A detailed descriptioon follows. Does anyone else get this, or have
any suggestions about how to fix it?
The card is a Diamond Stealth VRAM (version 3.3 according to the BIOS,
with 1 MB). It isn't listed at the Diamond web site, I suppose since
it is an older card (not the Stealth 64 everyone is posting
about). With any of Linux 1.2.8, 1.2.13, 1.3.28, Xfree86 3.1.1 or
3.1.2, I get the same results. If I switch back to a VC from X, then
back to X, I get black pixels scattered covering the screen. They are
somewhat random, but roughly organized in columns separated by 16
pixels. The dots go away if I use the twm refresh menu item, but
reappear if I switch back to a VC and then back to X again. Also,
there are some stripes that trail along with the mouse cursor. They're
not mouse droppings, since they follow the mouse, they don't remain on
the screen in places the mouse already moved away from. These dashes
are about 12 pixels long, one pixel tall, and located on 16 pixel
centers, horizontally. There are 4 dashes horizontally and many
vertically, fitting I guess in a 64x64 pixel area with the actual
mouse cursor at the top left. The stripes are usually black or white
(in my solid grey root window), but their color is somewhat correlated
with they cursor and background colors of wherever the mouse happens
to be. I see the same behavior in any mode, even 640x480. I turned off
my IDE card's game port, as well as ttyS01, and switched it between
lpt0/lpt1, thinking to check for hardware conflicts, but it made no
difference in the mouse stripes or black pixels. The only other card
in the system is an Intel Etherexpress (see my posting about it in the
networks group :-). It has memory mapping turned off. Originally, I
had a bus mouse which seemed to be correlated with black pixels on
sarting X, but it's gone now (I have a serial mouse on ttyS00), and I
have to switch to a VC first then back to see the black pixels now.
If I switch to a software cursor, then I get mouse droppings in a
small area around the cursor. I think they get cleared up by
In the VCs, there are no problems at all, no black pixels, no mouse
droppings or trailers (with gpm). OK, there is one big problem which
is that the fonts get messed up switching from X or after using an
SVGAlib program, but they can be fixed by typing 'setfont m'
blindly. This is only with 3.1.2, 3.1.1 didn't have this
problem. Occasionally switching back to X the screen will be shifted
in some random direction and I get to see the VGA font at the bottom,
but not usually (switching again usually cures it). I'm not sure if
this happened with 3.1.1 also.
Any suggestions? Does anyone else use this card without all this stuff
happening? The relevant part of my config file follows, then some
output from X. I normally use 1024x768cust. Using options "nolinear"
or "diamond" in various combinations makes no difference.
ModeLine "1024x768cust" 75 1024 1024 1224 1280 768 768 768 789
# SVGA server auto-detected chipset
Identifier "Diamond Stealth VRAM"
BoardName "Stealth VRAM"
# Clocks 25 28 33 40 25 28 33 40
Device "Diamond Stealth VRAM"
Monitor "MAG MX14S"
Modes "1024x768cust" "640x480" "640x480-28" "640x480-33" "800x600-40" "800x600" "1024x768i" "1024x768" "1024x768-90" "1024x768-100"
ViewPort 0 0
Virtual 1024 768
# Option "nolinear"
# Option "diamond"
# Option "sw_cursor"
Some X output for net.perusal:
(--) S3: chipset: 924
(**) S3: chipset driver: s3_generic
(**) S3: videoram: 1024k
(--) S3: Ramdac type: normal
(--) S3: Ramdac speed: 110
(**) S3: Using ICD2061A programmable clock
(--) S3: Maximum allowed dot-clock: 110.000 MHz
(**) S3: Mode "1024x768cust": mode clock = 75.000
(**) S3: Mode "640x480": mode clock = 25.000
(**) S3: Mode "640x480-28": mode clock = 28.000
(**) S3: Mode "640x480-33": mode clock = 33.000
(**) S3: Mode "800x600-40": mode clock = 40.000
(**) S3: Mode "800x600": mode clock = 50.000
(**) S3: Mode "1024x768i": mode clock = 44.000
(**) S3: Mode "1024x768": mode clock = 75.000
(**) S3: Mode "1024x768-90": mode clock = 90.000
(**) S3: Mode "1024x768-100": mode clock = 100.000
(--) S3: Using 6 bits per RGB value
(**) S3: Virtual resolution set to 1024x768
(--) S3: Using a banksize of 64k, line width of 1024
(--) S3: Using a single 64x64 area at (960,769) for expanding pixmaps
(--) S3: Using 8 planes of 960x255 at (0,769) aligned 8 as font cache
NASA/George C. Marshall Space Flight Center I'm not affiliated with
Code ES84 NASA.
Huntsville, AL 35812
My URL is http://xanth.msfc.nasa.gov/~raj/home.html