[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Bug#102193: gcc-3.0: compiled code with gcc 3.0 is slow and big :-(



On Mon, 25 Jun 2001, Matthias Klose wrote:

> funny report. what do you expect? please provide the relevant source
> as documented in /usr/share/doc/gcc-3.0/README.Bugs.gz.
> 
>  > I'm Gabor Lenart from Hungary and we're developing a movie player software
>  > for Linux. We tried to compile and benchmark mplayer (our player) with both
>  > of gcc 3.0 and gcc 2.95. We found out that gcc 3.0 produces SLOWER and BIGGER
>  > code than 2.95. Is it normal? If it is, I will refuse to use gcc 3.0 since
>  > I need maximum speed, since on my AMD K6-2 DVD playback is a very CPU
>  > sensitive situation.

FYI, the K6 support in gcc suffered a tad from bitrot, so it's not
nearly as good as it could be (which could explain poor performance
if you're compiling your code with the K6 scheduling/optimisation
options).  Work is being done on it now, though, to improve things 
hopefully for 3.0.1.  Athlon support is pretty good, though, if that's any
consolation.

Other than that, I agree with Matthias...we would need the code to really
determine if that's the true cause or not.

C



Reply to: