After plugging and unplugging video memory, the card can now display some garbled characters.

What is the problem causing this?

Normal graphics card displays like this
download/file.php?mode=view&id=241649

This seems to be a rather peculiar error the normal text characters seems to be corrupted, but the characters redefined to display the Energy Star logo seem to be not getting redefined but retrieving the data from uncorrupted character ROM?

ekkiller wrote on Yesterday, 02:36:

After plugging and unplugging video memory, the card can now display some garbled characters.

Definitely weird looking.

As Grzyb said, it would help a lot if you could dump the BIOS ROMs and upload them here. It could provide some valuable leads.

If you don't currently have a reader\programmer I'd recommend getting something like this. They are super handy if you are into retro computing.

EDIT: Alternatively, if the card still allows the system to boot aside from the garbled output you could set up a batch file that dumps the video BIOS so that you can type and run the command blind without having to read anything. I'm sure there is some program out there that can do this, though I haven't searched yet myself.

Hi,

While I also recommend to dump the ROM if you have the equipment, I pondered about this a bit more and there are two things that come to mind that can be used to debug the situation:

1. Check if "Video BIOS Shadow" feature of Award BIOS is active, some cards don't like it, so if it is on, turn it off, maybe it will work afterwards.
2. It is possible that this card simply doesn't support the character redefinition feature that AWARD BIOS uses to display the Energy Star logo in text mode, if you have a motherboard with a non AWARD BIOS or with an AWARD BIOS version too old to have this it might be worth it to try it.

As Grzyb said, it would help a lot if you could dump the BIOS ROMs and upload them here. It could provide some valuable leads.
[/quote]
I have several programmers and have dumped the BIOS of the card

MagefromAntares wrote on Yesterday, 06:11:

Hi, […]
Show full quote

Hi,

While I also recommend to dump the ROM if you have the equipment, I pondered about this a bit more and there are two things that come to mind that can be used to debug the situation:

1. Check if "Video BIOS Shadow" feature of Award BIOS is active, some cards don't like it, so if it is on, turn it off, maybe it will work afterwards.
2. It is possible that this card simply doesn't support the character redefinition feature that AWARD BIOS uses to display the Energy Star logo in text mode, if you have a motherboard with a non AWARD BIOS or with an AWARD BIOS version too old to have this it might be worth it to try it.

Thank you.
Indeed, this card is not compatible with AWRAD BIOS.
It displays normally on the AMI BIOS motherboard.

The graphics card is plugged into the 286 (AMI BIOS) motherboard and displays normally.

The graphics card is plugged into the 386 (AMI BIOS) motherboard and displays normally.

The graphics card is installed on a 386 motherboard(AWARD BIOS) , and there are issues with the display, but the characters are still legible in the CMOS setup interface.

ekkiller wrote on Yesterday, 12:37:

The graphics card is installed on a 386 motherboard(AWARD BIOS) , and there are issues with the display, but the characters are still legible in the CMOS setup interface.

It might be because the Setup Interface doesn't use redefined characters, if you wish to use that motherboard try setting the VIDEO in "Standard CMOS Setup" to "CGA 40" or "CGA 80" as CGA didn't had redefinable characters. This might mess up some programs detection of the video card however, they might believe the BIOS that it is only a CGA 😁.

No need, I have some 386 motherboards with AMI BIOS. I reckon there might be some compatibility issues with this graphics card for DOS games.

Googling for "CVGA/24" finds some Chinese stuff - looks like that card was used in Great Wall 386 microcomputers.
I guess it would be a good idea to try Baidu...

Is it possible that the card is intended for use on systems that use Chinese, Japanese or other non-latin based languages? I believe those tend to require higher resolution text\character modes so that characters are legible. If the character set this card is using is meant for an entirely different language and text display resolution it might result in these illegible broken up characters being displayed.

I just ask because some of the garbled characters look very intentional and look like parts of larger and more detailed characters.

The two HY62256ALP chips could be used for loading different character sets to the card from software or something like that... just spitballing.

EDIT: A brief search hasn't turned up any sites that feature similar ISA cards, but IBM PS/2 MCA cards intended for the Japanese market tended to have a lot of additional complexity. Some even had daughterboards for adding character ROMs.

It might be worth testing DOS/V to see if it displays properly on this card?

