1. 01 May, 2000 3 commits
    • Aaron Holtzman's avatar
      Oops. Forgot this. · 5e0a911a
      Aaron Holtzman authored
    • Aaron Holtzman's avatar
      SDL update from Ryan Gordon. Also add a flag to display_init to go · 6566acb0
      Aaron Holtzman authored
    • Aaron Holtzman's avatar
      Major core reworkage of mpeg2dec to allow libmpeg2 to be used by · 0193cc9c
      Aaron Holtzman authored
      The big change is that the mpeg2_decode_frame changed to mpeg2_decode_data
      and the fill_buffer callback was removed. mpeg2_decode_data now
      returns the number of frames finished during the call. This required a number
      of internal changes, and a lot of code was touched.
      Unfortunately this change lost us %5 in performance on my machine. Don't
      worry though, the optimizations we have planned in the next few weeks should
      make this seem like a drop in the bucket.
      Incidentally, I have been playing around the perf counter patches on my
      celeron 366. I have measured our CPU utilization at 1.5 CPI for a particular
      test clip. This is pretty terrible for a OOE cpu and just goes to show
      how hamstrung we are by our cache footprint.
  2. 26 Apr, 2000 2 commits
  3. 11 Apr, 2000 1 commit
  4. 04 Apr, 2000 3 commits
  5. 03 Apr, 2000 1 commit
  6. 02 Apr, 2000 2 commits
  7. 29 Mar, 2000 1 commit
    • Thomas Mirlacher's avatar
      cleanup & bugfixes · dda3f8b7
      Thomas Mirlacher authored
      1st try to cleanup paths
      ... and other thing i don't rember right now (probably i should commit in
      	shorter periods, or take notes)
  8. 27 Mar, 2000 1 commit
  9. 08 Mar, 2000 1 commit
  10. 13 Feb, 2000 1 commit
  11. 10 Feb, 2000 1 commit
    • Aaron Holtzman's avatar
      Quick hack to fix bistream so dent will stop complaining about · 90f42fa7
      Aaron Holtzman authored
      mpeg2dec not working directly in oms :)
      bistream.c has been changed to read chunks of data into a 64k buffer
      and hence get the required 32 bit align, and multiple of 4 bytes.
      And it doesn't even go slower (this is a case of give a little,
      take a little). It could probably be faster if I restructured
      the bitstream internals a bit.
  12. 31 Jan, 2000 1 commit
  13. 28 Jan, 2000 1 commit
    • Aaron Holtzman's avatar
      Some more motion_comp fixes from Peter Bajusz (from Jan 26). · 8c10fb7d
      Aaron Holtzman authored
      Note that when you send me patches, _please_ put [patch] in the
      subject line. This will prevent the hell I'm experiencing now
      trying to find sort out all the patches in my mailbox. Also,
      try and keep the patch size small (one feature per patch). This
      will also make my life easier :)
      Thanks Peter, for your help in fixing up the motion compensation.
  14. 25 Jan, 2000 1 commit
  15. 09 Jan, 2000 1 commit
  16. 08 Jan, 2000 4 commits
  17. 06 Jan, 2000 2 commits
  18. 05 Jan, 2000 1 commit
    • Aaron Holtzman's avatar
      Landed bp_frames branch. It doesn't work yet, but I needed · b98e3176
      Aaron Holtzman authored
      some of the code in order to move the codec into libmpeg2.a.
      b and p frames are turned off by default. At some point in
      the near future, I'll move all of the libmpeg2 files into
      their own directory, with possibily it's own configure
  19. 04 Jan, 2000 1 commit
  20. 03 Jan, 2000 4 commits
  21. 29 Dec, 1999 2 commits
    • Steaphan Greene's avatar
      · ae3d2903
      Steaphan Greene authored
      -Minor updates to the (still non-functional) Rage128 Liner mode.
      -Promised autoconf changes from Daniel Caujolle-Bert, include:
      the --with[out]-mga --with[out]-gatos and --with[out]-x11
      These each enable/disable each display type.
      Only one can be selected at once, and the default is mga.
      If mga or gatos is selected and not detected x11 will be used.
      ...I tested this a bit, and it seems to work fine, but I don't have a
      Matrox, so let me know if something doesn't work like it should.
      BTW- Has anyone gotten the GATOS display working besides me?
      --Insomnia (Stea Greene)
    • Steaphan Greene's avatar
      · a7039936
      Steaphan Greene authored
      	Well, here goes my first commit to LiVid's CVS....
      	-Added display_gatos.c to mpeg2dec.
      	-Added autoconf updates to mpeg2dec to detect GATOS.
      	The display_gatos thingy really doesn't work well yet for any of
      the Rage128s.  For any of these cards you have to modify the code (as I
      mentioned in my originaly message to livid-dev).  Eventually it'll be nice
      and generic, but before I can do that, I have to get it working better on
      the 128s.
      	I'll put up a change to the autoconf stuff to add a configure
      option to disable gatos detection for these cards tomorrow (software is
      currently a lot better than using the 128's BES).  I don't know autoconf
      very well, the autoconf changes are/will be written by Daniel
      Caujolle-Bert, another member of the GATOS team.
      	Hopefully I didn't break anything.
      						--Insomnia (Stea Greene)
  22. 24 Dec, 1999 1 commit
    • Aaron Holtzman's avatar
      Better g200 support from Real Ouellet. The card · d4a0ff7f
      Aaron Holtzman authored
      type is returned via the config ioctl now, so there
      are no g200 defines in display_mga_vid anymore.
      I have two more patches in my TODO pile. One is
      mmx support for dithering, and the other is Solaris
      vis optimization. These will have to wait until I
      get back home (29th) as I don't want to sort out
      patch conflicts remotely over a 28.8 dialup using MS
      telnet :)
      Merry Christmas!
  23. 21 Dec, 1999 1 commit
  24. 20 Dec, 1999 2 commits
  25. 18 Dec, 1999 1 commit
    • Aaron Holtzman's avatar
      g200 patch from Real Ouellet <realo@sympatico.ca> that I put into · de01eba0
      Aaron Holtzman authored
      0.1.5, but somehow got reverted along the way. You need to define
      IS_G200 1 in mga_vid.c in order to have this work. To make this
      work automagically, we need to add a chipset field to the mga_vid_config
      struct that the driver fills in and copies back to user space.
      I also removed all generated files from the repository. If you have
      a fresh checkout you need to run ./autogen.sh.