Re: [Patch, libfortran] Use high resolution clock if available, timing cleanup

[Janne Blomqvist <blomqvist dot janne at gmail dot com>]

On Mon, Jan 31, 2011 at 22:46, Steve Kargl
<sgk@troutmask.apl.washington.edu> wrote:
> On Mon, Jan 31, 2011 at 10:18:45PM +0200, Janne Blomqvist wrote:
>> 2011-01-31 ÂJanne Blomqvist Â<jb@gcc.gnu.org>
>>
>> Â Â Â * configure.ac: Check for clock_gettime().
>> Â Â Â * configure: Regenerated.
>> Â Â Â * config.h.in: Regenerated.
>> Â Â Â * intrinsics/time_1.h (__time_1): Rename to gf_cputime, add
>> Â Â Â times() fallback.
>> Â Â Â (gf_gettime): New function.
>> Â Â Â * intrinsics/cpu_time.c (__cpu_time_1): Update to call gf_cputime.
>> Â Â Â * intrinsics/date_and_time.c (date_and_time): Use gf_gettime.
>> Â Â Â * intrinsics/dtime.c (dtime_sub): Use gf_cputime.
>> Â Â Â * intrinsics/etime.c (etime_sub): Use gf_cputime.
>> Â Â Â * intrinsics/system_clock.c (system_clock_4): Use gf_gettime.
>> Â Â Â (system_clock_8): Use gf_gettime, increase count rate to allow
>> Â Â Â nanosecond precision, remove overflow prone branch.
>>
>
> OK.

Thanks for the quick review!

Sending        libgfortran/ChangeLog
Sending        libgfortran/config.h.in
Sending        libgfortran/configure
Sending        libgfortran/configure.ac
Sending        libgfortran/intrinsics/cpu_time.c
Sending        libgfortran/intrinsics/date_and_time.c
Sending        libgfortran/intrinsics/dtime.c
Sending        libgfortran/intrinsics/etime.c
Sending        libgfortran/intrinsics/system_clock.c
Sending        libgfortran/intrinsics/time_1.h
Transmitting file data ..........
Committed revision 169449.


> One comment. ÂWhy
>
> +AC_DEFINE([HAVE_FEENABLEEXCEPT],[1],[libm includes feenableexcept])])
>
> I did not see a use of fenv facilities in a quick scan of
> the patch.

That was just stuff surrounding the part my patch changes in configure.ac.

-- 
Janne Blomqvist