Merge pull request #237 from daineAMD/master

Merge develop into master for ROCm 3.7 feature complete

.githooks/pre-commit

@@ -4,9 +4,7 @@
 # are installed, and if so, uses the installed version to format
 # the staged changes.
 
-set -x
-
-format=/opt/rocm/hcc/bin/clang-format
+export PATH=/usr/bin:/bin:/opt/rocm/llvm/bin:/opt/rocm/hcc/bin
 
 # Redirect stdout to stderr.
 exec >&2
@@ -48,11 +46,11 @@ for file in $files; do
 done
 
 # if clang-format exists, run it on C/C++ files
-if [[ -x $format ]]; then
+if command -v clang-format >/dev/null; then
     for file in $files; do
        if [[ -e $file ]] && echo $file | grep -Eq '\.c$|\.h$|\.hpp$|\.cpp$|\.cl$|\.h\.in$|\.hpp\.in$|\.cpp\.in$'; then
-            echo "$format $file"
-            "$format" -i -style=file "$file"
+            echo "clang-format $file"
+            clang-format -i -style=file "$file"
             git add -u "$file"
         fi
     done

CMakeLists.txt

@@ -45,7 +45,7 @@ include( ROCMInstallTargets )
 include( ROCMPackageConfigHelpers )
 include( ROCMInstallSymlinks )
 
-set ( VERSION_STRING "0.30.0" )
+set ( VERSION_STRING "0.32.0" )
 rocm_setup_version( VERSION ${VERSION_STRING} )
 
 # Append our library helper cmake path and the cmake path for hip (for convenience)

bump_develop_version.sh

@@ -5,7 +5,7 @@
 # - run this script in master branch
 # - after running this script merge master into develop
 
-OLD_HIPBLAS_VERSION="0.30.0"
-NEW_HIPBLAS_VERSION="0.31.0"
+OLD_HIPBLAS_VERSION="0.32.0"
+NEW_HIPBLAS_VERSION="0.33.0"
 
 sed -i "s/${OLD_HIPBLAS_VERSION}/${NEW_HIPBLAS_VERSION}/g" CMakeLists.txt

bump_master_version.sh

@@ -6,7 +6,7 @@
 # - after running this script and merging develop into master, run bump_develop_version.sh in master and
 #   merge master into develop
 
-OLD_HIPBLAS_VERSION="0.29.0"
-NEW_HIPBLAS_VERSION="0.30.0"
+OLD_HIPBLAS_VERSION="0.31.0"
+NEW_HIPBLAS_VERSION="0.32.0"
 
 sed -i "s/${OLD_HIPBLAS_VERSION}/${NEW_HIPBLAS_VERSION}/g" CMakeLists.txt

clients/CMakeLists.txt

@@ -30,8 +30,17 @@ set(hipblas_f90_source_clients
   include/hipblas_fortran.f90
 )
 
+set(hipblas_f90_source_clients_solver
+  include/hipblas_fortran_solver.f90
+)
+
 if( BUILD_CLIENTS_TESTS OR BUILD_CLIENTS_SAMPLES )
-  add_library(hipblas_fortran_client ${hipblas_f90_source_clients})
+  if( BUILD_WITH_SOLVER)
+    add_library(hipblas_fortran_client ${hipblas_f90_source_clients} ${hipblas_f90_source_clients_solver})
+  else()
+    add_library(hipblas_fortran_client ${hipblas_f90_source_clients})
+  endif()
+
   add_dependencies(hipblas_fortran_client hipblas_fortran)
   include_directories(${CMAKE_BINARY_DIR}/include)
 endif( )

clients/common/cblas_interface.cpp

@@ -53,6 +53,18 @@ void zgetrs_(char*                 trans,
              int*                  ldb,
              int*                  info);
 
+void sgetri_(int* n, float* A, int* lda, int* ipiv, float* work, int* lwork, int* info);
+void dgetri_(int* n, double* A, int* lda, int* ipiv, double* work, int* lwork, int* info);
+void cgetri_(
+    int* n, hipblasComplex* A, int* lda, int* ipiv, hipblasComplex* work, int* lwork, int* info);
+void zgetri_(int*                  n,
+             hipblasDoubleComplex* A,
+             int*                  lda,
+             int*                  ipiv,
+             hipblasDoubleComplex* work,
+             int*                  lwork,
+             int*                  info);
+
 void sgeqrf_(int* m, int* n, float* A, int* lda, float* tau, float* work, int* lwork, int* info);
 void dgeqrf_(int* m, int* n, double* A, int* lda, double* tau, double* work, int* lwork, int* info);
 void cgeqrf_(int*            m,
@@ -3163,6 +3175,41 @@ int cblas_getrs<hipblasDoubleComplex>(char                  trans,
     return info;
 }
 
+// getri
+template <>
+int cblas_getri<float>(int n, float* A, int lda, int* ipiv, float* work, int lwork)
+{
+    int info;
+    sgetri_(&n, A, &lda, ipiv, work, &lwork, &info);
+    return info;
+}
+
+template <>
+int cblas_getri<double>(int n, double* A, int lda, int* ipiv, double* work, int lwork)
+{
+    int info;
+    dgetri_(&n, A, &lda, ipiv, work, &lwork, &info);
+    return info;
+}
+
+template <>
+int cblas_getri<hipblasComplex>(
+    int n, hipblasComplex* A, int lda, int* ipiv, hipblasComplex* work, int lwork)
+{
+    int info;
+    cgetri_(&n, A, &lda, ipiv, work, &lwork, &info);
+    return info;
+}
+
+template <>
+int cblas_getri<hipblasDoubleComplex>(
+    int n, hipblasDoubleComplex* A, int lda, int* ipiv, hipblasDoubleComplex* work, int lwork)
+{
+    int info;
+    zgetri_(&n, A, &lda, ipiv, work, &lwork, &info);
+    return info;
+}
+
 // geqrf
 template <>
 int cblas_geqrf<float>(int m, int n, float* A, int lda, float* tau, float* work, int lwork)