Bug 115673 - Default bpp set to 15 for Apple Powerbook Pismo (should be 16 or 24)
Summary: Default bpp set to 15 for Apple Powerbook Pismo (should be 16 or 24)
Status: RESOLVED FIXED
Alias: None
Product: SUSE LINUX 10.0
Classification: openSUSE
Component: SaX2 (show other bugs)
Version: Beta 4
Hardware: PowerPC All
: P5 - None : Normal
Target Milestone: ---
Assignee: Marcus Schaefer
QA Contact: E-mail List
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2005-09-07 19:17 UTC by Leah Cunningham
Modified: 2005-09-09 12:02 UTC (History)
1 user (show)

See Also:
Found By: Other
Services Priority:
Business Priority:
Blocker: ---
Marketing QA Status: ---
IT Deployment: ---


Attachments
hwinfo (141.07 KB, text/plain)
2005-09-07 19:19 UTC, Leah Cunningham
Details
15bpp-pismo (169.35 KB, image/png)
2005-09-08 16:00 UTC, Leah Cunningham
Details
24bpp-pismo (324.33 KB, image/png)
2005-09-08 16:01 UTC, Leah Cunningham
Details
15bpp-pism (169.35 KB, image/png)
2005-09-08 22:40 UTC, Leah Cunningham
Details
16bpp-pismo (596.14 KB, image/png)
2005-09-08 22:41 UTC, Leah Cunningham
Details

Note You need to log in before you can comment on or make changes to this bug.
Description Leah Cunningham 2005-09-07 19:17:53 UTC
Sax2 seems to want to set the bpp for the Powerbook Pismo by default to 15,
which works, but doesn't look very good.  It should be 16 or 24 for this system.
 I will attach hwinfo.  Let me know what else is needed.
Comment 1 Leah Cunningham 2005-09-07 19:19:02 UTC
Created attachment 49106 [details]
hwinfo
Comment 2 Marcus Schaefer 2005-09-08 08:18:15 UTC
In the code there is something like: 
 
# special case for PPC... need 15bit colordepth 
# ----------------------------------------------- 
if (($architecture =~ /ppc/i) && ($flag_list[$i] eq "DEFAULT")) { 
   $var{Screen}{$i}{DefaultDepth} = "15"; 
} 
 
so this happens intentionally. Olaf do you agree that this 
special handling isn't needed anymore ? Thanks 
Comment 3 Olaf Hering 2005-09-08 08:26:35 UTC
for r128:
mac on linux requires 15 or the macos colors look wrong.
dri requires 16, 24 doesnt work for me.
r128 cant do 32

16 should only be the default if there is the DRI/3D/whatever checkbox, which
isnt there yet.

I dont have a radeon, but I guess it will behave the same.

Leah, why is 15 wrong?
Comment 4 Leah Cunningham 2005-09-08 14:45:38 UTC
At least on the pismo 15 just doesn't look very good, there is quite a noticable
difference between 15 and 16.  On the pismo, I can use either 'ati' or 'r128'
driver.  Is the requirement for MOL specific to 'r128'?  I can see if I can get
a screenshot to demonstrate the difference.  

I haven't tried MOL yet as I need to go off and find my OSX cds...
Comment 5 Olaf Hering 2005-09-08 14:47:55 UTC
r128 is a subset of 'ati'.

I dont have a radeon, but I guess mol needs 15bit as well there.
Comment 6 Andreas Schwab 2005-09-08 14:49:52 UTC
I have no problem with mol. 
Comment 7 Olaf Hering 2005-09-08 14:53:26 UTC
ok, so we will never know how radeon handles 15, 16, 24 and 32 bit. 
Comment 8 Leah Cunningham 2005-09-08 16:00:57 UTC
Created attachment 49232 [details]
15bpp-pismo
Comment 9 Leah Cunningham 2005-09-08 16:01:34 UTC
Created attachment 49233 [details]
24bpp-pismo
Comment 10 Olaf Hering 2005-09-08 17:19:54 UTC
Marcus, do you know what r128 supports on i386?

DRI: 16 and 24, or just 16?
non-DRI: 15, 16, 24 and 32? 32bit is unsupported here.
Comment 11 Olaf Hering 2005-09-08 19:13:27 UTC
can you try 16bit as well?
Comment 12 Leah Cunningham 2005-09-08 22:40:42 UTC
Created attachment 49282 [details]
15bpp-pism
Comment 13 Leah Cunningham 2005-09-08 22:41:25 UTC
Created attachment 49283 [details]
16bpp-pismo
Comment 14 Marcus Schaefer 2005-09-09 07:25:37 UTC
I assume r128 driver requires 16bit while using 3D features and 
can do 8,15,16,24bit without dri enabled. Maybe Stefan knows more 
Comment 15 Stefan Dirsch 2005-09-09 07:30:52 UTC
Yes, I think so. Anyway, let's concentrate on 2D support. We don't support DRI 
on ppc. This needs to be configured manually on ppc. 
Comment 16 Marcus Schaefer 2005-09-09 07:33:08 UTC
ok, so what to do here ?. I would prefer to be able to remove that 
special 15bit code for ppc but according to Olaf this is still needed 
Comment 17 Stefan Dirsch 2005-09-09 07:38:10 UTC
Don't know. Better keep it for ppc for now. 
Comment 18 Marcus Schaefer 2005-09-09 07:45:23 UTC
ok 
Comment 19 Olaf Hering 2005-09-09 09:19:47 UTC
radeon doesnt support 32bit either, with or without dri.

the yast screen for monitor/resolution/depth did not offer 32bit (on r128)
anyway, either intentially or by accident.

dri works with 16 and 24 bit with radeon 9200 SE, and also with a r128 with 16MB.
Its up to sax to do the math and offer the only the DRI/resolution/depth
combinations that can work. The pismo has 8MB, so 16bit + DRI or 15/16/24 + no
DRI would be valid.

the 15bit or not check should be more like that:
if (arch == ppc) {
   case $card in
      r128|radeon|nv)
         default_depth = 16;
      *)
         default_depth = 15;
}
Comment 20 Marcus Schaefer 2005-09-09 09:25:37 UTC
we don't care about 32 bit anymore, so this is intentionally... no problem here 
Thanks for investigating and providing the information. Well I was under 
the impression we may be able to remove that "hack" but it seems that the 
code snippet needs to grow to a more hacked version :-) 
 
I'm not sure if this should be fixed for 10.0... how important is that ? 
Comment 21 Olaf Hering 2005-09-09 09:49:47 UTC
the older graphics chips can only do 15 bit, not 16. Either noone figured out
how to do 16 bit or they simply cant do it. Thats why only ati and nvidia is in
that list.

it depends how easy is to implement. the default should be 16, glxgears appears
to be faster with 16 bit than 24 bit, 560 vs. 400 fps.
Comment 22 Marcus Schaefer 2005-09-09 11:39:27 UTC
ok, I changed the code to use 16bit for radeon/nv/r128 on PPC  
Comment 23 Olaf Hering 2005-09-09 12:02:07 UTC
tested mol with 24bit+dri on a radeon with 32mb, that works as well. Maybe its
just macos that doesnt do 16bit for whatver reason. It usually offers 256, 32768
and 16,7 million colors.