SirNickity wrote:

Maybe the TAG RAM size?

It has two 8KB TAG chips, which is correct for the amount of cache according to the TH99 page. Having 8KB tag + 64KB cache or 16KB TAG + 128KB cache didn't seem to make any difference with regard to the "Megabyte #4 isn't being cached" message.

Is there any general rule with 386 cacheable RAM size? From what I can gather, every chipset is completely different...

Could it be a bug in the chipset or motherboard?

I presume you have remap memory enabled, so that will take the memory between 640K and 1024K that isn't being used for shadowing and maps it above rest of the memory at address 4meg. Maybe because of the remap, the caching doesn't work.

canthearu wrote:

I presume you have remap memory enabled, so that will take the memory between 640K and 1024K that isn't being used for shadowing and maps it above rest of the memory at address 4meg. Maybe because of the remap, the caching doesn't work.

The DTK Datatech BIOS on the system has almost no options whatsoever. The only "chipset options" listed are for Video and BIOS shadowing. Aside from that all I have are options for drives, boot order and memory settings for base and extended memory size (640K and 3328... 128K for shadowing is shown at startup but I have no options to change it).

At this point I'm going to overlook the odd cacheable memory issue because I have a much more pressing problem...

I can't get any 8bit soundcards to work in this system! There's hardly anything in this system so I don't think it's caused by a resource conflict because an SB Pro2 CT1600 works fine with the same settings. A have a Diamond Speedstar VGA Tseng ET4000AX and a SIIG Enhanced IDE card with the Parallel port set to IRQ7. I can boot from the hard drive to DOS 6.22, I can use a Parallel to CF reader for external storage and everything works fine except that I get no sound in any games when using any 8bit cards. I've tried my CT1350B, MediaVision Thunderboard and an Aztech SoundGalaxy BXII and they all give me random buzzing and static. All are set to IRQ5 and address 220, just like the CT1600.

I thought maybe it was related to some of the cards requiring -5v (I'm using a modern ATX PSU with one of the adapters that adds -5v), but using a tester card it shows -5v as being there and I can use my DMM on the -5v test points and it is showing about -4.85v. Last week when I first noticed the issue I tried using an old AT PSU and had the same problem, so it really doesn't seem to be -5v related. Also it doesn't look like the -5v pin on the Thunderboard is connected to anything so that definitely rules that out for that card. For what its worth, I tested the CT1350B last year and it worked fine, and I tested the BXII about two years ago and it also worked fine (if a bit noisy). The Thunderboard hasn't been tested until now however. I'll try testing the cards in another system just to be sure.

In the mean time... has anyone ever seen an issue like this?