From Newsgroup: comp.sys.ibm.ps2.hardware

Folks, the third Micro Channel system I ever bought was a Model
8590-0G9. Classic lines of the Model 70...

IBM had to take one for the team when it decided to abandon the
cable-less approach of the Model 70. DBA-ESDI support was limited to TWO internal drives. SCSI, OTOH, supported SEVEN devices, and had an
external port. If DBA-ESDI ever had an external port, it has never
caught my attention...

Cable management on the 90 sux. I'm not sure WHAT IBM had in mind for
the 5.25" drive bay, B: , quite possumbly a 5.25" floppy, a 3.5" floppy
or tape? Think of a floppy device in the B: bay, just unfold the floppy
cable to hook it up and the PSU is right there with power plugs.

Cable management for a CDROM is miserable. H8te the later thin CDROM
bezel when you are taking the cover off. The front does not stay in
place, but comes forward and pivots up at the rear. Model 77 does a much better job.

IBM driver support for the XGA suxd. 640x480 @ 256 screaming colors was
no better than VGA, resolution-wise. True, the VGA speed was improved,
but...

Then, the Omnimessiah sendt UZnal. He figured out the Direct Color thing
and added support for 800x600 @ 64K. PBUH!

Recently, Michal Necasek has been looking into better drivers for OS/2.
I turned him on to UZnal's Windows drivers, and started to purify myself
by the INMOS G200 XGA controller and the IBM tech ref on Video
Subsystems ['92]. Very odd to have documentation on an IBM ASIC...

One thing led to another, many horns of Jolt were quaffed and we leapt
from Model 90 to Model 90 throughout the riotous night. [like the "13th Warrior, but different]

Much meditation on the SIMMple G191 / G200 diagrams was performed
without the most Holy elixir, Jolt. I then noticed the F245 octal 3
state transceivers control the data flow to / from the Micro Channel
bus. Oh-ho, perhaps a weg to pin-out the G200 to the MCA bus for A0-A31
and D0-D31.

MAJ Tom whipped up a crutility [Crude Utility] to query the system as to
the slot size of the Model 90's planar XGA, it self-identifies as
"32-bit". Does that mean 32-bit Data -AND- 32-bit Address? Painful
probulation is indicated. MAJ Tom coyly suggested probing A31 and D31 to
prove support for 32-bit-ness.

http://ps-2.kev009.com/ohland/XGA/Direct_Color_Mode.html

I am at a loss as to which weg UZnal implemented "direct color". There
are TWO wegs to accomplish it.

The first is actual "direct color", where you bypass the palette and
directly write the 16bpp video directly to the video DAC. Unfortunately,
this disables the XGA co-processor. And it also requires the system
processor to shovel bits.

The second weg is chopping the 16bpp video into two 8bpp halves and now
you can use the XGA co-processor.

I wonder what the performance of 16bpp in direct color vs the
co-processor supported 8bpp is?

For the record, the XGA-2 CAN perform Direct Color if you wandt to, but
it has an 8-bit DAC, so it doesn't need to.
--- Synchronet 3.21a-Linux NewsLink 1.2