Update for restrict keywords

nuraiman · Sep 16, 2016 · 392a15a · 392a15a
1 parent 76b827f
commit 392a15a
Show file tree

Hide file tree

Showing 2 changed files with 1 addition and 2 deletions.
diff --git a/openmp45-opt/src/SparseMatrix_functions.hpp b/openmp45-opt/src/SparseMatrix_functions.hpp
@@ -519,7 +519,6 @@ void operator()(MatrixType& A,
 	MINIFE_GLOBAL_ORDINAL row_end   = 0;
 
         #pragma omp target teams distribute parallel for firstprivate(row_start, row_end)
-        #pragma clang loop vectorize(enable)
         for(MINIFE_LOCAL_ORDINAL row = 0; row < rows_size; ++row) {
                 row_start = Arowoffsets[row];
                 row_end   = Arowoffsets[row+1];

diff --git a/openmp45-opt/src/miniFE_info.hpp b/openmp45-opt/src/miniFE_info.hpp
@@ -8,6 +8,6 @@
 
 #define MINIFE_CXX "'/home/projects/pwr8-rhel72/ibm/clang/20160914/clang/bin/clang++'"
 #define MINIFE_CXX_VERSION "'clang version 4.0.0 (bbot:/home/bbot/repos/clang.git 36692876ee67dbbe648874e79144e523fdaf2ce5) (bbot:/home/bbot/repos/llvm.git 2ce3658b3be59a2be93850ddee9792d3265e18a1)'"
-#define MINIFE_CXXFLAGS "'-v -O3 -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda --cuda-path=/home/projects/pwr8-rhel72/cuda/8.0.27 -ffp-contract=fast -mcpu=power8 -mtune=power8 -fslp-vectorize-aggressive -Rpass=loop-vectorize -Rpass-missed=loop-vectorize -Rpass-analysis=loop-vectorize -Xcuda-ptxas -maxrregcount=32 '"
+#define MINIFE_CXXFLAGS "'-v -O3 -fopenmp -fopenmp-targets=nvptx64-nvidia-cuda --cuda-path=/home/projects/pwr8-rhel72/cuda/8.0.27 -ffp-contract=fast -fopenmp-nonaliased-maps -mcpu=power8 -mtune=power8 -fslp-vectorize-aggressive -Rpass-missed='[loop*]' -Rpass-analysis='[loop*]' -Rpass='[loop* | inline | licm | *vectorize*]' -Xcuda-ptxas -maxrregcount=32 '"
 
 #endif