Fix Windows build compatibility (MSVC cl.exe and clang-cl)

.gitignore

@@ -10,5 +10,6 @@ src/xc_integrator/local_work_driver/host/obara_saika/generator/*.x
 
 # Build directories
 build/
+build*/
 _build/
 cmake-build-*/

cmake/gauxc-dep-versions.cmake

@@ -1,20 +1,20 @@
 set( GAUXC_LINALG_MODULES_REPOSITORY https://github.com/wavefunction91/linalg-cmake-modules.git )
-set( GAUXC_LINALG_MODULES_REVISION  9d2c273a671d6811e9fd432f6a4fa3d915b144b8 )
+set( GAUXC_LINALG_MODULES_REVISION  222364df5e7639f371bf2f37ceb0f476301101a1 )
 
 set( GAUXC_CUB_REPOSITORY https://github.com/NVIDIA/cub.git )
 set( GAUXC_CUB_REVISION   1.10.0 )
 
 set( GAUXC_CUTLASS_REPOSITORY https://github.com/NVIDIA/cutlass.git )
 set( GAUXC_CUTLASS_REVISION v2.10.0 )
 
-set( GAUXC_EXCHCXX_REPOSITORY https://github.com/wavefunction91/ExchCXX.git )
-set( GAUXC_EXCHCXX_REVISION   v1.0.0 )
+set( GAUXC_EXCHCXX_REPOSITORY https://github.com/lorisercole/ExchCXX.git )
+set( GAUXC_EXCHCXX_REVISION   97338065120ca32d04ba77f45d615a9e92cc0d20 )
 
 set( GAUXC_GAU2GRID_REPOSITORY https://github.com/dgasmith/gau2grid.git )
 set( GAUXC_GAU2GRID_REVISION   v2.0.6 )
 
-set( GAUXC_INTEGRATORXX_REPOSITORY https://github.com/wavefunction91/IntegratorXX.git )
-set( GAUXC_INTEGRATORXX_REVISION   1369be58d7a3235dac36d75dd964fef058830622 )
+set( GAUXC_INTEGRATORXX_REPOSITORY https://github.com/lorisercole/IntegratorXX.git )
+set( GAUXC_INTEGRATORXX_REVISION   923125236ea5971ee9accdea39da552b8e322ff6 )
 
 set( GAUXC_HIGHFIVE_REPOSITORY https://github.com/highfive-devs/HighFive.git )
 set( GAUXC_HIGHFIVE_REVISION 805f0e13d09b47c4b01d40682621904aa3b31bb8 )

external/gau2grid/CMakeLists.txt

@@ -12,7 +12,12 @@ target_sources( gauxc PRIVATE
 
 
 target_compile_definitions( gauxc PRIVATE $<BUILD_INTERFACE:__GG_NO_PRAGMA> )
-target_include_directories( gauxc 
+if( CMAKE_C_COMPILER_ID MATCHES "Clang" )
+  # gau2grid uses __pragma(loop(ivdep)) for MSVC vectorization hints.
+  # clang-cl does not recognize this pragma and emits -Wunknown-pragmas.
+  target_compile_options( gauxc PRIVATE -Wno-unknown-pragmas )
+endif()
+target_include_directories( gauxc
   PRIVATE
     $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/generated_source>
 )

external/gau2grid/generated_source/gau2grid/gau2grid_pragma.h

@@ -34,7 +34,7 @@
     #define PRAGMA_RESTRICT                                  __restrict__
 
 #elif defined(__clang__) && defined(_MSC_VER)
-    // pragmas for MSVC
+    // pragmas for clang-cl
 
     #define ALIGNED_MALLOC(alignment, size)                  _aligned_malloc(size, alignment)
     #define ALIGNED_FREE(ptr)                                _aligned_free(ptr)

external/gau2grid/generated_source/gau2grid_deriv1.c

@@ -72,11 +72,9 @@ void gg_collocation_L0_deriv1(const unsigned long npoints, const double* PRAGMA_
     const double center_x = center[0];
     const double center_y = center[1];
     const double center_z = center[2];
-    double A;
-    double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -110,7 +108,7 @@ void gg_collocation_L0_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -130,7 +128,7 @@ void gg_collocation_L0_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];
@@ -225,11 +223,9 @@ void gg_collocation_L1_deriv1(const unsigned long npoints, const double* PRAGMA_
     const double center_x = center[0];
     const double center_y = center[1];
     const double center_z = center[2];
-    double A;
-    double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -263,7 +259,7 @@ void gg_collocation_L1_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -283,7 +279,7 @@ void gg_collocation_L1_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];
@@ -440,7 +436,7 @@ void gg_collocation_L2_deriv1(const unsigned long npoints, const double* PRAGMA_
     double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -474,7 +470,7 @@ void gg_collocation_L2_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -494,7 +490,7 @@ void gg_collocation_L2_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];
@@ -693,7 +689,7 @@ void gg_collocation_L3_deriv1(const unsigned long npoints, const double* PRAGMA_
     double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -727,7 +723,7 @@ void gg_collocation_L3_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -747,7 +743,7 @@ void gg_collocation_L3_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];
@@ -1004,7 +1000,7 @@ void gg_collocation_L4_deriv1(const unsigned long npoints, const double* PRAGMA_
     double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -1038,7 +1034,7 @@ void gg_collocation_L4_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -1058,7 +1054,7 @@ void gg_collocation_L4_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];
@@ -1388,7 +1384,7 @@ void gg_collocation_L5_deriv1(const unsigned long npoints, const double* PRAGMA_
     double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -1422,7 +1418,7 @@ void gg_collocation_L5_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -1442,7 +1438,7 @@ void gg_collocation_L5_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];
@@ -1860,7 +1856,7 @@ void gg_collocation_L6_deriv1(const unsigned long npoints, const double* PRAGMA_
     double AX, AY, AZ;
 
     // Build negative exponents
-    for (unsigned long i = 0; i < nprim; i++) {
+    for (unsigned long i = 0; i < (unsigned long)nprim; i++) {
         expn1[i] = -1.0 * exponents[i];
         expn2[i] = -2.0 * exponents[i];
     }
@@ -1894,7 +1890,7 @@ void gg_collocation_L6_deriv1(const unsigned long npoints, const double* PRAGMA_
                 S1[i] = 0.0;
             }
             } else {
-            unsigned int start_shift = start * xyz_stride;
+            unsigned long start_shift = start * xyz_stride;
 
             PRAGMA_VECTORIZE
             for (unsigned long i = 0; i < remain; i++) {
@@ -1914,7 +1910,7 @@ void gg_collocation_L6_deriv1(const unsigned long npoints, const double* PRAGMA_
         }
 
         // Start exponential block loop
-        for (unsigned long n = 0; n < nprim; n++) {
+        for (unsigned long n = 0; n < (unsigned long)nprim; n++) {
             const double coef = coeffs[n];
             const double alpha_n1 = expn1[n];
             const double alpha_n2 = expn2[n];