Feel free to delete them from the DUMP.TXT manually if you're uncomfortable with this.

Cheers,
Vintagefugg

Reply 1 of 11, by vintagefugg

User metadata
Rank Newbie
Rank
Newbie

I also don't have permission to edit posts so I guess I'll update here for now: I've added /NOFILES argument to UNIDUMP for reporters who are uncomfortable sending their autoexec.bat and config.sys.

Reply 2 of 11, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie

Hi, vintagefugg,
And thanks for not giving up on your UNISOUND fork with bugfixes, despite the hater's attacks in the original thread!

If nothing's changed in the forum's rules, when you reach twenty posts, you'll get the permissions to edit old posts and send PMs.

The word Idiot refers to a person with many ideas, especially stupid and harmful ideas.
This world goes south since everything's run by financiers and economists.
This isn't voice chat, yet some people overusing online communications talk and hear voices.

Reply 3 of 11, by vintagefugg

User metadata
Rank Newbie
Rank
Newbie

Addressing the PM from analog_programmer regarding their ES1869F and the previous release fix-3 here.
Please try the following:

1
$ UNISOUND /XESSB /LOG > XESSB.TXT
This will try 660h for the bypass. This is the port ESSCFG uses, then 800h. If you want to try a specific port only add an R value to BLASTER:
$ SET BLASTER=A220 I5 D1 R660
$ UNISOUND /XESSB /LOG > XESSB660.TXT

Don't add /XEA: it switches the ES1868 and ES1869 to each other's setup methods which means you only test 800h here. This is why your /XEA /XESSB run said nothing about 660h.

2
Also please do a wider PnP scan using the arguments JazeFox added a while ago to see if we can see a read port.
$ UNISOUND /F /LOG > F.TXT
$ UNISOUND /FS /LOG > FS.TXT
/F makes the read-port search try more ports.
/FS also starts the search at 20Bh and keeps isolation going through up to 80 missed bits instead of 4. /FS is supposed to help with chipsets where isolation is unreliable.
If the log then shows a card with a bad checksum, /XXF makes UNISOUND accept it anyway.

3
If you're comfortable running the new UNIDUMP, please do so after a power cycle with no sound drivers loaded (proprietary or unisound).
$ UNIDUMP > DUMP.TXT
$ UNIDUMP /ONLY:PNPBIOS,PNP /ISOLATE > ISOLATE.TXT
The first shows whether the PnP BIOS lists the card, and records the machine details we never got.
The second runs UNIDUMP's own isolation and reports any card it finds. That separates "the card doesn't answer isolation" from a problem in UNISOUND's scan.

4
Try ESS's own setup and then dump the card state for comparison.
Run ESSCFG or the card's own setup program, then:
$ UNISOUND /R /LOG > AFTERESS.TXT
$ UNIDUMP /ONLY:PNP > AFTERESS2.TXT
If ESSCFG can set the card up, it's reachable on that machine, and these logs show the settings and control port it used.

Reply 4 of 11, by analog_programmer

User metadata
Rank Oldbie
Rank
Oldbie

Thanks, I'll try these test with your latest fix4-version and report later by PM.

And one note - you can use "he" and "his" when pointing at me, not "them" or "their" 😁

The word Idiot refers to a person with many ideas, especially stupid and harmful ideas.
This world goes south since everything's run by financiers and economists.
This isn't voice chat, yet some people overusing online communications talk and hear voices.

Reply 5 of 11, by Darmok

User metadata
Rank Newbie
Rank
Newbie

@vintagefugg
I’ve tested fix-4, and the IDE port on the OPTi930 is now working. Thank you very much for your quick response to my request. Tomorrow, I plan to add another sound card—based on the OPTi925 chip—to my existing AWE64 and OPTi930 setup. I’m doing this for its built-in wavetable synthesizer and an additional IDE port, which will allow me to connect a couple more hard drives. Unfortunately, this card relies on OPTi-FM synthesis, so the OPTi930 remains essential.

Reply 6 of 11, by GigAHerZ

User metadata
Rank Oldbie
Rank
Oldbie

Thank you for working on this! Keep up the good work!

"640K ought to be enough for anybody." - And i intend to get every last bit out of it even after loading every damn driver!
A little about software engineering: https://byteaether.github.io/

Reply 7 of 11, by vintagefugg

User metadata
Rank Newbie
Rank
Newbie

No worries @GigAHerZ. I'm excited to see people using the fixes.

@Darmok I think you're going to have a rough time getting the OPTi 925 IDE working. I don't think it supports hard drives at all (happy to be proven wrong) but, more importantly, the IDE on it seems like it's probably broken in unisound when compared to the datasheet. I think unisound has set up the 82C925's IDE port wrongly since at least 0.82a: with C170 in BLASTER, the drive's command port ended up at 103h instead of 170h, so the CD-ROM driver probably won't find the drive. I've got a test build on github that changes the behaviour on that chip to copy what the real opti driver (cdsetup.sys) does. I don't have a 925 card and I can't find it in 86box so the test build is completely untested. The test build is attached to a branch specifically for this change and was produced by CI.

