#retrocomputing

They talk about a theoretical board design that could do up to 1280x1024.

I think they meant four ET2002's, as four attribute controllers don't make a lot of sense as it's not like rasterizing to a CRT is a parallelable operation.

This is the first mention of the ET2003 bit-slice controller I have seen anywhere. The bit-slice engine and the ability to coordinate with a co-processor (like the Z80 on the GB-1) were clearly features targeted at producing a IBM PGC clone.

The failure of the PGC to establish itself as a standard probably scuttled all those plans, and so the ET2000 was doomed to remain only the heart of an unassuming EGA clone, with its coprocessor support relegated to doing port translation for CGA titles.

Talk about a "you pass butter" moment.

The Internet Archive is really a priceless resource. Here's a note from a 1987 issue of "Computer Magazine" about what was fixed in a 4/6/1987 ROM revision for the ET2000.

I have seen photos of boards with this date printed on their ROMs, but nobody has dumped it yet.

If you collect video cards and don't dump their ROMs and upload them there is a special place in hell for you.

There's also a mention of the ET2000's "microsequencing engine". Whatever the hell that is.

I suspect that's another part of intended PGC emulation support.

This is basically a prototypical GPU - it has an 8088 CPU and 64K of ROM. Its video BIOS routines has support for primitive 3D graphics.

It can do 640x480 - in 256 colors. And this is before VGA even existed.

This card was designed for IBM by a company called Vermont Microsystems, and one of the designers would go on to be a cofounder of a little firm you may have heard of called Nvidia.

You don't talk to a PGC by your traditional IO pokey pokey. It provides a memory-mapped interface from C6000 to C63FF.

There's a lot of status bits to read and write, because the PGC is basically a whole-ass computer in your computer.