On Mon, 24 Sep 2007 00:40:32 -0700
Amit Uttamchandani <[EMAIL PROTECTED]> wrote:

> > hello ,
> > 
> > i wonder why some of packages compiled from source are faster than 
> > precompiled debs.
> > 
> > for example - why the ruby interpreter in all tested version (look here 
> > ,please   http://www.ruby-forum.com/topic/125344 ) 1.8.5,1.8.6, 1.9 is 
> > much more faster
> > 
> > when is compiled from source ? because of optimization for universal 
> > architecture in debs ?
> > 
> > 
> 
> Packages compiled from source are always going to be faster than their binary 
> counterparts. When you compile software from source there are a lot factors 
> that are involved, most importantly the compiler on your machine and your 
> architecture. For example, if you have the latest version on the gcc 
> compiler, your software will perform faster because of optimizations built in 
> to the current version of the compiler that you have. If I'm not mistaken 
> some packages in debian or other distros still use gcc ver 3 because it is 
> much more tested and stable. In quite a few cases, byte code from gcc 4 will 
> be faster than byte code from gcc 3.
> 
> I hope there was no error in what I was trying to say.

I don't know anything about Ruby, but the questions of why there aren't
more processor specific packages and whether it pays to compile
packages locally comes up often.  The conventional wisdom seems to be
that the kernel, libc, and various other compilation sensitive
packages, such as mplayer, are indeed available in multiple, optimized
flavors, and that for most other packages there isn't much to be gained
by tweaking the compilation.  The OP is asserting that he does see
major benefits from building his own Ruby and certain other packages;
I haven't experimented in this area.

> Amit Uttamchandani <[EMAIL PROTECTED]>

Celejar
--
mailmin.sourceforge.net - remote access via secure (OpenPGP) email
ssuds.sourceforge.net - A Simple Sudoku Solver and Generator


-- 
To UNSUBSCRIBE, email to [EMAIL PROTECTED] 
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]

Reply via email to