[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: profile-directed optimizations
From: |
Paolo Bonzini |
Subject: |
Re: profile-directed optimizations |
Date: |
Sun, 21 Sep 2008 16:44:49 +0200 |
User-agent: |
Thunderbird 2.0.0.16 (Macintosh/20080707) |
Bruno Haible wrote:
> Paolo Bonzini wrote:
>>> But the compiler does not know that fstrcmp is called millions of time and
>>> that this piece of code needs to be optimized for speed rather than for
>>> space.
>> If doing profile-directed optimization, it does know.
>
> Well, it _could_ know. But currently, GCC exploits the profile of previous
> runs only for branch probabilities. It does not use it for inlining decisions.
> Says the GCC 4.3.0 doc.
My experience is different. In fact, in 4.4.0 it will even use it for
instruction selection in some cases.
Paolo
- Re: [PATCH] Implement premature termination of compareseq., (continued)
- Re: [PATCH] Implement premature termination of compareseq., Bruno Haible, 2008/09/14
- Message not available
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Ralf Wildenhues, 2008/09/15
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Bruno Haible, 2008/09/15
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Ralf Wildenhues, 2008/09/19
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Bruno Haible, 2008/09/20
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Ralf Wildenhues, 2008/09/20
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Paolo Bonzini, 2008/09/21
- Re: profile-directed optimization, Bruno Haible, 2008/09/21
- Re: profile-directed optimization, Paolo Bonzini, 2008/09/21
- Re: profile-directed optimizations, Bruno Haible, 2008/09/21
- Re: profile-directed optimizations,
Paolo Bonzini <=
- Re: profile-directed optimizations, Bruno Haible, 2008/09/21
- Re: msgmerge speedup: fstrcmp and diffseq improvements, Bruno Haible, 2008/09/22