Soundblaster AWE32 (CT3990) and resource conflicts...

[TechGeek]

Is it just me or does the Soundblaster AWE32 (model CT3990) require a lot of resources to be able to properly run?

[eccerr0r]

define "a lot" ... with video cards mapping all video memory to address space, that could be said to be a lot more than any sound card...

[TechGeek]

I/O port ranges and configuration resources. On the 1993 Gateway 2000 80486 system, the card only works reliably in Slot 8, and it only seems to work well with the Adaptec AVA-1515 SCSI controller. Trying my UMC UM9008F-based or my Winbond W89C904AXF-based network card has only intermittent success, usually resulting in the system hanging when AWEUTIL.COM is executed to initialize the CT3990, or AWEUTIL returning 'ERR012: AWE32 initialization failed'.

Other strange things that I've noticed are:
Successful initialization on the first cold boot, but hanging on a warm boot or after a second cold boot.
Initialization failing, but when starting WfW3.11, the system immediately hard-hangs when any audio is played, rather than simply throwing an error stating that the soundcard isn't present.

Not sure if it's just a lack of configuration resources, weird interoperability quirks, or what it could be. I've went into the Intel PnP Configuration Utility, disabled the IDE and gameport controllers, and set the baseport to 0x280. That seems to make it play nice with the SCSI controller, which it didn't do previously, and it won't play nice with either network card, with or without the SCSI controller installed.

[stj]

make sure the irq isnt shared with anything.

[eccerr0r]

Same to DMA channels, been a while but thought SB boards used a DMA channel along with IRQ.

but yes there have been boards that have really inflexible IO port/IRQ/DMA/memory address switching that getting them to coexist with other boards have been close to or totally futile. Course it would be nice if all boards could be set to any arbitrary settings but that would increase the cost of the board ... and we know the beancounters win.

[stj]

i seem to remember the irq can be 5 or 7 and that clashes with the parallel port.

[eccerr0r]

LPT1 uses IRQ7. LPT2 typically uses IRQ5 but most people only have one parallel point and IRQ5 is probably the most common SB16 interrupt line.

[lti]

There was a "retro computer" trend a couple years ago of putting the sound card on IRQ 7 instead of 5. I don't know if there was some game that was hardcoded to IRQ 7. IRQ 5 works for me.

My Gateway Solo 2500 defaults to IRQ 11 if the CMOS battery dies (along with the more annoying problem of having to run an IDE auto-detect function in the BIOS - it defaults to "not installed" for both IDE drives). It doesn't really matter because the oddball NeoMagic sound chip has no DOS compatibility, even though it's based on a Yamaha OPL3-SAx (it even has the Yamaha control panel in Windows).

[stj]

if you look at your motherboard manual, you will see that many boards dont have the same interupts available on each slot.
so back in the isa days you sometimes had to swap cards around.

[TechGeek]

Originally Posted by stj

if you look at your motherboard manual, you will see that many boards dont have the same interupts available on each slot.
so back in the isa days you sometimes had to swap cards around.

That's the problem. There is no motherboard manual for the Micronics 09-00081/09-00135 boards. Chipset documentation is MIA as well, as that too is a Micronics custom chipset.

[stj]

https://theretroweb.com/motherboards...0081-xx-09-001
more than nothing

[TechGeek]

That's just some basic jumper configuration. I haven't found anything else regarding documentation.

[RJARRRPCGP]

Originally Posted by lti

There was a "retro computer" trend a couple years ago of putting the sound card on IRQ 7 instead of 5. I don't know if there was some game that was hardcoded to IRQ 7. IRQ 5 works for me.

My Gateway Solo 2500 defaults to IRQ 11 if the CMOS battery dies (along with the more annoying problem of having to run an IDE auto-detect function in the BIOS - it defaults to "not installed" for both IDE drives). It doesn't really matter because the oddball NeoMagic sound chip has no DOS compatibility, even though it's based on a Yamaha OPL3-SAx (it even has the Yamaha control panel in Windows).

If that's an early-2000s PC or a very-late-1990s PC, then it going to IRQ 11 can be totally bad news. IRQ 11 is for the video, at least on later PCs.

[Hondaman]

My first computer (386DX-40, 4 megs RAM, April 1992) did not have a sound card or speakers, or CD-ROM (although it had SIXTEEN 30-pin SIMM sockets which I think was rare. 16 SIMMS * 4 megs = 64 MB max). When I added a sound card and CD-ROM in 1994, as many did that year, I got the Sound Blaster 32 (not the AWE32). My motherboard featured "decoupled refresh" in the BIOS. When the motherboard was refreshing 8 of the SIMM sockets, it could read or write to/from the other 8 (I think that's what it was doing).

When "decoupled refresh" was turned on in the BIOS, the first time you asked the sound card to make any sound, some weird noise came out of the speakers and would not stop. The reboot button or pulling the power cord were the only ways to make it stop.

I don't know if this means it took lots of resources. I do know that I spent $500 for four 4-meg 30-pin SIMMS (16 MB total plus whatever else I already had) but they didn't work and I had to send them back. They were "Packard Bell Fake Parity Chips" that failed on many motherboards, including mine.

[TechGeek]

I've pulled the CT3990 out of the Gateway 2000 system and will be shelving it until I can get the ball rolling on an upcoming Socket 5 build, which will be a while down the road.

[lti]

Originally Posted by RJARRRPCGP

If that's an early-2000s PC or a very-late-1990s PC, then it going to IRQ 11 can be totally bad news. IRQ 11 is for the video, at least on later PCs.

It's from 1999. It has one of those NeoMagic graphics chips that everyone hates, but I haven't seen the compatibility problems that people talk about.