ATI video controller

ATI video controller

[Richard]

I have an ATI Mach32 video board that only seems to work in the basic
640x480 mode.  I have a similar, but newer, ATI board at home that worked
right away, so it's not that the ATI support in aux/vga and /lib/vgadb is
all wrong.  The problem board is a couple years old, clearly a mach32
board with a ics2494AM clock chip (but a BCT245 chip I take to be the ramdac).
At 800x600 and higher res, the screen is looks very much like the timing is
all wrong---patterns of dots/lines that repeat every so often and are
aligned vertically along a diagonal.

I've added a   clock=ics2494a      to no avail.

The reason I'm appealing to the mailing list rather than spending (quite some)
time trying timing parameters is that the fact that aux/vga
always comes up with  "controller not in /lib/vgadb"  even though it
clearly is (and correctly so and at the correct 0xC0085 for
LOCAL BUS ATI ULTRA PRO  which happens as well to match the original
0xC008F   ATI ULTRA PRO ).

Does anyone know if the "controller not in lib/vgadb" message can mean,
"yes, the controller is in the db, but I'm not happy with your description
of it"?

BTW, for the "what models work", my Gateway at home is a 4DX2-66V with
an ATI Mach32 VLB video board and two 405MB IDE drives  (and SB-16 and CDROM
which are untried with the 4 disk floppy distribution).

ATI video controller

[jmk]

	> board with a ics2494AM clock chip (but a BCT245 chip I take to be the ramdac).
the clock chip may be made by ICS but it contains an ATI specific table of frequencies.
there are, according to ATI, 3 possible clock chips on the mach32 which they call the
18811-[012] and i have seen a board with a clock-chip so labelled. the -[12] chips have
the same table. since i can't tell what the chip type is from software, the mach32 part
of aux/vga has a truncated table of frequencies whose entries match on all the chips:
 	25175000, 32000000, 40000000, 44900000, 65000000, 75000000
these frequencies match the common resolutions up to 1024x768. anything higher on the
mach32/64 requires using the accelerator and possibly programming the ramdac (this would
be a good thing to have as the mach64 is quite a zippy chip, hint).


	> The reason I'm appealing to the mailing list rather than spending (quite some)
	> time trying timing parameters is that the fact that aux/vga
	> always comes up with  "controller not in /lib/vgadb"  even though it
	> clearly is (and correctly so and at the correct 0xC0085 for
	> LOCAL BUS ATI ULTRA PRO  which happens as well to match the original
	> 0xC008F   ATI ULTRA PRO ).
	>	Does anyone know if the "controller not in lib/vgadb" message can mean,
	>	"yes, the controller is in the db, but I'm not happy with your description
	>	of it"?
no, it just means it couldn't find a matching address/string in the BIOS. this is
surprising given what you say. let's see the output of 'aux/vga -vp>somefile'.