Merge pull request #888 from mkrainiuk/lapacke_64

Add Index-64 API as extended API with _64 suffix for LAPACKE

Author: langou

CMakeLists.txt

@@ -107,8 +107,10 @@ else()
   set(LAPACKELIB "lapacke")
   set(TMGLIB "tmglib")
 endif()
-# By default build standard API and extended _64 API
+# By default build extended _64 API for supported compilers only
+set(INDEX64_EXT_API_COMPILERS "Intel|GNU")
 option(BUILD_INDEX64_EXT_API "Build Index-64 API as extended API with _64 suffix" ON)
+message(STATUS "Build Index-64 API as extended API with _64 suffix: ${BUILD_INDEX64_EXT_API}")
 
 include(GNUInstallDirs)
 

INSTALL/droundup_lwork.f

@@ -76,7 +76,8 @@ DOUBLE PRECISION FUNCTION DROUNDUP_LWORK( LWORK )
 *
       IF( INT( DROUNDUP_LWORK ) .LT. LWORK ) THEN
 *         Force round up of LWORK
-          DROUNDUP_LWORK = DROUNDUP_LWORK * ( 1.0D+0 + EPSILON(0.0D+0) )
+          DROUNDUP_LWORK = DROUNDUP_LWORK *
+     $                     ( 1.0D+0 + EPSILON(0.0D+0) )
       ENDIF
 *
       RETURN

INSTALL/sroundup_lwork.f

@@ -76,7 +76,8 @@ REAL             FUNCTION SROUNDUP_LWORK( LWORK )
 *
       IF( INT( SROUNDUP_LWORK ) .LT. LWORK ) THEN
 *         Force round up of LWORK
-          SROUNDUP_LWORK = SROUNDUP_LWORK * ( 1.0E+0 + EPSILON(0.0E+0) )
+          SROUNDUP_LWORK = SROUNDUP_LWORK *
+     $                     ( 1.0E+0 + EPSILON(0.0E+0) )
       ENDIF
 *
       RETURN

LAPACKE/CMakeLists.txt

@@ -67,12 +67,35 @@ if(LAPACKE_WITH_TMG)
 endif()
 list(APPEND SOURCES ${UTILS})
 
-add_library(${LAPACKELIB} ${SOURCES})
+add_library(${LAPACKELIB}_obj OBJECT ${SOURCES})
+set_target_properties(${LAPACKELIB}_obj PROPERTIES POSITION_INDEPENDENT_CODE ON)
+
+if(BUILD_INDEX64_EXT_API)
+  # 64bit Integer Extended Interface
+  set(SOURCES_64_C)
+  list(APPEND SOURCES_64_C ${SOURCES})
+  list(REMOVE_ITEM SOURCES_64_C src/lapacke_nancheck.c)
+  list(REMOVE_ITEM SOURCES_64_C utils/lapacke_make_complex_float.c)
+  list(REMOVE_ITEM SOURCES_64_C utils/lapacke_make_complex_double.c)
+  add_library(${LAPACKELIB}_64_obj OBJECT ${SOURCES_64_C})
+  set_target_properties(${LAPACKELIB}_64_obj PROPERTIES
+          POSITION_INDEPENDENT_CODE ON)
+  target_compile_options(${LAPACKELIB}_64_obj PRIVATE
+          -DLAPACK_ILP64
+          -DLAPACKE_API64
+          -DWeirdNEC
+          -DCBLAS_API64)
+endif()
+
+add_library(${LAPACKELIB} $<TARGET_OBJECTS:${LAPACKELIB}_obj>
+        $<$<BOOL:${BUILD_INDEX64_EXT_API}>: $<TARGET_OBJECTS:${LAPACKELIB}_64_obj>>)
+
 set_target_properties(
   ${LAPACKELIB} PROPERTIES
   LINKER_LANGUAGE C
   VERSION ${LAPACK_VERSION}
   SOVERSION ${LAPACK_MAJOR_VERSION}
+  POSITION_INDEPENDENT_CODE ON
   )
 target_include_directories(${LAPACKELIB} PUBLIC
     $<BUILD_INTERFACE:${LAPACK_BINARY_DIR}/include>

LAPACKE/example/CMakeLists.txt

@@ -3,12 +3,29 @@ add_executable(xexample_DGESV_colmajor example_DGESV_colmajor.c lapacke_example_
 add_executable(xexample_DGELS_rowmajor example_DGELS_rowmajor.c lapacke_example_aux.c lapacke_example_aux.h)
 add_executable(xexample_DGELS_colmajor example_DGELS_colmajor.c lapacke_example_aux.c lapacke_example_aux.h)
 
