Re: A new patch, good efficiency improvements.

For what it's worth the code I've been using to benchmark is here:
http://dev.int64.org/snips/tidybench.c


On 1/29/06, Geoff Air <[email protected]> wrote:
> Hi Cory,
>
> >>this means the TIDY_USE_MMAP only appears once in the
> >>current CVS code, in tidylib.c, around tidyDocParseFile(),
> >>and the static removed from initStreamIn() in
> >>streamio.c ...
> >
> >Keeping things in separate files sounds like a good idea.
>
> Attached to a separate email, since emails to the
> develop list can not have attachments, is a diff
> file, using a single new file fileioW32.c. If this is
> suitable to you, I hope you will put it up as a
> definitive change ...
>
> It seems the -N option of diff WILL include new files, but
> it has a strange way to show it is a NEW file ... it makes
> the date of the non-existent cvs file as date zero, like
> --- c:/FGCVS/tidy\src\fileioW32.c       Thu Jan 01 01:00:00 1970
>
> This diff patch also includes changes to tidy.dsp, the
> MSVC6 build file ... to make building with all versions
> of MSVC easy.
>
> >>     doctype = NewNode( doc->lexer );
> >>and this would fix this case.
> >
> >I was not aware of that.  Good catch.
>
> and it contains this 'fix' ... and I have only used
> GetFileSize(), but not quite as I originally errantly
> wrote it ;=()
>
> Also in pool.c, although you had used the allocation macro,
> MemAlloc, you had used 'free' ... now changed to MemFree ...
>
> >>As mentioned, just as a 'test' I have also built Tidy
> >>using MSVC6, thus have adjusted the DSW/DSP files
> >>accordingly ... I agree with using these as the basis for
> >>the WIN32/MSVC port, rather than the new SLN/VCPROJ files ...
> >
> >Just curious.. why?
>
> Just for backward compatibility ... you can use DSW/DSP with
> MSVC7, but can not use SLN/VCPROJ with MSVC6 ... and to add
> the SLN/VCPROJ files would mean modifying both for each, and
> every change ...
>
> It also includes tidytimer.[c|h] modules, so you can SEE the
> speed improvements ... but this would need to be 'fixed'
> for *nix platforms ... I do not know their high-performance
> timer function(s) ...
>
> It seems your #3 item, (void *) has already been changed in
> cvs ;=))
>
> I certainly agree your #1 (file mapping) should default
> to on in WIN32, as should the #2 (pool for Node allocation),
> and have made it so in the modified platform.h ...
>
> Everybody agrees we should keep the #ifdef in the code
> to the barest minimum, and putting this all in a separate
> module certainly does this ...
>
> Hope this helps ...
>
> Regards,
>
> Geoff.
>
> PS: Just to remind everyone the (WIN32) benefits of this
> patch - run on a 30MB xml file -
> Without patch
> Parse=9.5066, Out=7.3785, else=0.000066, total 16.89 secs.
> Parse=9.5062, Out=6.9299, else=0.000129, total 16.44 secs.
> Parse=9.4557, Out=6.7300, else=0.000063, total 16.19 secs.
> Parse=9.5349, Out=6.9790, else=0.000063, total 16.51 secs.
> Parse=9.6337, Out=6.8220, else=0.000064, total 16.46 secs.
> With patch
> Parse=5.6912, Out=7.4344, else=0.000061, total 13.13 secs.
> Parse=5.6227, Out=6.9251, else=0.000061, total 12.55 secs.
> Parse=5.6847, Out=6.6701, else=0.000082, total 12.35 secs.
> Parse=5.6951, Out=6.7353, else=0.000062, total 12.43 secs.
> Parse=5.6228, Out=6.9983, else=0.000070, total 12.62 secs.
>
> EOF - Tidy-21.doc
>
> _________________________________________________________________
> Express yourself instantly with MSN Messenger! Download today - it's FREE!
> http://messenger.msn.click-url.com/go/onm00200471ave/direct/01/
>
>


--
Cory Nelson
http://www.int64.org


-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems?  Stop!  Download the new AJAX search engine that makes
searching your log files as easy as surfing the  web.  DOWNLOAD SPLUNK!
http://sel.as-us.falkag.net/sel?cmd=lnk&kid3432&bid#0486&dat1642