Add PGI support to MKL

[ocaisa]

based on https://www.pgroup.com/support/link.htm#mkl

This relies on #1342 and easybuilders/easybuild-easyblocks#866

[hpcugentbot]

EasyBuild framework unit test suite FAILed.

See https://jenkins1.ugent.be/job/easybuild-framework-pr-builder/2856/console for more details.

Please fix the reported issues by pushing additional commits to the branch corresponding with this pull request; contact @boegel if you're not sure what to do.

[hpcugentbot]

EasyBuild framework unit test suite FAILed.

See https://jenkins1.ugent.be/job/easybuild-framework-pr-builder/2857/console for more details.

Please fix the reported issues by pushing additional commits to the branch corresponding with this pull request; contact @boegel if you're not sure what to do.

[hpcugentbot]

EasyBuild framework unit test suite FAILed.

See https://jenkins1.ugent.be/job/easybuild-framework-pr-builder/2858/console for more details.

Please fix the reported issues by pushing additional commits to the branch corresponding with this pull request; contact @boegel if you're not sure what to do.

[hpcugentbot]

EasyBuild framework unit test suite FAILed.

See https://jenkins1.ugent.be/job/easybuild-framework-pr-builder/2888/console for more details.

Please fix the reported issues by pushing additional commits to the branch corresponding with this pull request; contact @boegel if you're not sure what to do.

[boegel]

@ocaisa this looks wrong?

[boegel]

from the output of eb --from-pr 2900 HPL-2.1-pomkl-2016.03.eb -x:

  export BLAS_MT_STATIC_LIBS="libmkl_intel_lp64.a,libmkl_pgi_thread.a,libmkl_core.a"
  export BLAS_STATIC_LIBS="libmkl_intel_lp64.a,libmkl_sequential.a,libmkl_core.a"

the libmkl_intel_lp64.a should be libmkl_pgi_lp64.a instead?

However, there's no libmkl_pgi_lp64.a available...

[boegel]

based on https://software.intel.com/en-us/articles/intel-mkl-link-line-advisor, we should always link with libmkl_intel_lp64.a when using Intel or PGI compilers, and also for GCC C/C++; only with GCC Fortran libmkl_gf_lp64.a should be used instead

Also, libmkl_intel_thread.a should always be used, regardless of the compiler, even for GCC or PGI (which is not what we're doing now). I wonder why libmkl_pgi_thread.a and libmkl_gnu_thread.a are there at all then.

This is based on the advice provided for imkl 11.3.2 specifically...

[ocaisa]

But its not, didn't I give a reference to the pgi linker page?

On 2 May 2016 9:00 p.m., "Kenneth Hoste" <notifications@github.commailto:notifications@github.com> wrote:

In easybuild/toolchains/linalg/intelmkl.pyhttps://github.com//pull/1664#discussion_r61776195:

@@ -84,6 +85,7 @@ def _set_blas_variables(self):
interfacemap = {
TC_CONSTANT_INTELCOMP: 'intel',
TC_CONSTANT_GCC: 'gf',

•        TC_CONSTANT_PGI: 'intel',
  

@ocaisahttps://github.com/ocaisa this looks wrong?

—
You are receiving this because you were mentioned.
Reply to this email directly or view it on GitHubhttps://github.com//pull/1664/files/f88303f378701303559b06724b546944f8bf5a3f#r61776195

---

---

Forschungszentrum Juelich GmbH
52425 Juelich
Sitz der Gesellschaft: Juelich
Eingetragen im Handelsregister des Amtsgerichts Dueren Nr. HR B 3498
Vorsitzender des Aufsichtsrats: MinDir Dr. Karl Eugen Huthmacher
Geschaeftsfuehrung: Prof. Dr.-Ing. Wolfgang Marquardt (Vorsitzender),
Karsten Beneke (stellv. Vorsitzender), Prof. Dr.-Ing. Harald Bolt,

Prof. Dr. Sebastian M. Schmidt

---

[boegel]

please include https://www.pgroup.com/support/link.htm#mkl in a comment above this line?

it looks wrong at first sight, but you're right, it's correct

[hpcugentbot]

EasyBuild framework unit test suite PASSed (see https://jenkins1.ugent.be/job/easybuild-framework-pr-builder/3020/console for more details).

This pull request is now ready for review/testing.

Please try and find someone who can tackle this; contact @boegel if you're not sure what to do.

[boegel]

tested in combination with #1724 using HPL easyconfig from easybuilders/easybuild-easyconfigs#2900, works great, so going in, thanks @ocaisa!