Bad compile time complexity for large files ??? (fwd)

[Mike Stump]

On Nov 10, 2004, at 11:14 AM, Gabriel Dos Reis wrote:
> Date: Tue, 9 Nov 2004 11:36:29 +0100 (CET)
> From: Joris van der Hoeven <texmacs@math.u-psud.fr>
> To: gcc@gcc.gnu.org
> Subject: Bad compile time complexity for large files ???

> When the number of routines increases, I noticed a far more than
> linear time complexity for the compilation time.

If you are serious about this, abstract down the glue code, plot a 
graph of size v compile time, and see if it is bad (n^2 2^n...).  If it 
is submit it as a bug report.

By abstract down, I mean, take the full code, and keep trimming out 
elements such that the compile times stay bad.

Another approach might be to submit a profile of the compiler as it 
compiles your code.

Be sure to use a top of the tree 4.0.0 compiler, as there have been a 
lot of work done recently to improve the compiler in this regard.

I'll give you an example of known slow things.  Template instantiations 
are (were) done by a linear search, causing n^2 behavior.  Not sure if 
this impacts you, as you abstracted out too much detail for me to tell.

Another possibility, try -O0.

A -ftime-report might also be useful in figuring out what is going 
wrong.

And most of all, be sure that you have an appropriate amount of memory 
for the task at hand.  The garbage collector will run often, if you 
have a `small' machine.

And lastly, you'll want to get a yahoo or gmail account, if you can't 
manage to sort out your ISP problems.  See 
http://gcc.gnu.org/lists.html for some details...