EDIT2: Ah! I found something possibly related... AIC (AIC HK) seems to be Arnos Instruments and Computer Systems. The AR810 chip is found on some 386 boards, sometimes labeled SIRUS AR810, manufactured by Toshiba and with very similar printing to the AR310 (minus the Toshiba branding).

https://theretroweb.com/motherboards/s/arnos- … -a386-25-33-40c

https://theretroweb.com/motherboards/s/arnos- … own-ar810-3-486

https://theretroweb.com/motherboards?manufact … 45&showImages=1

So... it seems likely that the AR310 is some kind of glue logic that facilitates the use of the two different sets of memory and the SRAM, rather than acting as a graphics accelerator.

I am still leaning toward this being language\character related rather than graphics, but it is all just a guess at this point.

Matth79 wrote on 2026-04-16, 23:40:

I'd guess the mystery chip would be RAMDAC, or possibly clock gen

Nope, the UMC chip above the EPROM is the RAMDAC, a pretty generic one seen on many ISA VGA cards (I have the same on a Trident card). And even in the early 1990s, no way does a clock gen need to be that big.

Ozzuneoj wrote on Yesterday, 17:45:

Is it possible that the card is intended for use on systems that use Chinese, Japanese or other non-latin based languages? I bel […]
Show full quote

Is it possible that the card is intended for use on systems that use Chinese, Japanese or other non-latin based languages? I believe those tend to require higher resolution text\character modes so that characters are legible. If the character set this card is using is meant for an entirely different language and text display resolution it might result in these illegible broken up characters being displayed.

I just ask because some of the garbled characters look very intentional and look like parts of larger and more detailed characters.

The two HY62256ALP chips could be used for loading different character sets to the card from software or something like that... just spitballing.

EDIT: A brief search hasn't turned up any sites that feature similar ISA cards, but IBM PS/2 MCA cards intended for the Japanese market tended to have a lot of additional complexity. Some even had daughterboards for adding character ROMs.

It might be worth testing DOS/V to see if it displays properly on this card?

EDIT2: Ah! I found something possibly related... AIC (AIC HK) seems to be Arnos Instruments and Computer Systems. The AR810 chip is found on some 386 boards, sometimes labeled SIRUS AR810, manufactured by Toshiba and with very similar printing to the AR310 (minus the Toshiba branding).

[...]

So... it seems likely that the AR310 is some kind of glue logic that facilitates the use of the two different sets of memory and the SRAM, rather than acting as a graphics accelerator.

I am still leaning toward this being language\character related rather than graphics, but it is all just a guess at this point.

The language/character thing is very likely what's messing up the Award POST screen, and agree that it would be very interesting to see what it does with DOS/V.

But I find the idea that a different character set would need different RAM, including an SRAM cache, very hard to believe. For different character sets I'd expect multiple ROM banks, not multiple RAM banks, and definitely not fast, expensive SRAM cache. That kind of architecture really screams processor of some kind or other. Which implies it's a very different chip to the AR810. Agree that it looks like you've found the manufacturer though.

I'd second the idea suggested three weeks ago to try the WIn3.1 8514A driver and see what that does.

dionb wrote on Yesterday, 21:58:

The language/character thing is very likely what's messing up the Award POST screen, and agree that it would be very interesting to see what it does with DOS/V.

But I find the idea that a different character set would need different RAM, including an SRAM cache, very hard to believe. For different character sets I'd expect multiple ROM banks, not multiple RAM banks, and definitely not fast, expensive SRAM cache. That kind of architecture really screams processor of some kind or other. Which implies it's a very different chip to the AR810. Agree that it looks like you've found the manufacturer though.

I'd second the idea suggested three weeks ago to try the WIn3.1 8514A driver and see what that does.

Yeah, I was just spitballing ideas. When you put it that way I definitely agree. Having two sets of SRAM (16KB and 64KB) and two sets of DRAM (512KB and 768KB) seems like massive overkill for something that isn't doing any acceleration. There has to be more to it than that.

Also, since you mention it possibly being an 8514/A compatible card, I stumbled upon this thread. It mentions this EIZO "extended" 8514/A compatible card having 64KB SRAM for a hardware mouse cursor:
An EIZO branded extended 8514/A compatible CAD/GUI accelerator: The AA41
There is also yet another 8KB SRAM chip on that EIZO card (TC5588J) that isn't mentioned in that thread as far as I can tell. It's possible that the 16KB SRAM on the AIC card performs a similar function, whatever it may be.

If DOS is higher res, EG 640x400 instead of 320x200, it's most likely for DOS-based word processing IMO.

Definitely some processing of characters. If it were for accelerating windows it would have a very standard DOS, no character processing.

Storing characters in RAM space instead of ROM allows a program to load different character sets depending on the need.