<br><div class="gmail_quote">On Thu, Apr 3, 2008 at 12:54 PM, Steve Peters - Priority Electronics <<a href="mailto:steve@priorityelectronics.com">steve@priorityelectronics.com</a>> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div class="Ih2E3d"><br>
><br>
>I'm also holding off in the hopes that the issues with excessive CPU<br>
>usage that keep being reported here will be fixed.<br>
><br>
>I'll probably do what I usually do with Windows, and wait for the .1<br>
>release. ;)<br>
><br>
>____________________________________________<br>
<br>
<br>
</div>David, it's been my experience that the reason for the extra cpu usage is<br>
because the playback profiles give a higher quality playback. I was using a<br>
different deinterlacer on my mythbox with version 0.20.2, and my playback<br>
was not as smooth. I now use Yadif 2X, which I think is supposed to be the<br>
most cpu intensive, and during playback, my cpu is now at around 13% for SD<br>
Playback. It was at 10% before using kernel for deinterlacing. If I go back<br>
to kernel now on my 0.21 system, it goes backt o 10% cpu usage for playback.<br>
<br>
I wonder if many people are saying they're seeing higher cpu usage but are<br>
not realizing that the playback profiles may be the reason. This is an<br>
improvement in my opinion because better playback is worth a few percentage<br>
points of cpu usage.<br>
<font color="#888888"><br>
-Steve<br>
</font><div><div></div><div class="Wj3C7c"><br>
<br>
_______________________________________________<br>
mythtv-users mailing list<br>
<a href="mailto:mythtv-users@mythtv.org">mythtv-users@mythtv.org</a><br>
<a href="http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users" target="_blank">http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users</a><br>
</div></div></blockquote></div><br>I believe I have matched my settings on .21 to those I used on .20.2 and I use more cpu than I did
before. That was achieved by adding a simple playback group that
includes only one setting, which is to use XVMC, openGL/openGL for the
OSD, and no deinterlacer. All SD. That's how I ran for months on
.20.2 and there was no trouble at all, and now that I run that way on
.21, my CPU usages is up 10-20 percent (the difference between smooth playback, quick channel changes, and smooth OSD to skipping playback, slow
channel changes, jittery OSD on my 1.7ghz P4 frontend). <br>
<br>
I wish I could compare my two systems side-by-side, but I'm not going to revert to .20.2 just for the test. <br><br>The strange thing is I have an eMac that I switched from running OS X to Debian that seems to perform much better on .21 than it did with .20.2. I'm thinking about trying that switch on my P4 frontend just to see what happens there. <br>