Merge pull request #126 from DrTimothyAldenDavis/dev

Dev

ChangeLog

@@ -1,3 +1,12 @@
+Apr 10, 2022, SuiteSparse 5.12.0
+
+    * GraphBLAS v7.0.3:  see GraphBLAS/Doc/ChangeLog for details.
+        * performance: GrB_mxm, GrB_assign, and transpose
+        * bug fix: vector iterator for bitmap
+        * revised ACM TOMS submission: Doc/toms_parallel_grb2.pdf
+        * spec bug: GrB_Matrix_diag was implemented incorrectly,
+            thus requiring a version v7.x
+
 Mar 14, 2022, SuiteSparse 5.11.0
 
     * GraphBLAS v6.2.5:  see GraphBLAS/Doc/ChangeLog for changes.

GraphBLAS/CITATION.bib

@@ -0,0 +1,41 @@
+
+@article{GraphBLAS7,
+author = {Davis, Timothy A.},
+title = {Algorithm 10xx: SuiteSparse:GraphBLAS: Graph Algorithms in the Language of Sparse Linear Algebra},
+year = {2022},
+abstract= {SuiteSparse:GraphBLAS is a full parallel implementation of the GraphBLAS
+standard, which defines a set of sparse matrix operations on an extended
+algebra of semirings using an almost unlimited variety of operators and types.
+When applied to sparse adjacency matrices, these algebraic operations are
+equivalent to computations on graphs.  A description of the parallel
+implementation of SuiteSparse:GraphBLAS is given, including its novel parallel
+algorithms for sparse matrix multiply, addition, element-wise multiply,
+submatrix extraction and assignment, and the GraphBLAS mask/accumulator
+operation.  Its performance is illustrated by solving the graph problems in the
+GAP Benchmark and by comparing it with other sparse matrix libraries.},
+journal = {ACM Trans. Math. Softw.},
+month = {(under revision)},
+note={See GraphBLAS/Doc/toms_parallel_grb2.pdf},
+keywords = {GraphBLAS, Graph algorithms, sparse matrices}
+}
+
+@article{10.1145/3322125,
+author = {Davis, Timothy A.},
+title = {Algorithm 1000: SuiteSparse:GraphBLAS: Graph Algorithms in the Language of Sparse Linear Algebra},
+year = {2019},
+issue_date = {December 2019},
+publisher = {Association for Computing Machinery},
+address = {New York, NY, USA},
+volume = {45},
+number = {4},
+issn = {0098-3500},
+url = {https://doi.org/10.1145/3322125},
+doi = {10.1145/3322125},
+abstract = {SuiteSparse:GraphBLAS is a full implementation of the GraphBLAS standard, which defines a set of sparse matrix operations on an extended algebra of semirings using an almost unlimited variety of operators and types. When applied to sparse adjacency matrices, these algebraic operations are equivalent to computations on graphs. GraphBLAS provides a powerful and expressive framework for creating graph algorithms based on the elegant mathematics of sparse matrix operations on a semiring. An overview of the GraphBLAS specification is given, followed by a description of the key features and performance of its implementation in the SuiteSparse:GraphBLAS package.},
+journal = {ACM Trans. Math. Softw.},
+month = {dec},
+articleno = {44},
+numpages = {25},
+keywords = {GraphBLAS, Graph algorithms, sparse matrices}
+}
+

GraphBLAS/CMakeLists.txt

@@ -26,10 +26,10 @@ endif ( )
 set ( CMAKE_MACOSX_RPATH TRUE )
 
 # version of SuiteSparse:GraphBLAS
-set ( GraphBLAS_DATE "Mar 14, 2022" )
-set ( GraphBLAS_VERSION_MAJOR 6 )
-set ( GraphBLAS_VERSION_MINOR 2 )
-set ( GraphBLAS_VERSION_SUB   5 )
+set ( GraphBLAS_DATE "Apr 8, 2022" )
+set ( GraphBLAS_VERSION_MAJOR 7 )
+set ( GraphBLAS_VERSION_MINOR 0 )
+set ( GraphBLAS_VERSION_SUB   3 )
 
 message ( STATUS "Building SuiteSparse:GraphBLAS version: v" ${GraphBLAS_VERSION_MAJOR}.${GraphBLAS_VERSION_MINOR}.${GraphBLAS_VERSION_SUB}  " date: " ${GraphBLAS_DATE} )
 
@@ -511,6 +511,7 @@ if ( DEMO )
     add_executable ( wildtype_demo "Demo/Program/wildtype_demo.c" )
     add_executable ( reduce_demo   "Demo/Program/reduce_demo.c" )
     add_executable ( import_demo   "Demo/Program/import_demo.c" )
+    add_executable ( wathen_demo   "Demo/Program/wathen_demo.c" )
 
     # Libraries required for Demo programs
     target_link_libraries ( openmp_demo   PUBLIC graphblas graphblasdemo ${GB_CUDA} ${GB_RMM} )
@@ -520,6 +521,7 @@ if ( DEMO )
     target_link_libraries ( wildtype_demo PUBLIC graphblas ${GB_CUDA} ${GB_RMM} )
     target_link_libraries ( reduce_demo   PUBLIC graphblas ${GB_CUDA} ${GB_RMM} )
     target_link_libraries ( import_demo   PUBLIC graphblas graphblasdemo ${GB_CUDA} ${GB_RMM} )
+    target_link_libraries ( wathen_demo   PUBLIC graphblas graphblasdemo ${GB_CUDA} ${GB_RMM} )
 
 else ( )
 

