[MPlayer-dev-eng] [PATCH] fix for -srate bug

Ed Wildgoose lists at wildgooses.com
Wed Oct 20 10:50:51 CEST 2004


Jan Knutar wrote:

>On Tuesday 19 October 2004 19:39, Ed Wildgoose wrote:
>  
>
>>I dont see that I need that much 
>>CPU for video to be honest though?
>>    
>>
>
>Well most other people don't see why they need that much CPU for audio ;-)
>
>A nice chain such as spp,scale,unsharp,noise eats alot of CPU and having
>the perceptive video quality reduced a few magnitudes at the cost of a magnitude
>of CPU usage, and more, in the audio scaler for a marginal, mostly unnoticeable
>audio quality increase, is not acceptable for the majority of people, which is why
>Rich is upset :-)
>  
>

Well, true, but I have never suggested that this be the default for 
everyone.  So I *don't* see why he is upset?

In any case.  Some tests "seem" to suggest that we are talking about 
1-2% vs 2-4% cpu.  I don't think this is enough for anyone to get 
excited about, especially since it would be an option only for those who 
want it.

As for how many people would notice.  I think we are in a chicken and 
egg situation.  If you drop by a few of the amateur home theatre forums 
you will see a large change from people buying $10,000 scalers and some 
move from external audio processors to a PC based system (which is as 
good, but for less money).  Inevitably this change is mostly to a 
windows based system because, well likely it is what they know...  But I 
think also because although linux is very close to being an option, we 
don't *quite* yet have all the tools together to build a high end video 
system for general purpose use (please don't bite here, I'm not trying 
to create an argument on this point)

I'm sure loads of people would like to jump in and say that they have a 
nice video system working now... Yeah, well so do I.  It doesn't change 
the fact though that to really compete with $10K of commercial 
electronics (which we can nearly do on linux), we still have a few rough 
edges to polish off.  But I really think that we are very close!

So I would claim that the situation is that with the current resampler 
at 1-2% then mplayer does not compete with even a cheap outboard 
surround processor.  But with a higher quality one (which might double 
that CPU requirement?) then we probably have something which can compete 
with a high end surround processor.  This is my current situation, that 
I would like a really high end system and since my machine sits at <20% 
CPU playing videos I see no reason not to turn on all the quality knobs 
to see what I can get it to do (no point having it sit idle just to show 
a really low load meter?)

Anyway, I have no intention of spending that kind of cash on consumer 
electronics!  However, I would like to get that kind of quality.... 

Does that make sense?




More information about the MPlayer-dev-eng mailing list