From mboxrd@z Thu Jan  1 00:00:00 1970
From: Jan Hubicka <jh@suse.cz>
To: Brad Lucier <lucier@math.purdue.edu>
Cc: jh@suse.cz, gcc@gcc.gnu.org
Subject: Re: Timing information for CFG manipulations
Date: Sun, 14 Oct 2001 01:18:00 -0000
Message-id: <20011014101656.A1749@atrey.karlin.mff.cuni.cz>
References: <200110140332.f9E3W6I16651@banach.math.purdue.edu>
X-SW-Source: 2001-10/msg00800.html

Thanks,
>  cfg construction      : 269.67 (34%) usr   0.32 ( 0%) sys 270.02 (26%) wall

This is new problem. I guess it is related to the CFG freeing code...

>  flow 2                : 267.66 (34%) usr 251.44 (98%) sys 519.08 (49%) wall

I will care this one next week (I am at vacation right now).

>  44.47    348.50   348.50  7450804     0.05     0.05  remove_edge

This probably makes clear where we spend all the time.  We are now deallocating
each edge in the CFG before rebuild and this may cause problems on walking the
lists.

I will need to add function like "free outgiong edges"...

>  32.36    602.16   253.66     9556    26.54    26.54  sbitmap_vector_alloc
Do you have any idea, in what state is the ebitmap stuff?

Thanks for the infromation!
Honza