VLC 2.0 has 20% higher CPU usage?

[JohnMcPayne]

I tried several video files and formats and noticed that the new version has a significant higher cpu usage than the previous version.
Can someone else look at this?
I had CPU usage around 35% with the old version, which was my main reason to use VLC because of its very low power consumption.
Now once a month I use to travel by train, it lasts 4h30min and till now, every time my battery was full enough for an additional hour. After I updated to 2.0 my Macbook Pro went into Sleep after 4 hours with an empty battery. So I tried to figure out why. I found out that now playing video files the CPU consumption has risen to about 55%.

If this is the case, then my question to the developers:
Why?

Please try to keep power consumption as low as possible.

[Jean-Baptiste Kempf]

Can you try to delete libsamplerate_plugin.dylib and check if this is better?

[JohnMcPayne]

Hey!

I deleted the file and this solved the problem. Now I have 35-40% CPU usage. Thank you.

[Jean-Baptiste Kempf]

Please, can you try deleting libspeex_dsp_plugin.dylib too to see if the CPU gets lower?

[JohnMcPayne]

There is no such file:

libspeex_plugin.dylib and libspeex_resampler_plugin.dylib, but no such with dsp.

[Jean-Baptiste Kempf]

libspeex_resampler_plugin.dylib one and please report.

[JohnMcPayne]

It jumps around, between 30% and 40%. It seems to stay around 30% more often than before. What are those plugins for? Is this a bug?

[Jean-Baptiste Kempf]

So, this is better, right?

[JohnMcPayne]

Yes, right. Deleting the first file helped the most -> about 10% fewer CPU usage.
Deleting the second file helped a little -> about 3-5% fewer CPU usage.

Deleting both I am back at the average CPU usage of the previous version of VLC.

[Jean-Baptiste Kempf]

It should be lower than previous version, in fact.

[RTouris]

I also noticed such a behaviour from v. 2.0...I deleted all the old prefs via the script provided in the image and launched the new version / compared on similar files...On a MacBookPro 2,1 (2.33C2D / Radeon X1600) CPU usage went from 15% on v. 1.2.12 up to 30-35% on v. 2.0...Will try the resolution above and report back asap.

[Jean-Baptiste Kempf]

Try to remove libsamplerate_plugin.dylib and libspeex_resampler_plugin.dylib too.

[JohnMcPayne]

Can you recommend something I could try? Maybe if I completely remove VLC and reinstall new.

Should I only drag the App into the trash bin or are there any hidden pref files I should delete by hand?

[Jean-Baptiste Kempf]

Do what I said. Remove those 2 modules and check again.

[misterpieter]

i have a simular problem: my macbook is much slower with the new VLC
but I cant seem to find the 'libsamplerate_plugin.dylib' on the mac i even searched in the hidden files...
where is it located?

[kdean]

Right-click on VLC application and select "Show Package Contents", then go to Contents/MacOS/plugins

[Jean-Baptiste Kempf]

Ok, another great solution seems to be to use the linear profile and not the main one for the resampler.

[JohnMcPayne]

@Jean Baptiste:
As I mentioned above I already deleted both files, but I have slightly higher CPU usage as with the previous version. I will try to use the linear profile.

Do I get a lesser quality with this profile?

[Jean-Baptiste Kempf]

It should be better quality than without those 2 dylibs.

[JohnMcPayne]

Yea, this helped. Thank you!
Now I have under 30% CPU usage most of the time. So the CPU usage change between VLC 1.xx and 2.0 is just a matter of sound quality, is that right?

[Jean-Baptiste Kempf]

Yea, this helped. Thank you!
Now I have under 30% CPU usage most of the time. So the CPU usage change between VLC 1.xx and 2.0 is just a matter of sound quality, is that right?

Yes.
Can you confirm the results with SRC and linear? and Zero Order Hold?

[JohnMcPayne]

Yes, I can confirm that.

With Zero Order Hold I get the best results which vary between 25-30% load.

[Jean-Baptiste Kempf]

That would be even lower than 1.1.11 ?

[JohnMcPayne]

Yes.
I had about 30% with 1.x.x.

[Jean-Baptiste Kempf]

Good. We seem to have a solution then...