That third EPROM is Translation ROM - provides register-level CGA and Hercules modes at normal VGA frequencies.

I don't have such a ROM from Genoa, but these are from other cards based on ET4000AX:
- Orchid: https://minuszerodegrees.net/rom/bin/Orchid/O … _nmc27c256b.BIN
- Targa: Re: Trident SVM hercules on VLB TVGA9200CXr

I haven't tried to run anything CGA or hercules. That explains why it seems to work without the chip. What would you say the odds are on either of those roms working without issues? I'm trying to determine if I roll the dice ordering a different video card or spend money on an EPROM programmer.

Translation ROM is described in the Data Book from Tseng Labs - so I don't think it's specific to a particular card model.
Also, an EPROM programmer may come in handy for other things as well...

BTW: see also https://theretroweb.com/expansioncards/s/tseng-labs-tli-9112 -> BIOS
That XL ROM is identical with the one from Orchid.

Ordered a T48 programmer and some chips. Will report back if it works. 🤞

Just got my programmer... Can someone tell me what chip is under the missing one in the first post? Is that the bios? It's socketed so I want to dump it to Retro Web.

ynk1121 wrote on 2026-09-16, 00:43:

Just got my programmer... Can someone tell me what chip is under the missing one in the first post? Is that the bios? It's socketed so I want to dump it to Retro Web.

It's a "27C256" - 32KB EPROM.

The missing upper EPROM chip is meant for the BIOS also, mkarcher once mentioned in one of his posts that ET4000 cards can access video BIOS using a 16 bit data bus, but for that to happen, a card must have two eproms with the BIOS image split into even and odd bytes. That Genoa card of yours is configured to access BIOS using only 8 bit data bus and the BIOS image is whole. I have not yet looked into this 16 bit BIOS data bus and image splitting idea, so I'm not 100% sure if ET4000 chip can handle it. The only ET4000 card that I own is using two BIOS chips and a translation ROM, but the BIOS images duplicate and are not split into even and odd bytes, it's just two identical whole images. However, card won't display a picture if any of the two BIOS EPROMs are removed, can't imagine it accessing BIOS using a 16 bit data bus, but who knows.

By the way, your card has a translation ROM, it's just soldered.

SSTV2 wrote on 2026-09-16, 23:31:

By the way, your card has a translation ROM, it's just soldered.

So that empty socket isn't for Translation ROM ?
Well, it's easy to check - run VDIAG.EXE -> CONFIGURATION SUMMARY, does it display "Translation ROM: Present"?

SSTV2 is correct... It's not a translation rom missing. I tried every mode imaginable and zero issues. So now how do I get this to use the full 16 bit data bus? 🤔 I uploaded the chip under the missing one to The Retro Web if anyone wants to check it out. Words are readable in the dump so it's not a split bios. What's really odd is the jumper is set to bios 16 like all the pics I've seen online of those with all chips populated. I removed the jumper and no change. 🤷‍♂️ Can I split the bios in two or put a second copy in the missing socket?

I asked Ai because this is a bit over my head. Can anyone confirm this correct? It spit out "00" as a result which indicates its running in 16 bit? Is that possible with just one bios rom?

Interesting, if a 16 bit data bus setting for ROM is controlled by ET4000's internal registers, then a second identical BIOS chip from Genoa would make no difference, as that BIOS simply doesn't set the required register. I looked up the datasheet for ET4000, it appears that one of the indexed (3D4h) CRTC registers sets 16 bit data bus option for ROM - register 37h (VSCONF2) bit 4 (see pg. 137). I tried to read it on my card using debug, but every time got inconsistent data when trying to read back port 3D5h, unlocking the CRTC registers didn't work either because registers 3BFh and 3D8h simply would not set on my card.

I tried to unlock (see pg. 102 "KEY") upper CRTC registers using these debug commands, which are equivalent to the given mnemonics:
-O 3BF 3
-O 3D8 A0

Register 3BFh would not change from 00h, 3D8h would write, but bit 2 was permanently stuck at 1.

Reading VSCONF2 should look like this:
-O 3D4 37
-I 3D5

Same as above - index port 3D4h could not be written to.

3C4h/3C5h ports could be written to or read back, but those "Timing Sequencer" registers don't set ROM bus width, like Ai claimed.

You could try burning two copies of this card, its BIOS image should be compatible with Genoa, as it uses an identical RAMDAC. If your card would fail to start with just one BIOS ROM, then in theory it should operate using 16 bit data bus with two copies.