GraphBLAS/CUDA/CMakeLists.txt

@@ -7,7 +7,7 @@ cmake_minimum_required(VERSION 3.20.1)
 
 project(GRAPHBLAS_CUDA VERSION 0.1 LANGUAGES CXX CUDA)
 
-set(CMAKE_CUDA_FLAGS "-cudart static -lineinfo -G")
+set(CMAKE_CUDA_FLAGS "-cudart=static -lineinfo -G")
 set(CMAKE_CXX_STANDARD 17)
 set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} --std=c++17 -fPIC ")
 set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -DGBNCPUFEAT")
@@ -16,7 +16,7 @@ set(CMAKE_C_STANDARD 11)
 
 message(STATUS "${CMAKE_CXX_FLAGS}")
 
-file(GLOB GRAPHBLAS_CUDA_SOURCES "*.cu" "*.c")
+file(GLOB GRAPHBLAS_CUDA_SOURCES "*.cu" "*.c" "*.cpp")
 
 add_library(graphblascuda SHARED
             ${GRAPHBLAS_CUDA_SOURCES}
@@ -34,12 +34,18 @@ set(GRAPHBLAS_CUDA_INCLUDES
 
 message(STATUS "${GRAPHBLAS_CUDA_INCLUDES}")
 
-target_include_directories(graphblascuda PUBLIC  ${CUDA_INCLUDE_DIRECTORIES} ${GRAPHBLAS_CUDA_INCLUDES})
+target_include_directories(graphblascuda PUBLIC  ${CUDAToolkit_INCLUDE_DIRS} ${GRAPHBLAS_CUDA_INCLUDES})
 set_target_properties(graphblascuda PROPERTIES POSITION_INDEPENDENT_CODE ON)
 set_target_properties(graphblascuda PROPERTIES CUDA_SEPARABLE_COMPILATION ON)
 set_target_properties(graphblascuda PROPERTIES CUDA_ARCHITECTURES "75")
 
-target_link_libraries(graphblascuda nvrtc cuda)
+target_link_libraries(graphblascuda CUDA::nvrtc CUDA::cudart_static)
+
+install ( TARGETS graphblascuda
+        LIBRARY       DESTINATION ${CMAKE_INSTALL_LIBDIR}
+        PUBLIC_HEADER DESTINATION ${CMAKE_INSTALL_INCLUDEDIR}
+        ARCHIVE       DESTINATION ${CMAKE_INSTALL_LIBDIR} )
+
 
 # 1. Execute enumify/stringify/jitify logic to compile ptx kernels and compile/link w/ relevant *.cu files.
 
@@ -77,8 +83,8 @@ foreach(var ${CUDA_TEST_SUITES})
             # easily ignore them from the build
             add_custom_command(
                     OUTPUT
-                    ${CMAKE_CURRENT_BINARY_DIR}/${var}_${semiring}_${kernel}_test_instances.hpp
-                    ${CMAKE_CURRENT_BINARY_DIR}/${var}_${semiring}_${kernel}_cuda_tests.cu
+                    ${CMAKE_CURRENT_BINARY_DIR}/${var}_${semiring}_test_instances.hpp
+                    ${CMAKE_CURRENT_BINARY_DIR}/${var}_${semiring}_cuda_tests.cpp
                     DEPENDS
                     jitFactory.hpp
                     COMMAND python3 ${CMAKE_CURRENT_SOURCE_DIR}/test/testGen_cmake.py "\"${CMAKE_CURRENT_SOURCE_DIR}\"" "\"${var}\"" "\"${CUDA_TEST_MONOIDS}\""
@@ -87,7 +93,7 @@ foreach(var ${CUDA_TEST_SUITES})
             )
 
             # Construct final list of files to compile (in parallel)
-            list(APPEND CUDA_TEST_CPP_FILES ${CMAKE_CURRENT_BINARY_DIR}/${var}_${semiring}_${kernel}_cuda_tests.cu)
+            list(APPEND CUDA_TEST_CPP_FILES ${CMAKE_CURRENT_BINARY_DIR}/${var}_${semiring}_cuda_tests.cpp)
         endforeach()
     endforeach()
 endforeach()
@@ -100,7 +106,13 @@ FetchContent_Declare(
 )
 # For Windows: Prevent overriding the parent project's compiler/linker settings
 set(gtest_force_shared_crt ON CACHE BOOL "" FORCE)
-FetchContent_MakeAvailable(googletest)
+FetchContent_GetProperties(googletest)
+if(NOT googletest_POPULATED)
+    FetchContent_Populate(googletest)
+    add_subdirectory(${googletest_SOURCE_DIR} ${googletest_BINARY_DIR} EXCLUDE_FROM_ALL)
+endif()
+
+#FetchContent_MakeAvailable(googletest EC)
 
 
 #file(MAKE_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR}/external_includes)
@@ -131,14 +143,18 @@ add_dependencies(graphblascuda_test gtest_main)
 add_dependencies(graphblascuda_test rmm_wrap)
 
 target_link_libraries(graphblascuda_test
+        PUBLIC
         graphblas
         graphblascuda
-        gtest_main
-        nvrtc
-        cuda
-        ${ADDITIONAL_DEPS})
-
-target_include_directories(graphblascuda_test PUBLIC
+        CUDA::cudart_static
+        CUDA::nvrtc
+        ${ADDITIONAL_DEPS}
+        PRIVATE
+        gtest_main)
+
+target_include_directories(graphblascuda_test
+        PUBLIC
         ${ADDITIONAL_INCLUDES}
-        ${CUDA_INCLUDE_DIRECTORIES}
+        ${CUDAToolkit_INCLUDE_DIRS}
         ${GRAPHBLAS_CUDA_INCLUDES})
+