#1008 NORM Update.: Play MPEG-1 video is not smooth via MPEGPlayer.
Zarro Boogs per Child
bugtracker at laptop.org
Thu Nov 8 11:27:56 EST 2007
#1008: Play MPEG-1 video is not smooth via MPEGPlayer.
-----------------------------+----------------------------------------------
Reporter: chiaying.lin | Owner: etoys
Type: enhancement | Status: new
Priority: normal | Milestone: Update.2
Component: etoys-activity | Version:
Resolution: | Keywords:
Verified: 0 |
-----------------------------+----------------------------------------------
Changes (by bert):
* milestone: Opportunity => Update.2
Comment:
(via email from John !McIntosh )
Bert, I have a changed mpeg support file here.
The decoder logic would use this, and other construct
{{{
d[0] = s[0]; d[1] = s[1]; d[2] = s[2]; d[3] = s[3];
d[4] = s[4]; d[5] = s[5]; d[6] = s[6]; d[7] = s[7];
}}}
Basically moving one byte at a time, however a few years back I discovered
this was a bottleneck in the mpeg frame performance and changed it to
{{{
t1 = *(long *) s;
t2 = *(((long *) s)+1);
*(long *) d = t1;
*(((long *) d)+1) = t2;
}}}
This made like a 30% performance gain in the powerPC frame rate.
however because I could not test on INTEL I had coded it as __POWERPC__
change only, plus at the time MMX was being used for Windows.
In looking at the OLPC stuff I'm wonder if we could use the same code on
INTEL.
Thus I tried on my !MacIntel box and it seems to be ok.
I've included the source file, perhaps you could compile up for OLPC and
send us back a Squeak VM?
You might want to see if it improves the OLPC EToy MPEG playback first
tho?
Look for the changes via
{{{
#ifdef __LONGMOVEMENTVERSUSBYTE__
}}}
Attachment: reconstruct.c
--
Ticket URL: <http://dev.laptop.org/ticket/1008#comment:9>
One Laptop Per Child <http://dev.laptop.org>
OLPC bug tracking system
More information about the Bugs
mailing list