This is the mail archive of the
mailing list for the GCC project.
Re: genautomata progress bars tweak
- From: law at redhat dot com
- To: Vladimir Makarov <vmakarov at redhat dot com>
- Cc: Andreas Jaeger <aj at suse dot de>, Segher Boessenkool <segher at kernel dot crashing dot org>, Zack Weinberg <zack at codesourcery dot com>, gcc-patches at gcc dot gnu dot org
- Date: Mon, 12 Jan 2004 12:09:20 -0700
- Subject: Re: genautomata progress bars tweak
- Reply-to: law at redhat dot com
In message <4002E8E7.36E1DE69@redhat.com>, Vladimir Makarov writes:
>> I wouldn't object if the progress bar was off by default, but could be
>> turned on by a switch or magic in the automata -- similarly for the
>> automata state information (which as I said it useful sometimes).
>> >It was a requirement from people who wrote DFA description.
>> Understood -- though that was probably due in part to the earlier versions
>> of genautomata being rather slow and the developers themselves not being
>> aware of the kinds constructs which would cause the automata to blow up.
>It will be an option which can be placed in a .md file or/and given on
>genattrtab command line (it is a standard genautomata mechanism).
>> genautomata is quite a bit faster now :-) Which reminds me, didn't Zdenek
>> have a change to speed up genautomata further (or was that genattrtab?)
>There were some patches to speed up genautomata.c. But even before this
>the most of time for i386 was spent in other parts of genattrtab. So I
>think this impression are mostly from Zdenek's work.
Well, at least when I was working on the PPro definition, we burned a ton
of time in genautomata.c for the prototype PPro defintion.
Given the recent interest in the PPro DFA bits, it might be worth seeing
if the proposed changes cut down the time in genautomata.