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: [C++ PATCH] Fix PR 45, 3784


Boy is this going to slow down compiles! :-( Not your fault, I'm just noting it. The solution is to canonicalize types, if only for cp_tree_equal!

Abraham's has a test case at:

http://users.rcn.com/abrahams/instantiation_speed/index.html

Compile with REPEATS=100 for fun. If you really slowed it down, you may want to start with 10 or 20. I think that should be able help characterize the slowdown, if you're interested.

If anyone wants to speed it up, before you start, I have patches that fix most of the problems, except for the canonicalization part. That part would be useful to work on. If not, I suspect I'll get a chance come back to this at some point.

Note - I didn't try this out on a compile with your change in it, but I have seen how often we come through this code with the testcase, I suspect a good sized hit on compile speed, but I could be wrong.

On Friday, January 3, 2003, at 09:22 AM, Nathan Sidwell wrote:
2002-12-26  Nathan Sidwell  <nathan@codesourcery.com>

	PR c++/45, c++/3784
	* tree.c (cp_tree_equal, TEMPLATE_PARM_INDEX): The types must be
	the same too.






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