-target_link_libraries(xexample_DGESV_rowmajor ${LAPACKELIB})
-target_link_libraries(xexample_DGESV_colmajor ${LAPACKELIB})
-target_link_libraries(xexample_DGELS_rowmajor ${LAPACKELIB})
-target_link_libraries(xexample_DGELS_colmajor ${LAPACKELIB})
+target_link_libraries(xexample_DGESV_rowmajor ${LAPACKELIB} ${BLAS_LIBRARIES})
+target_link_libraries(xexample_DGESV_colmajor ${LAPACKELIB} ${BLAS_LIBRARIES})
+target_link_libraries(xexample_DGELS_rowmajor ${LAPACKELIB} ${BLAS_LIBRARIES})
+target_link_libraries(xexample_DGELS_colmajor ${LAPACKELIB} ${BLAS_LIBRARIES})
 
 add_test(example_DGESV_rowmajor ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGESV_rowmajor)
 add_test(example_DGESV_colmajor ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGESV_colmajor)
 add_test(example_DGELS_rowmajor ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGELS_rowmajor)
 add_test(example_DGELS_colmajor ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGELS_colmajor)
+
+if(BUILD_INDEX64_EXT_API)
+  add_executable(xexample_DGESV_rowmajor_64 example_DGESV_rowmajor_64.c lapacke_example_aux.c lapacke_example_aux.h)
+  add_executable(xexample_DGESV_colmajor_64 example_DGESV_colmajor_64.c lapacke_example_aux.c lapacke_example_aux.h)
+  add_executable(xexample_DGELS_rowmajor_64 example_DGELS_rowmajor_64.c lapacke_example_aux.c lapacke_example_aux.h)
+  add_executable(xexample_DGELS_colmajor_64 example_DGELS_colmajor_64.c lapacke_example_aux.c lapacke_example_aux.h)
+
+  target_link_libraries(xexample_DGESV_rowmajor_64 ${LAPACKELIB} ${BLAS_LIBRARIES})
+  target_link_libraries(xexample_DGESV_colmajor_64 ${LAPACKELIB} ${BLAS_LIBRARIES})
+  target_link_libraries(xexample_DGELS_rowmajor_64 ${LAPACKELIB} ${BLAS_LIBRARIES})
+  target_link_libraries(xexample_DGELS_colmajor_64 ${LAPACKELIB} ${BLAS_LIBRARIES})
+
+  add_test(example_DGESV_rowmajor_64 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGESV_rowmajor_64)
+  add_test(example_DGESV_colmajor_64 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGESV_colmajor_64)
+  add_test(example_DGELS_rowmajor_64 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGELS_rowmajor_64)
+  add_test(example_DGELS_colmajor_64 ${CMAKE_RUNTIME_OUTPUT_DIRECTORY}/xexample_DGELS_colmajor_64)
+endif()

LAPACKE/example/example_DGELS_colmajor_64.c

@@ -0,0 +1,93 @@
+/*
+   LAPACKE Example : Calling DGELS using col-major layout
+   =====================================================
+
+   The program computes the solution to the system of linear
+   equations with a square matrix A and multiple
+   right-hand sides B, where A is the coefficient matrix
+   and b is the right-hand side matrix:
+
+   Description
+   ===========
+
+   In this example, we wish solve the least squares problem min_x || B - Ax ||
+   for two right-hand sides using the LAPACK routine DGELS. For input we will
+   use the 5-by-3 matrix
+
+         ( 1  1  1 )
+         ( 2  3  4 )
+     A = ( 3  5  2 )
+         ( 4  2  5 )
+         ( 5  4  3 )
+    and the 5-by-2 matrix
+
+         ( -10 -3 )
+         (  12 14 )
+     B = (  14 12 )
+         (  16 16 )
+         (  18 16 )
+    We will first store the input matrix as a static C two-dimensional array,
+    which is stored in col-major layout, and let LAPACKE handle the work space
+    array allocation. The LAPACK base name for this function is gels, and we
+    will use double precision (d), so the LAPACKE function name is LAPACKE_dgels.
+
+    lda=5 and ldb=5. The output for each right hand side is stored in b as
+    consecutive vectors of length 3. The correct answer for this problem is
+    the 3-by-2 matrix
+
+         ( 2 1 )
+         ( 1 1 )
+         ( 1 2 )
+
+    A complete C program for this example is given below. Note that when the arrays
+     are passed to the LAPACK routine, they must be dereferenced, since LAPACK is
+      expecting arrays of type double *, not double **.
+
+
+   LAPACKE Interface
+   =================
+
+   LAPACKE_dgels (col-major, high-level) Example Program Results
+
+  -- LAPACKE Example routine --
+  -- LAPACK is a software package provided by Univ. of Tennessee,    --
+  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
+*/
+/* Calling DGELS using col-major layout */
+
+/* Includes */
+#include <stdio.h>
+#include <lapacke_64.h>
+#include "lapacke_example_aux.h"
+
+/* Main program */
+int main (int argc, const char * argv[])
+{
+   /* Locals */
+   double A[5][3] = {{1,2,3},{4,5,1},{3,5,2},{4,1,4},{2,5,3}};
+   double b[5][2] = {{-10,12},{14,16},{18,-3},{14,12},{16,16}};
+   int64_t info,m,n,lda,ldb,nrhs;
+
+   /* Initialization */
+   m = 5;
+   n = 3;
+   nrhs = 2;
+   lda = 5;
+   ldb = 5;
+
+   /* Print Entry Matrix */
+   print_matrix_colmajor_64( "Entry Matrix A", m, n, *A, lda );
+   /* Print Right Rand Side */
+   print_matrix_colmajor_64( "Right Hand Side b", n, nrhs, *b, ldb );
+   printf( "\n" );
+
+   /* Executable statements */
+   printf( "LAPACKE_dgels_64 (col-major, high-level) Example Program Results\n" );
+   /* Solve least squares problem*/
+   info = LAPACKE_dgels_64(LAPACK_COL_MAJOR,'N',m,n,nrhs,*A,lda,*b,ldb);
+
+   /* Print Solution */
+   print_matrix_colmajor_64( "Solution", n, nrhs, *b, ldb );
+   printf( "\n" );
+   exit( info );
+} /* End of LAPACKE_dgels Example */

