No Next Item No Next Conference Can't Favor Can't Forget Item List Conference Home Entrance    Help
View Responses


Grex Micros Item 26: here's a problem for you...
Entered by keats on Thu Apr 16 14:16:04 UTC 1992:

i have a 386/20 with cache, and not too long ago, i added an ati ultra
graphics accelerator card. when i changed over from standard vga to 800x
600 with 256 colors, something interesting (read, "annoying") happened.
any graphic i saved, be it a .bmp or even an icon, had its colors changed:
browns became yellows, light greens dark greens, and other color changes
i cannot recall specifically. if i attempted (in 256 mode) to reedit these
graphics and change the colors back, when i saved any graphic, it changed
back to the wrong colors. in effect, i can display 256 colors but i can only
edit and save as few as 8. 

well, i reinstalled the .bmp files from their original sources, and since
one doesn't normally edit those, they're safe as they are for the time being.
but i have a huge icon library that had roughly a dozen or a dozen and a
half icons distorted. i thought about it for a while, and went back to 16-
color mode. i edited and saved the icon library, and behold, the graphics
retained their proper colors. 

it gets more complicated. when i went back to 256-color mode, i opened the
icon library. the graphics were in the correct colors (so far, so good). 
but when i attempted to load an icon onto the screen, sure enough, the 
graphic loaded with the wrong colors (the library, btw, retained the graphics
in their correct colors).

i tried copying the icon library to a floppy under a different name and then
loading it back on in 256-color mode to see if that would make the driver
respect the color distinctions. it didn't. 

i have, naturally, called ati, and this stumped the online representative.
i'm awaiting a call back, but i'm not very optimistic. got any ideas about
why this happens and how i can fix it?

14 responses total.



#1 of 14 by steve on Thu Apr 16 20:56:40 1992:

   Is the software you're using incapable of using 256 colors?  This
smells of the classic 'color pallate mashing' problem.  It has been
the case in the past that a lot of DOS software has acted weirdly when
hardware is upgraded, etc.  So I'm not sure where the problem is, but
ATI should be able to help out eventually.  If they can't, list all the
software here.  My collection of DOS stuff is growing quite rapidly these
days.


#2 of 14 by keats on Thu Apr 16 23:08:10 1992:

the two programs in question are windows and norton desktop. what's weird
is that the functions i'm using (paintbrush, icon editor) only use 16 colors,
anyway, and that if i put the card in 16-color mode, it all works fine. it's
when i tell the card that it can use 256 colors that it does this.

i think both programs are compatible with 256, because afterdark, which uses
256 at svga, runs fine on the computer (through the same shell as would the
paintbrush or icon editor). 

i haven't noticed any other irregularities...yet.


#3 of 14 by jdg on Fri Apr 17 01:24:10 1992:

Icons for windows are 16-color images.  I've found (back when I had SVGA)
that I couldn't easily manipulate 16-color bitmap graphics when in 256
color mode.  Keep your various SYSTEM.INI files around to make switching
easier.


#4 of 14 by bad on Fri Apr 17 02:02:28 1992:

I've actually had the same thing happen with Windows icons (using 
Icondraw), and an ATI SVGA chipset. Happened sporadically, and Windows
was too klunky anyway, so I never bothered to look into it too much further.
I think I only had it happen once, if at all, with paintbrush, when I used
some other SW program to convert 256-c GIFs.


#5 of 14 by mistik on Fri Apr 17 02:15:16 1992:

What happens if you turn all the caching off?  I read something on usenet
that some of the caching hardware has problems, about 10% of the 386/486
motherboards are affected, Gateway was mentioned, but they also said
other brands had this problem as well.  It surfaced as a PKZIP problem,
when they traced it back to IO errors.  Reportedly, it seems that the
CPU does not relize that memory was updated by DMA, and the cache does
not get reloaded.  Reportedly, this happens only under certain
timimg conditions, your board might be aggrevating it.

So if it goes away when you turn off caching, then you know what it is.
Otherwise, I don't know.


#6 of 14 by keats on Fri Apr 17 05:57:25 1992:

i'll try turning off the caching, though i don't have pkzip anywhere on
the computer currently. in the meantime, i'm grateful for all possible
suggestions. 

btw, i called ati again today and they told me they'd get back to me.
(they didn't.) i'll let you know what happens if they do.


#7 of 14 by bad on Fri Apr 17 06:27:28 1992:

Caching problems? Hmmn...now *that* is interesting.


#8 of 14 by keats on Fri Apr 17 17:25:02 1992:

i tried turning off my cache. the system, booting up and working in its 
environment, was outstripped by a rival molasses coprocessor, but the
graphics error still occurred. thanks for the suggestion, but it's got
to be something else. and as ati _still_ hasn't called me back, i'm open
to more good suggestions.


#9 of 14 by mistik on Sat Apr 18 01:05:34 1992:

The problem is not related to pkzip, dma and cache working at the same time
seem to cause a problem on some machines.  Since you turned caching off,
it must not be related to that problem.


#10 of 14 by rcurl on Sun Mar 14 05:16:49 2004:

Here's another problem for you.

I have a Star SG10 printer that worked fine with an 8086 machine running
DOS 3.3. I have tried to run Print Screen with it from a 486 DX2 machine
running DOS 6.3. It doesn't work. Why, and can this be corrected?



#11 of 14 by keesan on Mon Mar 15 03:54:26 2004:

We have another Star printer that won't print screen with 2 486s or 3
pentiums, and we asked Rane to test his out.  Ours also won't print from a
DOS text editor, or WORD for DOS.  The person using it claims it does print
perfectly from her 486 DX2 66MHz AT&T desktop, both PrtScr and Pine (type Y
to Prynt this mail).  And we have one pentium that does Pine PrYnt but two
that won't, despite changing phone numbers to grex, printers, cables, and
modems (2 or 3 of each).  


#12 of 14 by albaugh on Mon Mar 15 18:58:08 2004:

(12 years minus 1 month between responses - yee haw! :-)


#13 of 14 by rcurl on Mon Mar 15 20:42:05 2004:

It was the most appropriate existing item in this cf. I'm an item
conservationist. 

In regard to the topic - of printer compatibiity with later OSs - could
it be simply that a printer that grew up with an 8 MHz processor could
not find happiness in a 66 Mhz family?


#14 of 14 by keesan on Tue Mar 16 00:02:31 2004:

We never used my Star with anything faster than 8MHz, so you could be right.
I kept my 1985 XT for 12 or 13 years but replaced the printer with a 24-pin
a few years after buying the computer.  I still use the 24-pin printer.

Our other printing problem was due to Jim putting on an old version of kermit
because it was smaller.  All versions are pretty small but he wanted to get
lots of other things on his boot disk.  So now we have to find some way to
replace the kermit on our friend's computers and she does not know how to
download files (we taught her a few times) or use floppy disks, and it is a
3 hour bike ride from here to there.  

Response not possible - You must register and login before posting.

No Next Item No Next Conference Can't Favor Can't Forget Item List Conference Home Entrance    Help

- Backtalk version 1.3.30 - Copyright 1996-2006, Jan Wolter and Steve Weiss