[MPlayer-dev-eng] Re: Compile options

[Michael Niedermayer]

Hi

On Sun, Sep 17, 2006 at 04:02:09PM +0000, Carl Eugen Hoyos wrote:
> Andrew Savchenko <Bircoph <at> list.ru> writes:
> 
> [...]
> 
> > It is rather strange that higher optimization level produces worse code 
> > on your machine.
> 
> AFAICR this was always true for gcc3.x and -O3.
> 
> > According to your earlier post, you are using gcc 
> > version 4.0.2 20050901 (prerelease), maybe you should upgrade to 
> > release?
> 
> Thanks for suggesting, the results are even more convincing (for my personal
> future use) with gcc 4.0.3 release:
> 
> -O4:
> BENCHMARKs: VC:  58.442s VO:  18.009s A:   0.000s Sys:   2.152s =   78.603s
> BENCHMARKs: VC:  58.477s VO:  18.052s A:   0.000s Sys:   2.076s =   78.605s
> BENCHMARKs: VC:  58.487s VO:  18.120s A:   0.000s Sys:   2.076s =   78.683s
> 
> -O2:
> BENCHMARKs: VC:  55.917s VO:  18.127s A:   0.000s Sys:   2.442s =   76.487s
> BENCHMARKs: VC:  56.172s VO:  18.253s A:   0.000s Sys:   2.032s =   76.458s
> BENCHMARKs: VC:  55.869s VO:  18.192s A:   0.000s Sys:   2.001s =   76.062s

this is getting quite interresting, id suggest the following if theres
some interrest in figuring out where these differences come from
* check that _all_ compiler flags are equal between both testers
* use the same video and same svn version of mplayer and libavcodec
  ideally both of you should get a fresh checkout of mplayer&lavc
* check that the compiler versions are identical
* check that no shared lavc is used
* both should get a official clean gcc release and comile them yourself
  to avoid possible messup by the distributions gcc maintainer

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

In the past you could go to a library and read, borrow or copy any book
Today you'd get arrested for mere telling someone where the library is