http://gcc.gnu.org/bugzilla/show_bug.cgi?id=49498
--- Comment #10 from Jeffrey A. Law <law at redhat dot com> 2011-07-07 21:26:57 UTC --- -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On 07/07/11 13:32, xinliangli at gmail dot com wrote: > Yes. Jump threading is inherently an iterative process. It was decided > some time ago to remove the iterative step as it doesn't buy a whole lot > in terms of code quality and it's a fairly significant compile-time cost. > > >> Does it need to be iterative? Any simple example to show dependency of >> jump-threading analysis on actual transformation of some other >> jump-threading instance (e.g, opportunity only gets exposed after >> transformation)? > > > > During my investigations noticed the same thing and manually reran DOM > from within GDB and verified it further simplified the CFG, but > rerunning DOM isn't really going to be an option. > > >> Is incremental DOM update a choice? I looked at it a few years ago. I never came up with anything even remotely close to working, even at a conceptual level. While the PHI nodes represent places where interesting things happen, I wasn't able to make any leap that would allow me to use some kind of worklist algorithm to pick up the secondary optimization opportunities. During the course of those investigations I came across Bodik's work, and only recently did the light come on with how Bodik's work might translate well into how we handle jump threading. I've been drawn off that work temporarily, but expect to return to it shortly. Jeff -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.11 (GNU/Linux) Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/ iQEcBAEBAgAGBQJOFiSHAAoJEBRtltQi2kC7PhQH/A7NwXEghs+FUlszsb7rnB9p lZkz1hrMSaVTPgMS2ewOwp6nedpK3pshOFVJHFcLxw1SjOjNePVYccH9XRXSwqSL wiuiLYTv+SfObC7icvTkQuJ0N8eJ6apxo5hBvb3d599DifwAaZHaq/3RoUslcH2O sNSfoQbJGiDSfw3HxAK36IgxRvx3Khe/1GoEMV1W40kiD9tB6wgH5YyAq1Iq6ZHR Ckgqjy/fR8IZtTruSNNCA76erYXOSUc2etQoassYLGvvwsG/htqA1c3D86TQFIfP 0K55jDWMpdOmuxEp0JrUEE0g6jVuOXHY2DPIWAt/YDY2YJwTflYSEjd8wYYqYPM= =/taX -----END PGP SIGNATURE-----
