This is the mail archive of the gcc-patches@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: Speedup CSE by 5%


My results were similar concerning speedup to CSE, but I don't have
exact numbers anymore, but the macro hackery is notably ugly.  Perhaps
factoring out some of code into inline functions would make it easier to
deal with.

Yes, that's what my patch does (like the edge vector iterators and Zdenek's bmp_iter). My code is not very ugly; that means, as ugly as simulated recursion can be, but not more. :-)


The bootstrap I'm timing is with all usages of for_each_rtx converted to FOR_EACH_RTX except those in combine.c (only because I'm working on another combine.c patch) and loop.c (because I expect it to die in 4.1 at last).

I thought about a static stack, but alias.c's usages of for_each_rtx are very recursive. Maybe a free list will do the job.

Paolo


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