LAPACKE/example/example_DGELS_rowmajor_64.c

@@ -0,0 +1,93 @@
+/*
+   LAPACKE Example : Calling DGELS using row-major layout
+   =====================================================
+
+   The program computes the solution to the system of linear
+   equations with a square matrix A and multiple
+   right-hand sides B, where A is the coefficient matrix
+   and b is the right-hand side matrix:
+
+   Description
+   ===========
+
+   In this example, we wish solve the least squares problem min_x || B - Ax ||
+   for two right-hand sides using the LAPACK routine DGELS. For input we will
+   use the 5-by-3 matrix
+
+         ( 1  1  1 )
+         ( 2  3  4 )
+     A = ( 3  5  2 )
+         ( 4  2  5 )
+         ( 5  4  3 )
+    and the 5-by-2 matrix
+
+         ( -10 -3 )
+         (  12 14 )
+     B = (  14 12 )
+         (  16 16 )
+         (  18 16 )
+    We will first store the input matrix as a static C two-dimensional array,
+    which is stored in row-major layout, and let LAPACKE handle the work space
+    array allocation. The LAPACK base name for this function is gels, and we
+    will use double precision (d), so the LAPACKE function name is LAPACKE_dgels.
+
+    thus lda=3 and ldb=2. The output for each right hand side is stored in b as
+    consecutive vectors of length 3. The correct answer for this problem is
+    the 3-by-2 matrix
+
+         ( 2 1 )
+         ( 1 1 )
+         ( 1 2 )
+
+    A complete C program for this example is given below. Note that when the arrays
+     are passed to the LAPACK routine, they must be dereferenced, since LAPACK is
+      expecting arrays of type double *, not double **.
+
+
+   LAPACKE Interface
+   =================
+
+   LAPACKE_dgels (row-major, high-level) Example Program Results
+
+  -- LAPACKE Example routine --
+  -- LAPACK is a software package provided by Univ. of Tennessee,    --
+  -- Univ. of California Berkeley, Univ. of Colorado Denver and NAG Ltd..--
+*/
+/* Calling DGELS using row-major layout */
+
+/* Includes */
+#include <stdio.h>
+#include <lapacke_64.h>
+#include "lapacke_example_aux.h"
+
+/* Main program */
+int main (int argc, const char * argv[])
+{
+   /* Locals */
+   double A[5][3] = {{1,1,1},{2,3,4},{3,5,2},{4,2,5},{5,4,3}};
+   double b[5][2] = {{-10,-3},{12,14},{14,12},{16,16},{18,16}};
+   int64_t info,m,n,lda,ldb,nrhs;
+
+   /* Initialization */
+   m = 5;
+   n = 3;
+   nrhs = 2;
+   lda = 3;
+   ldb = 2;
+
+   /* Print Entry Matrix */
+   print_matrix_rowmajor_64( "Entry Matrix A", m, n, *A, lda );
+   /* Print Right Rand Side */
+   print_matrix_rowmajor_64( "Right Hand Side b", n, nrhs, *b, ldb );
+   printf( "\n" );
+
+   /* Executable statements */
+   printf( "LAPACKE_dgels_64 (row-major, high-level) Example Program Results\n" );
+   /* Solve least squares problem*/
+   info = LAPACKE_dgels_64(LAPACK_ROW_MAJOR,'N',m,n,nrhs,*A,lda,*b,ldb);
+
+   /* Print Solution */
+   print_matrix_rowmajor_64( "Solution", n, nrhs, *b, ldb );
+   printf( "\n" );
+   exit( 0 );
+} /* End of LAPACKE_dgels Example */

