[MPlayer-dev-eng] Re: Compile options

[Andrew Savchenko]

Hi

> I repeated the test with -benchmark -nosound.
> Result: -O2 is faster as expected after the first test.
> Could anybody confirm this?
>
> Again multiple tries, again -O4 was never faster than any -O2 try.
>
> -O4:
> BENCHMARKs: VC:  58.629s VO:  18.218s A:   0.000s Sys:   2.617s =  
> 79.464s BENCHMARK%: VC: 73.7808% VO: 22.9261% A:  0.0000% Sys: 
> 3.2931% = 100.0000%
>
> BENCHMARKs: VC:  57.930s VO:  18.243s A:   0.000s Sys:   2.120s =  
> 78.293s BENCHMARK%: VC: 73.9912% VO: 23.3014% A:  0.0000% Sys: 
> 2.7074% = 100.0000%
>
> BENCHMARKs: VC:  58.757s VO:  18.281s A:   0.000s Sys:   2.003s =  
> 79.041s BENCHMARK%: VC: 74.3374% VO: 23.1289% A:  0.0000% Sys: 
> 2.5337% = 100.0000%
>
> -O2:
> BENCHMARKs: VC:  56.918s VO:  18.192s A:   0.000s Sys:   2.549s =  
> 77.659s BENCHMARK%: VC: 73.2924% VO: 23.4256% A:  0.0000% Sys: 
> 3.2820% = 100.0000%
>
> BENCHMARKs: VC:  56.920s VO:  18.465s A:   0.000s Sys:   2.565s =  
> 77.950s BENCHMARK%: VC: 73.0208% VO: 23.6888% A:  0.0000% Sys: 
> 3.2904% = 100.0000%
>
> BENCHMARKs: VC:  55.889s VO:  18.347s A:   0.000s Sys:   2.016s =  
> 76.252s BENCHMARK%: VC: 73.2949% VO: 24.0614% A:  0.0000% Sys: 
> 2.6437% = 100.0000%

Iteresting, for your system O2 is clearly faster than O4. But in mine 
system things are on the contrary. Maybe it is system-dependant.

It is rather strange that higher optimization level produces worse code 
on your machine. According to your earlier post, you are using gcc 
version 4.0.2 20050901 (prerelease), maybe you should upgrade to 
release? It is seems to be a bug in your compiler. I'm using 
gcc-4.0.2-8.fc4 (as I've mentioned in one of previous letters), it was 
built 20051130.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 191 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20060917/d2035ea8/attachment.pgp>