Remove 8 MB of RAM. I own a 486 by EPSON with the same graphics chip. Windows 95 is unable to use more than 8-bit with this graphics chip when you have more than 16 MB RAM installed. No idea why.

Not really an option, sorry. I can do 24MB or 8MB, nothing in between. It's got 8MB onboard and the module is for 16MB. Oh well. 256 colors it is I guess. I wonder if Windows 3.11 has the same issue?

Are you trying to set 16-bit colour at 800x600? That might not work on a 1 MB card (pretty sure it didn't on my Speedstar Pro VLB 1 MB). Did you try 640x480?

800x600 @ 16 bit needs 937.5 KB of video ram. So 1 MB is enough.

As I said, I know this problem and it is caused by more than 16 MB RAM. Should jaZz_KCS read this, he can confirm the issue. He found it as well.

derSammler wrote:

800x600 @ 16 bit needs 937.5 KB of video ram. So 1 MB is enough.

Yea, I couldn't do the maths...
No seriously. Just because the framebuffer theoretically fits in VRAM doesn't mean the mode will actually work under Windows.
Again, if you have only tried it in 800x600, try it in 640x480.

The display of the T2130CS can only do 640x480 anyway, at least the models I know of. Unlikely that he tried setting it to 800x600.

though I have tried several monitor drivers and currently am using the Laptop Display Panel - 800x600 driver.

are you sure the vga chipset in this machine comes with a 16/24bit DAC? it could be a case of a 8bit DAC and thus 16/24bit modes are not available in it's BIOS
I have a couple 1MB isa cards with only a 8bit/256color dac

@Scali:

Feel free to continue ignoring that using Chips 65545 with Windows 95 and more than 16 MB RAM will not give you a 16-bit screen mode due to whatever bug. Makes no sense to discuss this any further.

derSammler wrote:

Feel free to continue ignoring that using Chips 65545 with Windows 95 and more than 16 MB RAM will not give you a 16-bit screen mode due to whatever bug. Makes no sense to discuss this any further.

Yea whatever. You're the one arguing against any other advice, and pushing this as the only possible answer (including ignoring the OP stating using an 800x600 driver).
I never said that the > 16 MB thing couldn't be the issue. I just said it may not necessarily be the only issue.

Not the first time you're being this obtuse in a discussion either. And it wouldn't be the first time you'd be wrong either.

I tried the 800x600 driver just as something else to test. I kept it in a 640x480 resolution since the panel's not good for anything past that -- it'll actually set to it but you have to mouse around to get to all of the screen, and I find that utterly useless, so I kept it at 640x480. It didn't change anything.

Another piece of the puzzle. I tried Win3.11 on here as well, since I didn't have much better to do, and got another setback -- the driver, officially provided by Toshiba from their website itself, does not work. I can try to set a 32K or 64K color mode (at 640x480 of course) and it bombs out saying the VBIOS doesn't support this mode. 256 colors... works fine, although their choice of default color scheme and wait cursor are questionable. Very odd. I wonder if I can update the VBIOS? And I suppose there's no use to running Windows at all on there, if I can't get higher than 256 colors out of it... Maybe I'll keep 3.11 around for the giggles, maybe not... I could very well just install Word 6.0 for DOS and get about as much utility out of that as what I had planned for Windows. 1995 is a bit late for anyone to be selling machines artificially hampered like this, so I guess since Win3.11 has lower memory requirements than 95 I'll pull the upgrade module out just to see if it'll work without it. I'm not expecting it to, but if it does, oh boy am I in for a world of pain trying to find an 8MB module to replace this 16...

EDIT: One quick test later and I can confirm, it still doesn't work. I guess it's down to my VBIOS mysteriously not supporting 32/64k color modes, despite it having plenty of video RAM to do it with. Why does it have such a large amount of video RAM if it can't push a higher color depth when even the screen is noted as supporting it? Contemporary machines running accelerators in the same family, also from Toshiba, although these came with Pentiums, do support higher bit depths.

I can confirm the strange quirk regarding C&T65545 chipsets and more than 16MB of memory resulting in the official drivers not working above 8 bits.... As well as other acceleration related weeaboos happening. I have experienced similar behaviour on my cash register 486 that has the same chipset. When installing more than 16MB of memory a certain range of features of the C&T65545 are not working anymore resulting in the following errors (that I can remember):

- Upon booting W95, it will say the GPU driver is incompatible and you have to choose anew. Even when selecting the same one as before, the higher bitmodes as well as resolution seem to be unavailable / crash upon try.
- Upon booting W3.1, it will crash with the C&T65545 driver selected.
- DOS VESA modes of various kinds cannot be intitialized and games that were working before (for example with SciTech or UniVESA or alike) do crash/exit. Basically VESA support is botched in a big way.
There were others that I cannot remember

Could it be an issue with the linear frame buffer? With SciTech there are options to check where the framebuffer is located and it can usually be moved.

Sedrosken wrote:

Not really an option, sorry. I can do 24MB or 8MB, nothing in between.

Have you tried with 8MB yet?
Should be plenty for running Win3.1, and acceptable for Win95.
Good enough for troubleshooting.

Tried 8MB on both. Doesn't fix the issue. I might look into Scitech to see if there's anything I can do, but at this point I mostly consider the matter solved if not exactly satisfactorily. Given that Device Manager warns that the IO addresses are in non-standard locations, Scitech may be able to do something. What's a good version for Windows 95 and a 486?

Sedrosken wrote:

Tried 8MB on both. Doesn't fix the issue. I might look into Scitech to see if there's anything I can do, but at this point I mostly consider the matter solved if not exactly satisfactorily.

The theory that the card only has an 8-bit DAC seems likely now.
If that theory is true, then the card will only support palette-based modes, with 256 colours max. And it could be that the BIOS actively blocks any other modes, because they would not work.

Sedrosken wrote:

Given that Device Manager warns that the IO addresses are in non-standard locations, Scitech may be able to do something. What's a good version for Windows 95 and a 486?

I'm not sure if that would work. Scitech is basically a TSR that adds/corrects the VESA BIOS Extensions. But Windows 95 does not use these. Windows drivers interface directly with the hardware (in which case it doesn't matter what the addresses are, or if the card is even remotely VGA-compatible or anything). At best you could use a generic VESA driver, but I'm not sure if Windows 95 came with one. Performance will likely be poor if there is a VESA driver, but there is a small chance that it would enable the hi-color mode, if Scitech can make it work at all (you could test Scitech under DOS first I suppose, with some image viewer or such, that supports VESA)

I have checked the datasheet and the 65540/65545 controllers have integrated truecolor ramdac but via registers they can be limited to support old flatpanels and their color depth - the datasheet states some dstn ones that have 256k color limitation and newer ones that support the full 16,7mil range.
Maybe the panel used in this system only supports 8bit and thus the controller is locked to that for the internal flatpanel. If there is an external vga port there is a good chance it will support higher color depths.

who knows what's really going on.... 😵