This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: some profiling numbers
- From: Mike Stump <mrs at apple dot com>
- To: Steven Bosscher <s dot bosscher at student dot tudelft dot nl>
- Cc: graydon hoare <graydon at redhat dot com>, gcc at gcc dot gnu dot org
- Date: Wed, 25 Jun 2003 19:45:45 -0700
- Subject: Re: some profiling numbers
On Wednesday, June 25, 2003, at 04:47 PM, Steven Bosscher wrote:
One thing you point out is that the slowdown is more or less a result
of GCC just getting bigger. Does that mean that the extra passes over
the whole chain of insns or trees may be blamed for much of the
slowdown? Is there any way to measure how many times different GCC
versions walks large chains of insns/trees, and where?
Yes. Devang posted a graphic that had this info in it. He
instrumented TREE_CODE and plotted address on the Y axis and time on
the X axis. You can watch the compiler stop through all of memory,
over and over and over again. Each time, you know it is pulling out of
DRAM. This is why passes cost 2% of the compile time, or so, it takes
that much time to hit everything.