Getting the build (needs a GitHub account; the download is only offered when you're logged in):
1. Open https://github.com/dafugg/unisound_continued/ … uns/43666156964
2. Under "Artifacts" at the bottom, click UNISOUND.COM. You get a zip with UNISOUND.COM inside.
3. The banner on the test build will say 0.82a-unofficial-fix-5-rc1 when you run it

I've attached the zip here but it'd be nice if we could eventually shift temporary builds to github artifacts because (1) I'm not sure the license options for file attachments here encompass this unisound situation and (2) I can clean them up on github when they are wrong / stale while I can't edit posts or attachments here to avoid confusing people.

I think by default both the opti driver and unisound will use IRQ 15 for both the 925 and the 930 IDE port so you'll also need to deal with that. The 930 is *fixed* at IRQ15 and port 170h/376H. Even the original cdsetup.sys docs say you can't change it. The 925 is PnP so you will need to use C1E8 in BLASTER to shift the 925 to port 1E8h/3EEh and IRQ 11 OR you can use C168 for 168h/36Eh on IRQ 10. Alternatively you can just use Y to override the IRQ after the port if you don't want the usual pairs.

I think you'll also have to set up the 925 first and then the 930 if you want to use them both with unisound. I think /NOPNP might still find the 925 at F8Ch and set it up instead of the 930.
If you get this working I'd love to update UNISOUND.TXT to include instructions about intermixing PnP and non-PnP cards. At present the docs say that unisound only supports one card if it's non-pnp.

Reply 8 of 11, by InterClaw

User metadata
Rank Newbie
Rank
Newbie

@vintagefugg Could you add an option to reverse stereo in your build? You briefly mentioned that you needed more input from user csal with a Quake problem. Well, I don't have a problem with Quake with my SB16, but there is a known bug in the game Terminal Velocity where the stereo sound is simply reversed, with no way of correcting it. I'd like to flip the channels before running the game and then restoring them afterwards.

I'm thinking something like this.

@UNISOUND.COM /RS1
@TV.EXE
@UNISOUND.COM /RS0

Reply 9 of 11, by vintagefugg

User metadata
Rank Newbie
Rank
Newbie

@interclaw I dug around with AI parsing datasheets and there aren't *any* cards that UNISOUND supports which support arbitrarily swapping stereo channels as far as I can tell.

Instead, I tasked AI to go find the mixer code in the copy of TV I own. It looks like the developer swapped the left and right channel byte order. They write right channel byte first when they should write the left channel byte first in both the stereo and surround modes. I've put up a github repo with a tool that will patch that behaviour to fix stereo channels in terminal velocity v1.2: https://github.com/dafugg/terminal_velocity_stereo_fixer

Reply 10 of 11, by InterClaw

User metadata
Rank Newbie
Rank
Newbie
vintagefugg wrote on Yesterday, 05:31:

@interclaw I dug around with AI parsing datasheets and there aren't *any* cards that UNISOUND supports which support arbitrarily swapping stereo channels as far as I can tell.

Instead, I tasked AI to go find the mixer code in the copy of TV I own. It looks like the developer swapped the left and right channel byte order. They write right channel byte first when they should write the left channel byte first in both the stereo and surround modes. I've put up a github repo with a tool that will patch that behaviour to fix stereo channels in terminal velocity v1.2: https://github.com/dafugg/terminal_velocity_stereo_fixer

Wow, that was quick. I let my DOS machine patch the file and compared the results with a manually updated file in a hex editor in Windows and the results are the same. I had version 1.2.

And the stereo output is now corrected! Much better solution to the problem and therefore a bit off topic here maybe then. I linked to your post in the other thread about the stereo issue itself.

Reply 11 of 11, by Darmok

User metadata
Rank Newbie
Rank
Newbie

Great work, vintagefugg! With fix5-rc1, there’s no need to use cdsetup.sys in config.sys.
If the OPTi95 is set to NOPNP mode via a jumper, unisound detects it but fails to see the OPTi930. When the OPTi925 is in PnP mode, I’ve been using the following sequence of commands:
SET BLASTER=A240 I7 D3 P330 J0 C168 F0
UNISOUND.COM /V90 /VL00 /VF99 /VC10
SET BLASTER=A220 I5 D1 P330 F388 C170
UNISOUND.COM /NOPNP /V90 /VL00 /VF99 /VC10

All resources are allocated correctly, which is confirmed after Win95 boots up. I added the secondary IDE port manually, while the quaternary port was detected automatically. A CD-ROM drive and an LS-120 drive are connected to the secondary port and work without any issues. I couldn't get anything to work on the quaternary port, but as it turns out, that’s due to a fault in my specific OPTi925 card. Consequently, to connect two additional hard drives, I used a simple Creative CT1860 IDE controller, configured as a quaternary port.

I have to stop further attempts to use this OPTi925 card because only the wavetable synthesizer, FM synthesizer, and mixer are functional. The SB Pro and WSS features do not work; the computer freezes as soon as there is an attempt to access their resources. The sndinit.exe utility crashes with a strange error message stating that it cannot open a log file for writing. It appears that the section responsible for handling interrupts is faulty. Perhaps, if I have time, I’ll try to track down the fault, but in all likelihood, the OPTI925 chip itself is defective.