This is the mail archive of the
mailing list for the GCC project.
Re: IRA improvements 2/4
- From: Vladimir Makarov <vmakarov at redhat dot com>
- To: Jeff Law <law at redhat dot com>
- Cc: gcc-patches <gcc-patches at gcc dot gnu dot org>
- Date: Wed, 07 Jul 2010 11:51:54 -0400
- Subject: Re: IRA improvements 2/4
- References: <4C3481AF.email@example.com> <4C34A0F4.firstname.lastname@example.org>
Jeff Law wrote:
On 07/07/10 07:31, Vladimir Makarov wrote:
The following patch speeds up some IRA code by compressing live
ranges even more. The idea that if we have several subsequent deaths
(or births) and there are no births (or deaths) between them, we can
use one point for all these deaths (or births).
Is the patch ok to commit the patch to the trunk?
2010-07-07 Vladimir Makarov <email@example.com>
* ira-live.c: Include sbitmap.h.
(remove_some_program_points_and_update_live_ranges): Use sbitmaps.
Compress live ranges even more.
Presumably you switched to use sbitmaps because typical map density of
points where something is either born or dies makes them more
efficient than bitmaps?
Right, they are very dense. Therefore I use sbitmaps.