Here's a picture of open cubic player.

The Video RAM issue is strange. Those cards do not use linear framebuffer stuff that can depend on hardware design or bus type, so the OLB ET4000AX should look exactly like an ISA ET4000AX to the BIOS, just faster. Thus I don't understand why the RAM size detection fails with the newer BIOS. The Open Cubic Player clearly is a BIOS issue: OCP tries to switch into 50-line mode by loading an 8x8 font, but the BIOS you use does not resize the character cell to 8x8 after loading that font, so you stay in 80x25 mode with the top half of the characters being from an 8x8 font, and the bottom half the characters being the bottom half of 8x16 characters.

As you said, you had issues with the video RAM - maybe you still have video RAM issues and the new BIOS falls back to 256K because it detects that the full 1M does not work correctly.

For VESA support, I recommend you use the original TSENG VESA driver, which should be obtainable at various BBS mirror sites. Look for TLIVESA.COM (TLI = Tseng Labs Incorporated).

Like the Cirrus 542x chips, the ET4000AX only has a 16-bit CPU bus interface, so this will limit the throughput.

There are two variants of the Opti Local Bus: In one variant, the OLB slots are fully ISA compatible, but lack in local bus support, in the other variant, the 486 local bus is quite completely mirrored to the slot, but some DMA and IRQ channels are no longer available. Only the "full local bus/limit ISA" variant of OLB supports the "BS16" signal that signals to the processor that the request was handled by a 16-bit peripheral and if the request crossed a 16-bit boundary (a full 32-bit request does so), the 486 needs to repeat the "second half". You got the OLB ET4000 variant that splits 32-bit cycles on the card, and does not need BS16 support, so it should be universally compatible with all OLB implementations. I got a simpler one that requires BS16 support on the mainbaord, but at the same time, I got an OLB mainboard that only has the wrong slot variant...

mkarcher wrote on 2025-12-16, 00:48:

As you said, you had issues with the video RAM - maybe you still have video RAM issues and the new BIOS falls back to 256K because it detects that the full 1M does not work correctly.

I tried BIOSes that had the same date stamp as well, no difference. But I'll try to swap all the RAM and see if that makes a difference.

mkarcher wrote on 2025-12-16, 00:48:

For VESA support, I recommend you use the original TSENG VESA driver, which should be obtainable at various BBS mirror sites. Look for TLIVESA.COM (TLI = Tseng Labs Incorporated).

Yea I did find v1.2 and it works. Not a lot of modes, but at least enough to get OCP's hi-res spectrum analyser to work. However: While Scitech/UniVBE gives me 1024x768@60Hz, the TLIVESA driver only gives me interlaced output, so unless there's a way to solve that, I'm stuck with UNIVBE (and 20KB less conventional RAM, as it seems to be impossible to load it high)... :D

mkarcher wrote on 2025-12-16, 00:48:

Like the Cirrus 542x chips, the ET4000AX only has a 16-bit CPU bus interface, so this will limit the throughput.

Yeah, but it sure seems to make the most of those 16 bits. I really need to figure out why this motherboard doesn't work at much more than 40MHz FSB; the chipset definitely supports 50MHz so I don't know what's keeping it back..

mkarcher wrote on 2025-12-16, 00:48:

There are two variants of the Opti Local Bus: In one variant, the OLB slots are fully ISA compatible, but lack in local bus support, in the other variant, the 486 local bus is quite completely mirrored to the slot, but some DMA and IRQ channels are no longer available. Only the "full local bus/limit ISA" variant of OLB supports the "BS16" signal that signals to the processor that the request was handled by a 16-bit peripheral and if the request crossed a 16-bit boundary (a full 32-bit request does so), the 486 needs to repeat the "second half". You got the OLB ET4000 variant that splits 32-bit cycles on the card, and does not need BS16 support, so it should be universally compatible with all OLB implementations. I got a simpler one that requires BS16 support on the mainbaord, but at the same time, I got an OLB mainboard that only has the wrong slot variant...

This might explain why an earlier[1] (or later?) variant of the board has a 16/32-bit LB jumper right next to the slot - mine does not[2].

[1] https://theretroweb.com/motherboards/s/austin … omputer-486-s14
[2] https://theretroweb.com/motherboards/s/austin … 486-s14-rev-1-1

ltning wrote on 2025-12-17, 00:19:

While Scitech/UniVBE gives me 1024x768@60Hz, the TLIVESA driver only gives me interlaced output, so unless there's a way to solve that, I'm stuck with UNIVBE (and 20KB less conventional RAM, as it seems to be impossible to load it high)... 😁

TLIVESA just gives you the referesh rate in VESA modes that you also get from the Video BIOS using classic Tseng mode numbers. You can call DMODE with a suitable command line to change the refresh rate of SVGA modes. The printed documentation just refers you to the online help of DMODE. The Tseng Labs DMODE utility was distributed on the driver disk with that card, and a hopefully suitable version of that tool can be found at http://ftp.oldskool.org/pub/drivers/TSENG/ET4 … 00/dos/UTILITY/

ltning wrote on 2025-12-17, 00:19:

I really need to figure out why this motherboard doesn't work at much more than 40MHz FSB; the chipset definitely supports 50MHz so I don't know what's keeping it back..

