Merge pull request #1238 from davydden/pkg/mkl_update

mkl: add provide blas/lapack

1 files changed, 37 insertions, 1 deletions

diff --git a/var/spack/repos/builtin/packages/mkl/package.py b/var/spack/repos/builtin/packages/mkl/package.py
index 454e78d29c..6ea64f5313 100644
--- a/var/spack/repos/builtin/packages/mkl/package.py
+++ b/var/spack/repos/builtin/packages/mkl/package.py
@@ -9,7 +9,13 @@ class Mkl(IntelInstaller):
 
     Note: You will have to add the download file to a
     mirror so that Spack can find it. For instructions on how to set up a
-    mirror, see http://software.llnl.gov/spack/mirrors.html"""
+    mirror, see http://software.llnl.gov/spack/mirrors.html.
+
+    To set the threading layer at run time set MKL_THREADING_LAYER
+    variable to one of the following values: INTEL, SEQUENTIAL, PGI.
+    To set interface layer at run time, use set the MKL_INTERFACE_LAYER
+    variable to LP64 or ILP64.
+    """
 
     homepage = "https://software.intel.com/en-us/intel-mkl"
 
@@ -18,6 +24,11 @@ class Mkl(IntelInstaller):
     version('11.3.3.210', 'f72546df27f5ebb0941b5d21fd804e34',
             url="file://%s/l_mkl_11.3.3.210.tgz" % os.getcwd())
 
+    # virtual dependency
+    provides('blas')
+    provides('lapack')
+    # TODO: MKL also provides implementation of Scalapack.
+
     def install(self, spec, prefix):
 
         self.intel_prefix = os.path.join(prefix, "pkg")
@@ -26,3 +37,28 @@ class Mkl(IntelInstaller):
         mkl_dir = os.path.join(self.intel_prefix, "mkl")
         for f in os.listdir(mkl_dir):
             os.symlink(os.path.join(mkl_dir, f), os.path.join(self.prefix, f))
+
+    def setup_dependent_package(self, module, dspec):
+        # For now use Single Dynamic Library:
+        # To set the threading layer at run time, use the
+        # mkl_set_threading_layer function or set MKL_THREADING_LAYER
+        # variable to one of the following values: INTEL, SEQUENTIAL, PGI.
+        # To set interface layer at run time, use the mkl_set_interface_layer
+        # function or set the MKL_INTERFACE_LAYER variable to LP64 or ILP64.
+
+        # Otherwise one would need to specify several libraries
+        # (e.g. mkl_intel_lp64;mkl_sequential;mkl_core), which reflect
+        # different interface and threading layers.
+
+        name = 'libmkl_rt.%s' % dso_suffix
+        libdir = find_library_path(name, self.prefix.lib64, self.prefix.lib)
+
+        self.spec.blas_shared_lib   = join_path(libdir, name)
+        self.spec.lapack_shared_lib = self.spec.blas_shared_lib
+
+    def setup_dependent_environment(self, spack_env, run_env, dependent_spec):
+        # set up MKLROOT for everyone using MKL package
+        spack_env.set('MKLROOT', self.prefix)
+
+    def setup_environment(self, spack_env, env):
+        env.set('MKLROOT', self.prefix)