Very Computer

Whooooaa - one of my machines works with the new server
from XFree86 (3.1.2A beta) - its got an AT&T 20C408 clock
chip, but the other seems to not work ....

It says its got a 'BedRock' clock chip, seems to accept
the programmed clocks, finds some modes which fit the clocks
and starts up. The screen stays black, and requires a 3-finger
special to recover. It also thinks the card is EISA, perhaps
causing the lack of image.

Ive seen one other posting with this same problem (EISA for
PCI, BedRock clock, 3 finger get out). I assume that the cards
are different, but what is an ATI-264CT?

To add to the mystery. I tried the XInside demo. Lovely, nice
working in 16 bit too. It uses the board type xpression2.xqa
from the ATI directory. This indicates that the card is actually
    Board          = "Graphics Xpression (Mach64/DRAM)";
    GraphicsChip   = "ATI88800";
    Ramdac         = "STG1702";

The XFree86 README.Mach64 says it that XF86_Mach64 supports the
STG1702 ramdac. Why can XInside use it ok, but XFree86 not see
it - and what is this BedRock chip.

Below are the reports from DOS Install stuff, and X probing.
WHAT card do I have? Any help would throw some light somewhere.

The DOS install stuff reports:
    Controller Type      PCI ATI-264CT
    Video BIOS Version   1.060, 1995/08/03
and the X -probeonlys salient points are:
    (**) Mach64: Option "no_bios_clocks"
    (**) Mach64: Option "program_clocks"
    (--) Mach64: card type: EISA
    (--) Mach64: Clock type: BedRock
    (--) Mach64: Maximum allowed dot-clock: 80.000 MHz
    (**) Mach64: Mode "1024.80": mode clock =  80.000
    (**) Mach64: Mode "800x600": mode clock =  50.000
    (**) Mach64: Virtual resolution: 1280x1024
    (**) Mach64: videoram: 2048k
    (--) Mach64: Using hardware cursor
    (--) Mach64: Using 4 MB aperture
    (--) Mach64: Ramdac is Internal
    (--) Mach64: Using 6 bits per RGB value

--
-------------------------------------------------------
Linux - reaches the parts other beers fail to reach.
WWW: http://www.cms.dmu.ac.uk/~rl
WWW: http://www.cms.dmu.ac.uk/CCSR
-------------------------------------------------------

The XQA file is our board description file (eXtended Query Architecture.
No, don't ask why).  Many board vendors keep the same board name, and
then change stuff, like the DAC, graphics chip, memory, clock, etc.
Some board vendors keep the hardware the same, and then change the board
name...

Anyway, our XQA file lists the STG1702, because that was what the first
Xpression had on it.  We silently recognise the new DACs that ATI puts
on the Xpression boards (e.g. before the AT&T 20C408, it was the
Chrontel 8398, I think).   Admittedly, we're not precognitive.  We do
have to issue a patch on the Anon-FTP Server to support new boards under
the old names.

As for the 264CT, yes, this is a special version of the Mach64 with an
integrated DAC, a bit like the Cirrus chips or the new Oak 111 chipset.
This should work with the patches for the ATI module in the Update C
module (which has been embedded in the demo version and on our diskettes
since late October).

You can find out what we *really* think is on the board, by looking at
our Server text output.  Just start the Server alone (Xaccel at the
command line) and then kill it with <Ctl-Alt-Backspace> (Terminate
Server With Extreme Prejudice).  You should be faced with a text screen
and a line that says "ATI,2:" and then the remaining stuff tells you
what graphics chip we found, which DAC, memory, PCI linear address,
clock frequencies and other stuff that you should never have to worry
about if you had a properly functioning Server :-)  The "ATI,2:" bit
tells us that you have the latest released version of the ATI driver
module, from Update C.

If you've been having problems with our Server, and the line says
"ATI,1:", grab the Update C package from the Anon-FTP Server in URL
ftp://ftp.xinside.com/accelx/1.2/patches/ .  You'll want README.update_c
and one of the compressed versions of update_c.tar .

The alternative to not precisely following the XQA file,was to make a
board description file for each board.  We think that customers should
not *have* to know more than is printed on the end of the box that the
graphics board came in...  And besides, with the rate at which some
vendors change components, we'd have pages of some board types alone,
with little qualifiers about the exact DAC type, memory chips, and so
on.  We decided we could make life easier, at the expense of not
explicitly listing every DAC, graphics chip and so on.

Cheers, JeremyC.
--
Jeremy Chatfield, +1 303/298-7478, FAX:+1 303/298-1406,

       X Inside Inc, 1801 Broadway, 17th Floor, Denver, CO 80202
   Commercial X Server - for more information please try these services

ftp://ftp.xinside.com

ATI-264CT is the low end Mach64 controller from ATI. I understand that
it has got an integrated RAMDAC and clock. I asked the XFree86 if this
chip was supported in the new beta, and they replied that it wasn't.

I have the same card, same VIDEO BIOS version, and am using XInsid. I
am very satisfied with it.

Quote:>To add to the mystery. I tried the XInside demo. Lovely, nice
>working in 16 bit too. It uses the board type xpression2.xqa
>from the ATI directory. This indicates that the card is actually
>    Board          = "Graphics Xpression (Mach64/DRAM)";
>    GraphicsChip   = "ATI88800";
>    Ramdac         = "STG1702";

>The XFree86 README.Mach64 says it that XF86_Mach64 supports the
>STG1702 ramdac. Why can XInside use it ok, but XFree86 not see
>it - and what is this BedRock chip.

Both XFree86 and X Inside report a lot of information which is
wrong. But XInside works, nevertheless.

Quote:>    (--) Mach64: card type: EISA

This is a minor bug, but the effect is that it probably selects a
wrong memory aperture address. Add the following to the device section
of XF86Config (using whichever address scanpci suggests):

    MemBase     0xf0000000

Although the developers don't expect it to work, there's still a
chance.

Good luck

Jon
--

Div. of Computer Systems and Telematics         Phone: +47 73 59 43 27
Norwegian Institute of Technology               Fax:   +47 73 59 69 73
Trondheim - Norway