This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH]: Cache immediate dominators
- From: David Edelsohn <dje at watson dot ibm dot com>
- To: Jan Hubicka <hubicka at kam dot mff dot cuni dot cz>
- Cc: Daniel Berlin <dberlin at dberlin dot org>, gcc-patches at gcc dot gnu dot org, mrs at apple dot com
- Date: Mon, 19 Aug 2002 13:15:56 -0400
- Subject: Re: [PATCH]: Cache immediate dominators
>>>>> Jan Hubicka writes:
Jan> I believe this is specific to SSA that is only so heavy user of the
Jan> function.
Jan> Concerning the implementation I was thinkig about slightly different
Jan> approach. Currently the et_forest is using nodes for basic blocks and
Jan> occurences for each edge to the father in the tree.
Jan> THere is exactly one occurence (or two one for the empty tree?) for each
Jan> non-root node representing the edge to the father. It should make sense
Jan> to put occurences into the nodes structure itself saving some memory
Jan> allocation overhead and making father lookup constant as well.
Yes, I do see a 10% compilation speed improvement with the patch
when I used -fssa. However, when SSA is not used, the compilation speed
is the same or slightly worse with the patch allocating the cache.
David