[MPlayer-dev-eng] Re: Make -nortc the default?

[Corey Hickey]

Vladimir Mosgalin wrote:
>> I cannot detect any difference between -rtc and -nortc. My test 
>> methodology is to:
>>
>> 1. Use this file:
>> http://images.apple.com/movies/us/hd_gallery/gl1800/1080p/bbc-japan_1080p.mov
>>
>> 2. mplayer bbc-japan_1080p.mov -rtc
>> (observe that mplayer says my system is too slow at "A:  28.1")
> 
> Mplayer's cpu statistics doesn't include cpu wasted on rtc. Use top to
> measure difference.

Those apple h.264 files aren't very good for watching with top.
MPlayer's CPU usage is so variable that it's hard to see subtle trends.

>> 3. mplayer bbc-japan_1080p.mov -nortc
>> (observe exactly the same result)
> 
> It's a bit tricky. You see, when mplayer's cpu usage is close to 100%,
> looks like it doesn't spend much time using rtc, so the difference
> between -rtc and -nortc becomes marginal. In other words, -nortc won't
> help you in this case, and you won't notice the difference with
> -benchmark too. But with rtc mplayer eats 10-15% of cpu that could be
> given to other processes on your system.

Ok, I'm seeing a difference now. I tried playing a few less
cpu-intensive videos and watched top. I see a difference now, but not
that great a difference; only 2-4% of total CPU usage.

>> I also tried with -v to make sure mplayer was using the requested timing 
>> sources. It was indeed.
>>
>> Is there a better way to test this?
>>
>> Athlon64 3400+ (a bit overclocked)
> 
> I see big difference on similar cpu (A64 @ 2.6Ghz)

Out of curiosity, are you using HPET for the RTC?

$ grep HPET /path/to/your/kernel/.config

> ..though I have 32-bit system and 2.6.16.

If you want better performance, build a 64-bit mplayer. I haven't
benchmarked recently, but I do remember seeing a substantial performance
advantage; somewhat more of a difference than the RTC usage makes now.

-Corey