[cig-commits] [commit] devel: added a comment about -DFORCE_VECTORIZATION (567cb26)
cig_noreply at geodynamics.org
cig_noreply at geodynamics.org
Tue Apr 22 16:37:47 PDT 2014
Repository : ssh://geoshell/specfem3d_globe
On branch : devel
Link : https://github.com/geodynamics/specfem3d_globe/compare/9f9faf71ddb9dd81afedf5968b94ef1b47757f75...01a32b6d397bdd20adfe526f8bbd9b282ba43542
>---------------------------------------------------------------
commit 567cb265c435d1e46b899eb1f78c513813ff1561
Author: Dimitri Komatitsch <komatitsch at lma.cnrs-mrs.fr>
Date: Wed Apr 23 01:36:14 2014 +0200
added a comment about -DFORCE_VECTORIZATION
>---------------------------------------------------------------
567cb265c435d1e46b899eb1f78c513813ff1561
flags.guess | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/flags.guess b/flags.guess
index 3c0c454..c0afe83 100644
--- a/flags.guess
+++ b/flags.guess
@@ -5,6 +5,11 @@
# can add -DUSE_SERIAL_CASCADE_FOR_IOs to the compiler options to make the mesher output mesh data
# to the disk for one MPI slice after the other, and to make the solver do the same thing when reading the files back from disk.
+# can also add -DFORCE_VECTORIZATION to force vectorization and unrolling of some critical loops, however this breaks
+# range checking options at run time (for instance -check all for Intel ifort) and, more importantly, modern compilers
+# vectorize the SPECFEM3D_GLOBE code very well and in practice this option only makes the code 3% to 5% faster in the best case,
+# therefore we suggest not activating it.
+
###########################################################################################################################
###########################################################################################################################
###########################################################################################################################
More information about the CIG-COMMITS
mailing list