Philip Reames 6bd16b5850
[SimplifyCFG] Fix a cost modeling oversight in branch commoning ...
The cost modeling was not accounting for the fact we were duplicating the instruction once per predecessor. With a default threshold of 1, this meant we were actually creating #pred copies.
Adding to the fun, there is *absolutely no* test coverage for this. Simply bailing for more than one predecessor passes all checked in tests.
llvm-svn: 341001
2018-08-30 00:03:02 +00:00
2017-10-28 18:43:07 +00:00
2017-11-22 20:43:00 +00:00
2017-11-13 01:47:52 +00:00
2018-04-26 13:37:00 +00:00
2017-12-28 17:02:34 +00:00
2018-08-28 01:18:29 +00:00
2018-05-09 02:40:45 +00:00
2018-08-30 00:03:02 +00:00
2017-10-28 18:43:07 +00:00
2018-01-19 17:13:12 +00:00
2018-06-18 23:20:08 +00:00
2017-09-12 23:28:11 +00:00
2018-05-18 08:52:15 +00:00
2018-05-08 23:28:15 +00:00
2017-10-28 18:43:07 +00:00
2017-09-01 19:54:08 +00:00
2017-10-31 19:03:49 +00:00
2018-05-09 02:40:45 +00:00
2017-11-13 01:47:52 +00:00
2018-07-09 22:27:23 +00:00
2017-11-03 21:08:13 +00:00
2018-08-12 18:32:53 +00:00
2017-10-31 19:03:49 +00:00
2017-12-09 00:21:41 +00:00
2017-12-14 22:05:20 +00:00
2018-03-22 17:07:51 +00:00
2018-07-09 22:27:23 +00:00
2017-08-09 18:56:26 +00:00
2017-12-31 16:47:16 +00:00
2018-05-09 02:40:45 +00:00
2017-10-28 18:43:07 +00:00
2017-08-29 20:06:24 +00:00
2017-10-28 18:43:07 +00:00
2018-05-09 02:40:45 +00:00
2018-04-12 20:58:24 +00:00
2017-10-31 19:03:49 +00:00
2017-12-14 22:05:20 +00:00
2018-05-09 02:40:45 +00:00
2017-10-31 19:03:49 +00:00
2017-11-09 23:17:52 +00:00
2018-02-14 06:58:08 +00:00
2017-12-05 14:14:00 +00:00
2017-11-13 01:47:52 +00:00
2018-07-09 22:27:23 +00:00
2018-07-09 22:27:23 +00:00
2018-02-08 07:16:29 +00:00
2018-07-09 22:27:23 +00:00
6bd16b5850