This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: [Bug optimization/13629] [tree-ssa][Regression] dominator optimizations can pessimize code
- From: Steven Bosscher <stevenb at suse dot de>
- To: Daniel Berlin <dberlin at dberlin dot org>, Jeff Law <law at redhat dot com>
- Cc: "gcc at gcc dot gnu dot org list" <gcc at gcc dot gnu dot org>
- Date: Fri, 16 Jan 2004 10:27:27 +0100
- Subject: Re: [Bug optimization/13629] [tree-ssa][Regression] dominator optimizations can pessimize code
- Organization: SUSE Labs
- References: <20040109121229.13629.steven@gcc.gnu.org> <20040116023724.16128.qmail@sources.redhat.com> <244E3AA4-47D8-11D8-AECF-000A95DA505C@dberlin.org>
On Friday 16 January 2004 04:57, Daniel Berlin wrote:
> > Long term we may want to handle this with partial dead code elimination
> > since real code often has partial dead code which we would like to be
> > able to move to eliminate it on paths where it's not needed.
>
> Errr, PDCE is O(n^3).
No, only O(N^2) with the new value graph algorithm. Not that I like
algorithms with quadratic behavior... ;-)
> Or are you referring to partial dead store elimination (Like the IA64
> compiler), which is usually based on some sort of PRE framework.
Dan, would partial dead store elimination help for the test case in the
PR audit log?
Gr.
Steven