LAPACKE/example/example_DGESV_colmajor.c

@@ -41,41 +41,41 @@ int main(int argc, char **argv) {
 
         /* Locals */
         lapack_int n, nrhs, lda, ldb, info;
-		int i, j;
+        int i, j;
         /* Local arrays */
-		double *A, *b;
-		lapack_int *ipiv;
+        double *A, *b;
+        lapack_int *ipiv;
 
         /* Default Value */
-	    n = 5; nrhs = 1;
+        n = 5; nrhs = 1;
 
         /* Arguments */
-	    for( i = 1; i < argc; i++ ) {
-	    	if( strcmp( argv[i], "-n" ) == 0 ) {
-		    	n  = atoi(argv[i+1]);
-			    i++;
-		    }
-			if( strcmp( argv[i], "-nrhs" ) == 0 ) {
-				nrhs  = atoi(argv[i+1]);
-				i++;
-			}
-		}
+        for( i = 1; i < argc; i++ ) {
+            if( strcmp( argv[i], "-n" ) == 0 ) {
+                n  = atoi(argv[i+1]);
+                i++;
+            }
+            if( strcmp( argv[i], "-nrhs" ) == 0 ) {
+                nrhs  = atoi(argv[i+1]);
+                i++;
+            }
+        }
 
         /* Initialization */
         lda=n, ldb=n;
-		A = (double *)malloc(n*n*sizeof(double)) ;
-		if (A==NULL){ printf("error of memory allocation\n"); exit(0); }
-		b = (double *)malloc(n*nrhs*sizeof(double)) ;
-		if (b==NULL){ printf("error of memory allocation\n"); exit(0); }
-		ipiv = (lapack_int *)malloc(n*sizeof(lapack_int)) ;
-		if (ipiv==NULL){ printf("error of memory allocation\n"); exit(0); }
+        A = (double *)malloc(n*n*sizeof(double)) ;
+        if (A==NULL){ printf("error of memory allocation\n"); exit(0); }
+        b = (double *)malloc(n*nrhs*sizeof(double)) ;
+        if (b==NULL){ printf("error of memory allocation\n"); free(A); exit(0); }
+        ipiv = (lapack_int *)malloc(n*sizeof(lapack_int)) ;
+        if (ipiv==NULL){ printf("error of memory allocation\n"); free(A); free(b); exit(0); }
 
         for( i = 0; i < n; i++ ) {
                 for( j = 0; j < n; j++ ) A[i+j*lda] = ((double) rand()) / ((double) RAND_MAX) - 0.5;
-		}
+        }
 
-		for(i=0;i<n*nrhs;i++)
-			b[i] = ((double) rand()) / ((double) RAND_MAX) - 0.5;
+        for(i=0;i<n*nrhs;i++)
+            b[i] = ((double) rand()) / ((double) RAND_MAX) - 0.5;
 
         /* Print Entry Matrix */
         print_matrix_colmajor( "Entry Matrix A", n, n, A, lda );
@@ -91,12 +91,20 @@ int main(int argc, char **argv) {
 
         /* Check for the exact singularity */
         if( info > 0 ) {
-                printf( "The diagonal element of the triangular factor of A,\n" );
-                printf( "U(%" LAPACK_IFMT ",%" LAPACK_IFMT ") is zero, so that A is singular;\n", info, info );
-                printf( "the solution could not be computed.\n" );
-                exit( 1 );
+            printf( "The diagonal element of the triangular factor of A,\n" );
+            printf( "U(%" LAPACK_IFMT ",%" LAPACK_IFMT ") is zero, so that A is singular;\n", info, info );
+            printf( "the solution could not be computed.\n" );
+            free(A);
+            free(b);
+            free(ipiv);
+            exit( 1 );
+        }
+        if (info <0) {
+            free(A);
+            free(b);
+            free(ipiv);
+            exit( 1 );
         }
-        if (info <0) exit( 1 );
         /* Print solution */
         print_matrix_colmajor( "Solution", n, nrhs, b, ldb );
         /* Print details of LU factorization */