This is the mail archive of the
fortran@gcc.gnu.org
mailing list for the GNU Fortran project.
Re: [Patch fortran] PR41113 and PR41117 - Unnecessary invocations of internal_pack
- From: Paul Richard Thomas <paul dot richard dot thomas at gmail dot com>
- To: Steve Kargl <sgk at troutmask dot apl dot washington dot edu>
- Cc: VandeVondele Joost <vondele at pci dot uzh dot ch>, fortran at gcc dot gnu dot org
- Date: Sun, 3 Jan 2010 17:20:21 +0000
- Subject: Re: [Patch fortran] PR41113 and PR41117 - Unnecessary invocations of internal_pack
- References: <Pine.A41.4.63.0912311211410.557240@idaix01.uzh.ch> <20091231221440.GB58634@troutmask.apl.washington.edu>
Steve and Joost,
I will put right the problems that Dominique found and will resubmit
at the end of the week. I am in Tenerife until the 7th with access
via a gamers` shop. Boy, some strange people come in here....
I too would be very interested to know the effect on timing.
Whilst here, I have been removing the bit-rot from the memory checking
patch so that I can use it to fix allocatable component memory leaks.
I think that I am there now but I am impressed/appalled by how much
gfortran has changed in one year.
Happy New Year to you all
Paul
On Thu, Dec 31, 2009 at 10:14 PM, Steve Kargl
<sgk@troutmask.apl.washington.edu> wrote:
> On Thu, Dec 31, 2009 at 12:16:37PM +0100, VandeVondele Joost wrote:
>>
>> I've tested the Paul's patch compiling CP2K, and it has a very significant
>> effect on the number of calls to
>> _gfortran_internal_pack/_gfortran_internal_unpack in the *orginal dumps :
>>
>> unpatched: 4252
>> ? patched: 1276
>>
>> If the patch is correct, it seems worthwhile.
>>
>> With best wishes for the New Year,
>>
>> Joost
>
> Thanks for testing. ?Given the complexity of cp2k, I suspected
> that it would be a good test. ?Did you happen run your benchmark
> and compare unpatched to patched results?
>
> Happy New Years
>
> --
> Steve
>
--
The knack of flying is learning how to throw yourself at the ground and miss.
--Hitchhikers Guide to the Galaxy