40MHz is pretty good. Most VLB cards only work up to 40MHz. They also have a >33MHz jumper and additional wait state jumper.

MikeSG wrote on 2025-12-17, 12:28:

ltning wrote on 2025-12-17, 00:19:

I really need to figure out why this motherboard doesn't work at much more than 40MHz FSB; the chipset definitely supports 50MHz so I don't know what's keeping it back..

40MHz is pretty good. Most VLB cards only work up to 40MHz. They also have a >33MHz jumper and additional wait state jumper.

Ah but this is regardless of the actual VGA card. Even with plain ISA it won't work. Either the diag board shows -- --, -- 01, or it stops or hangs at some point in the early boot process.

mkarcher wrote on 2025-12-17, 06:45:

ltning wrote on 2025-12-17, 00:19:

While Scitech/UniVBE gives me 1024x768@60Hz, the TLIVESA driver only gives me interlaced output, so unless there's a way to solve that, I'm stuck with UNIVBE (and 20KB less conventional RAM, as it seems to be impossible to load it high)... :D

TLIVESA just gives you the referesh rate in VESA modes that you also get from the Video BIOS using classic Tseng mode numbers. You can call DMODE with a suitable command line to change the refresh rate of SVGA modes. The printed documentation just refers you to the online help of DMODE. The Tseng Labs DMODE utility was distributed on the driver disk with that card, and a hopefully suitable version of that tool can be found at http://ftp.oldskool.org/pub/drivers/TSENG/ET4 … 00/dos/UTILITY/

But DMODE is only for the W32 series, I think? At least it has never worked on the AX chip - it exits saying it needs a W32 (confirmed on this card). Perhaps there's a similar tool for the AX series, or an older version?

mkarcher wrote on 2025-12-16, 00:48:

There are two variants of the Opti Local Bus: In one variant, the OLB slots are fully ISA compatible, but lack in local bus support, in the other variant, the 486 local bus is quite completely mirrored to the slot, but some DMA and IRQ channels are no longer available. Only the "full local bus/limit ISA" variant of OLB supports the "BS16" signal that signals to the processor that the request was handled by a 16-bit peripheral and if the request crossed a 16-bit boundary (a full 32-bit request does so), the 486 needs to repeat the "second half". You got the OLB ET4000 variant that splits 32-bit cycles on the card, and does not need BS16 support, so it should be universally compatible with all OLB implementations. I got a simpler one that requires BS16 support on the mainbaord, but at the same time, I got an OLB mainboard that only has the wrong slot variant...

Any way or idea on how to identify which variant the slots & cards are ? Do they all use the same connector as an EISA slot?

I have this
https://theretroweb.com/expansioncards/s/unkn … wn-et4000ax-olb
and this
https://theretroweb.com/motherboards/s/pcchips-m423

but haven't gotten round to connecting it yet and now I wondering if they might not be compatible ?

weedeewee wrote on 2025-12-21, 12:00:

I have this
https://theretroweb.com/expansioncards/s/unkn … wn-et4000ax-olb

Hey, would you have a chance to dump the VGA BIOS of your card? Just the one chip closest to the bus connector; it's a 27C256 (probably M27C256B@DIP28 in minipro) if you have an (E)EPROM reader..

I just uploaded mine to TRW (should appear as soon as someone has time to add it to the page), but was hoping someone has a different one because this BIOS is ... buggy as hell. :)

ltning wrote on 2025-12-21, 12:13:

weedeewee wrote on 2025-12-21, 12:00:

I have this
https://theretroweb.com/expansioncards/s/unkn … wn-et4000ax-olb

Hey, would you have a chance to dump the VGA BIOS of your card? Just the one chip closest to the bus connector; it's a 27C256 (probably M27C256B@DIP28 in minipro) if you have an (E)EPROM reader..

I just uploaded mine to TRW (should appear as soon as someone has time to add it to the page), but was hoping someone has a different one because this BIOS is ... buggy as hell. 😀

it identified as an NM27C256@DIP28 😉
anyway, here ya go.

Copyright(c)1988 Tseng Laboratories, Inc. 07/01/91 V2.00X
OEM_T07/05/91
Enhanced VGA BIOS MODEL:2148S Date : 05/30/92

mmmh... I already posted it before Re: ET4000 and Sierra Hi-color DAC weird behaviour?

anyway good thing I read it out again, because apparently I would have set myself up for some headache since I placed the eprom back the wrong way ! DOH !

edit:
just checked your post on TRW.
It's the same bios .

weedeewee wrote on 2025-12-21, 12:45:

just checked your post on TRW.
It's the same bios .

Damn, then I can't wait for you to report how your card works .. If you have the same symptoms I have, at least I know my card is no more broken than yours ;)

ltning wrote on 2025-12-21, 13:02:

weedeewee wrote on 2025-12-21, 12:45:

just checked your post on TRW.
It's the same bios .

