This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [patch, libfortran] Speed up / orthogonalize in_pack and in_unpack


On Wed, 2008-03-19 at 10:11 +0000, FX wrote:

> OK

Committed, together with Dominique's correction of a typo as noted
in the PR.  Thanks!

For the record, here is an indication of the execution speed advantage
for real(10) on an i686-pc-linux-gnu:

$ cat foo.f90
program main
  real(kind=10) a(10000)
  a = 100.
  do i=1,10000
    call foo(a(1:10000:2))
  end do
end program main

subroutine foo(a)
  real(kind=10) a(5000)
  a(1) = 1.
  a(499) = 1.
end subroutine foo

$ gfortran-4.3 -O3 -static foo.f90
$ time ./a.out

real	0m3.654s
user	0m3.644s
sys	0m0.000s
$ gfortran -O3 -static foo.f90
$ time ./a.out

real	0m0.719s
user	0m0.700s
sys	0m0.000s



Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]