This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: Serious performance regression -- some tree optimizer questions

From: Daniel Berlin <dberlin at dberlin dot org>
To: Ulrich Weigand <uweigand at de dot ibm dot com>
Cc: gcc at gcc dot gnu dot org
Date: Fri, 17 Dec 2004 16:29:37 -0500 (EST)
Subject: Re: Serious performance regression -- some tree optimizer questions
References: <200412172124.iBHLOwF3010084@53v30g16.boeblingen.de.ibm.com>

there appear to be some optimizations that would appear quite
straightforward to me that no pass is currently performing.
For example, the gimple code contains sequences of the form:
s1 = a + b
s2 = s1 + c
t1 = a + c
t2 = t1 + b
but no pass recognizes that s2 == t2 ...

Uh, both of the value numbering passes (DOM and PRE) should recognize this.

Any suggestions how to further investigate this?  Is this just
the way things are with the current tree optimizers, or is this
supposed to work and we just need to find the bug?

File bug reports with code we can compile.

Also, can you send me the dump file for -fdump-tree-pre-vops-details-blocks-stats?

I'd like to see these code sequences you don't think it is optimizing.

References:
- Serious performance regression -- some tree optimizer questions
  - From: Ulrich Weigand

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]