| View previous topic :: View next topic |
| Author |
Message |
mrpringle
Apprentice
Joined: 27 May 2007
Posts: 284
|
 Posted: Sat Aug 09, 2008 12:15 am Post subject: MythTV internal player CPU intensive compared with mplayer |
 |
|
Hi,
The title sums it up basically. My HTPC systems specs consist of a 2.4GHz core 2 Duo with 2GB DDR2-800 RAM, Geforce 9600GT and a winfast DTV1000-T. Basically when I play back digital TV at 1920x1080 the CPU usage is around 50% when I use the standard internal MythTV player. If I open the recording in mplayer with -vo xv the CPU usage is around 20% for the same recording.
Is there any way to reduce the CPU usage in the internal player. For the tests I turned off de-interlacing in the internal player and I used the standard mpeg playback method.
Is it normal for the internal player to have over double the CPU utilisation compared with mplayer?
Thanks |
|
| Back to top |
|
 |
bunder
Bodhisattva
Joined: 10 Apr 2004
Posts: 5934
|
| Posted: Sat Aug 09, 2008 1:12 am Post subject: |
|
|
while i don't do digital cable and large resolutions, my frontend uses 10% of one "core" on my p4-3.0e ht. not sure how that scales.
cheers
_________________
| Neddyseagoon wrote: | | The problem with leaving is that you can only do it once and it reduces your influence. |
banned from #gentoo since sept 2017 |
|
| Back to top |
|
|
tld
Veteran
Joined: 09 Dec 2003
Posts: 1816
|
| Posted: Sat Aug 09, 2008 8:50 pm Post subject: |
|
|
I'll bet that mplayer isn't doing any de-interlacing and myth is. That'd make a huge difference. 1080i definitely requires de-interlacing (unfortunately even for those of us with 1080i native displays...thanks for nothing nVidia). The de-interlacing method your mythtv profile is using will make a big differnce too.
One way to tell is to compare mythtv to mplayer with a 720p recorded show (which mythtv won't be deinterlacing).
Tom |
|
| Back to top |
|
|
tld
Veteran
Joined: 09 Dec 2003
Posts: 1816
|
| Posted: Sat Aug 09, 2008 8:55 pm Post subject: |
|
|
I have a 3 Ghz P4 with hyper-threading enabled. I'm using bob de-interlacing. Here's what top looks like playing a 1080i show:
| Code: | top - 16:51:42 up 5 days, 3:14, 1 user, load average: 0.48, 0.11, 0.07
Tasks: 58 total, 1 running, 57 sleeping, 0 stopped, 0 zombie
Cpu(s): 24.5%us, 0.7%sy, 0.0%ni, 73.5%id, 0.0%wa, 0.7%hi, 0.7%si, 0.0%st
Mem: 514112k total, 508228k used, 5884k free, 3108k buffers
Swap: 987988k total, 78040k used, 909948k free, 220692k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4153 mythtv 20 0 429m 242m 120m S 52 48.2 339:50.38 mythfrontend
4148 root 20 0 265m 221m 101m S 9 44.1 107:09.18 X
25811 tom 20 0 2228 1080 840 R 0 0.2 0:00.02 top
1 root 20 0 1588 544 468 S 0 0.1 0:03.08 init |
It looks like mine is using around 50% of one "cpu" (the fake hyper-threaded kind).
Also note that that's without XvMC. I'm using libmpeg2. The CPU usage is a little higher if I use ffmpeg as a decoder.
Tom |
|
| Back to top |
|
|
mrpringle
Apprentice
Joined: 27 May 2007
Posts: 284
|
| Posted: Sun Aug 10, 2008 1:12 am Post subject: |
|
|
| tld wrote: | | I'll bet that mplayer isn't doing any de-interlacing and myth is. That'd make a huge difference. 1080i definitely requires de-interlacing (unfortunately even for those of us with 1080i native displays...thanks for nothing nVidia). The de-interlacing method your mythtv profile is using will make a big differnce too. |
To be honest I can't really notice any difference when I play back recorded TV using mplayer compared with the internal player, even for 1080 content. I also turned off deinterlacing in the mythtv options because I didn't really notice much difference and the linear deinterlacing method was using about 10% more CPU.
| Code: |
top - 16:51:42 up 5 days, 3:14, 1 user, load average: 0.48, 0.11, 0.07
Tasks: 58 total, 1 running, 57 sleeping, 0 stopped, 0 zombie
Cpu(s): 24.5%us, 0.7%sy, 0.0%ni, 73.5%id, 0.0%wa, 0.7%hi, 0.7%si, 0.0%st
Mem: 514112k total, 508228k used, 5884k free, 3108k buffers
Swap: 987988k total, 78040k used, 909948k free, 220692k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
4153 mythtv 20 0 429m 242m 120m S 52 48.2 339:50.38 mythfrontend
4148 root 20 0 265m 221m 101m S 9 44.1 107:09.18 X
25811 tom 20 0 2228 1080 840 R 0 0.2 0:00.02 top
1 root 20 0 1588 544 468 S 0 0.1 0:03.08 init
|
I suppose my CPU usage is probably quite similar in comparison (although I use ffmpeg playback and no de-interlacing) because I get about 50% CPU usage on a single core when playing back 1080 content.
| Quote: | | Also note that that's without XvMC |
I've read xvmc is supposed to reduce the cpu load during playback, but when I use xmvc it hasn't made any significant difference to CPU utilisation. Why is this? |
|
| Back to top |
|
|
tld
Veteran
Joined: 09 Dec 2003
Posts: 1816
|
| Posted: Sun Aug 10, 2008 3:35 pm Post subject: |
|
|
[quote="mrpringle"] | tld wrote: |
| Quote: | | Also note that that's without XvMC |
I've read xvmc is supposed to reduce the cpu load during playback, but when I use xmvc it hasn't made any significant difference to CPU utilisation. Why is this? |
Good question. My frontend has a GeForce 7100GS and XvMC gives me no big improvement either. I've heard others report the same thing. I have no idea why. Just as well actually. For one thing I don't want to count on nVidia for anything I don't have to, and for another, I hate being stuck with a black and white OSD in mythtv.
Yet on an older machine I have with an old GeForce4 MX 420 and the old drivers, XvMC can make the difference between not being able to play a file at all and having the CPU doing almost nothing.
Tom |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Multimedia
