UFO: Alien Invasion Issue Tracker
star_faded.png
Please log in to bookmark issues
bug_report_small.png
OPEN  Bug report #2562  -  PPC: Software lighting is broken
Posted Jun 12, 2010 - updated Jan 28, 2013
arisian (arisian) has been working on this issue since January 28, 2013 (20:47)
Issue details
  • Type of issue
    Bug report
  • Status
     
    Open
  • Assigned to
     arisian
  • Type of bug
    Not triaged
  • Likelihood
    Not triaged
  • Effect
    Not triaged
  • Posted by
     keybounce
  • Owned by
    Not owned by anyone
  • Estimated time
    Not estimated
  • Category
    Battlescape
  • Resolution
    Not determined
  • Priority
    3. Normal
  • Reproducability
    Not determined
  • Severity
    Not determined
  • Targetted for
    icon_milestones.png Not determined
  • Complexity
    icon_customdatatype.png Not determined
  • Platform
    icon_customdatatype.png Not determined
  • Architecture
    icon_customdatatype.png Not determined
Issue description
Item 2562 imported from sourceforge.net tracker on 2013-01-28 19:39:33

As reported in thread http://ufoai.ninex.info/forum/index.php?topic=4877 and elsewhere:

I have a PPC mac. When using a map downloaded that was compiled on a 386 system, the lighting is horrible -- the map surface is indecipherable / nothing is recognizable. When using a locally compiled map, the lighting is mostly good -- a few odd patches here and there.

Usually.

Sometimes it is perfect. In at least one case, an i386 compiled map would flicker between perfect and horrible.

I have two machines. One has a 9300 (Mac mini) video card and cannot enable GLSL shaders. The other, an iBook g4, has a 9550 mobility, can turn on GLSL shaders, has perfect display when GLSL shaders are turned on, but generates one frame every 25 seconds.

As I said in that thread: We know that the PPC compiled maps work fine on i386, so it's probably not the map editor or the map compiler (yet i386 compiled maps are so much worse on the PPC); we know that i386 with GLSL turned off looks good, so that leaves something about PPC software light generation.

I am out of my league here; I don't know what to look for, or what values are reasonable. Can someone who knows this please add some debugging output to the code so this can be identified? Thank you

(Odd: Group does not have 2.3. It has 2.2.1, and 2.4, but not 2.3.)

Os: 10.5.8

Svn version: Everything from around 28xxx to 30088. Note that this was also present early in the 2.3 work back last august. I thought it had vanished at one point (sadly I was away from following UfoAI for almost 9 months)

NB: This also affects lighting on the geoscape -- everything is too dark to see anything.
Comments Ported from Sourceforge  ⇑ top
arisian (2010-06-19 13:34:05)  ⇑ top
I'm mostly familiar with the GLSL lighting, not the software lighting, since I only coded the former, but I'll try to figure out what's going on. I suspect that the reason the 9550 mobility renders one frame every 25 seconds is that the driver is actually using software-emulation to do the rendering. The ATI drivers will switch over to software rendering any time they detect something that the card can't handle.

The i386 to PPC conversion issue could be an issue of endian-ness; I'll try to take a look at the file I/O code to see if it handles little-endian and big-endian systems properly.
keybounce (2010-08-11 21:52:14)  ⇑ top
Bump

Have you had a chance to look at this yet?
tlh2000 (2010-08-21 16:55:14)  ⇑ top
@keybounce: could we maybe get a teamviewer session or any other remote debugging session? i don't own a big endian system and it would really help to see and feel the things. please join the irc channel and let's talk about it.
aduke1 (2012-10-01 00:50:16.269000)  ⇑ top
- **milestone**: 2.3 --> 2.3.x

Steps to reproduce this issue
Nothing entered.