Damn, then I can't wait for you to report how your card works .. If you have the same symptoms I have, at least I know my card is no more broken than yours 😉

Which mainboard are you using ?

weedeewee wrote on 2025-12-21, 13:14:

ltning wrote on 2025-12-21, 13:02:

Damn, then I can't wait for you to report how your card works .. If you have the same symptoms I have, at least I know my card is no more broken than yours ;)

Which mainboard are you using ?

This one:
https://theretroweb.com/motherboards/s/austin … 486-s14-rev-1-1

weedeewee wrote on 2025-12-21, 12:00:

Any way or idea on how to identify which variant the slots & cards are ? Do they all use the same connector as an EISA slot?

Yes, both OLB variants use the EISA connector. For the mainboard, identifying is easy if you can test for continuity. Look at this thread for more more information: Re: OPTi / Orchid Local Bus variants if you are into technical details, the patent I linked is worth a read as well.

weedeewee wrote on 2025-12-21, 12:00:

but haven't gotten round to connecting it yet and now I wondering if they might not be compatible ?

That card is also mentioned further down in the thread I linked, at it is what I would call a "universal" card, that will work in both kinds of OLB slots. The most notable feature on your card is the presence of 2 8-bit latch chips (the 74HCT374 chips), which are not present on my card. Latches are required when you translate 16-bit ET4000 reads into a 32-bit 486 read, because the 486 processor samples all 32 bits at once, while the ET4000 can only output 16-bit at once. So your card reads the first 16 bits into the 74HCT374 chips and saves the bits in those chips, then reads the second 16 bits directly to 16 bits of the local bus, while outputting the 16 bits from the latches on the other half of the local bus, and then tells the 486 chip that the requested data is ready. As my card has no latches, it can't do that.

Instead, my card reads only the first 16 bits, and then activates pin D4 on the (E)ISA connector intending to signal: "Hey, wait a second, I only serviced the first 16 bits of your request, please issue another requerst for the remaining 16 bits". On my system, activating D4 triggers IRQ11 and does not cause a second bus request.

mkarcher wrote on 2025-12-21, 19:29:

Yes, both OLB variants use the EISA connector. For the mainboard, identifying is easy if you can test for continuity. Look at th […]
Show full quote

weedeewee wrote on 2025-12-21, 12:00:

Any way or idea on how to identify which variant the slots & cards are ? Do they all use the same connector as an EISA slot?

Yes, both OLB variants use the EISA connector. For the mainboard, identifying is easy if you can test for continuity. Look at this thread for more more information: Re: OPTi / Orchid Local Bus variants if you are into technical details, the patent I linked is worth a read as well.

weedeewee wrote on 2025-12-21, 12:00:

but haven't gotten round to connecting it yet and now I wondering if they might not be compatible ?

That card is also mentioned further down in the thread I linked, at it is what I would call a "universal" card, that will work in both kinds of OLB slots. The most notable feature on your card is the presence of 2 8-bit latch chips (the 74HCT374 chips), which are not present on my card. Latches are required when you translate 16-bit ET4000 reads into a 32-bit 486 read, because the 486 processor samples all 32 bits at once, while the ET4000 can only output 16-bit at once. So your card reads the first 16 bits into the 74HCT374 chips and saves the bits in those chips, then reads the second 16 bits directly to 16 bits of the local bus, while outputting the 16 bits from the latches on the other half of the local bus, and then tells the 486 chip that the requested data is ready. As my card has no latches, it can't do that.

Instead, my card reads only the first 16 bits, and then activates pin D4 on the (E)ISA connector intending to signal: "Hey, wait a second, I only serviced the first 16 bits of your request, please issue another requerst for the remaining 16 bits". On my system, activating D4 triggers IRQ11 and does not cause a second bus request.

I already saw that thread, even commented in it.
Good to know that the card will likely work (if mainboard & card are still functional, since I haven't tested either yet).

@ltning There's also some photos of our card in that thread, and maybe I should find a way to digitize the manual that came with it 🤔.

Can the BIOS extension of this card be of use to you? I know nothing of the status of the card -- I do not have a motherboard to test it with, I made a reasearch once upon a time and concluded it could cause damage if plugged in to the PISA of ABIT AB-PB4.

georgel wrote on 2025-12-21, 21:47:

Can the BIOS extension of this card be of use to you? I know nothing of the status of the card -- I do not have a motherboard to test it with, I made a reasearch once upon a time and concluded it could cause damage if plugged in to the PISA of ABIT AB-PB4.

The attachment IMG_5370.JPG is no longer available

The attachment IMG_5371.JPG is no longer available

I'm gonna say, Yes it can be of use.
I also do not immediatly see this card on theretroweb.
Would you mind adding it?

@mkarcher I just noticed that your OLB vga card is also not listed on TRW... 😀

You may upload the pictures wherever you like. Here is the BIOS extension of the pictured card.

Oh-la-la, I have much to play with tomorrow, it seems :-)