Merge pull request #433 from DrTimothyAldenDavis/dev2

v7.2.1.beta2

.github/workflows/build.yaml

@@ -22,7 +22,7 @@ jobs:
     # https://docs.github.com/en/actions/using-github-hosted-runners/about-github-hosted-runners
     runs-on: ubuntu-latest
 
-    name: ubuntu (${{ matrix.compiler }} ${{ matrix.cuda }} CUDA)
+    name: ubuntu (${{ matrix.compiler }} ${{ matrix.cuda }} CUDA ${{ matrix.openmp }} OpenMP)
 
     strategy:
       # Allow other runners in the matrix to continue if some fail
@@ -31,6 +31,7 @@ jobs:
       matrix:
         compiler: [gcc, clang]
         cuda: [with, without]
+        openmp: [with]
         include:
           - compiler: gcc
             compiler-pkgs: "g++ gcc"
@@ -51,6 +52,14 @@ jobs:
               -DENABLE_CUDA=On
               -DCUDAToolkit_INCLUDE_DIR="/usr/include"
               -DCMAKE_CUDA_COMPILER_LAUNCHER="ccache"
+          - compiler: gcc
+            compiler-pkgs: "g++ gcc"
+            cc: "gcc"
+            cxx: "g++"
+            ccache-max: 600M
+            cuda: without
+            openmp: without
+            openmp-cmake-flags: "-DNOPENMP=ON"
 
     env:
       CC: ${{ matrix.cc }}
@@ -75,7 +84,7 @@ jobs:
         # used in action/cache/restore and action/cache/save steps
         id: ccache-prepare
         run: |
-          echo "key=ccache:ubuntu:${{ matrix.compiler }}:${{ matrix.cuda }}:${{ github.ref }}:$(date +"%Y-%m-%d_%H-%M-%S"):${{ github.sha }}" >> $GITHUB_OUTPUT
+          echo "key=ccache:ubuntu:${{ matrix.compiler }}:${{ matrix.cuda }}:${{ matrix.openmp }}:${{ github.ref }}:$(date +"%Y-%m-%d_%H-%M-%S"):${{ github.sha }}" >> $GITHUB_OUTPUT
 
       - name: restore ccache
         # setup the GitHub cache used to maintain the ccache from one job to the next
@@ -85,8 +94,8 @@ jobs:
           key: ${{ steps.ccache-prepare.outputs.key }}
           # Prefer caches from the same branch. Fall back to caches from the dev branch.
           restore-keys: |
-            ccache:ubuntu:${{ matrix.compiler }}:${{ matrix.cuda }}:${{ github.ref }}
-            ccache:ubuntu:${{ matrix.compiler }}:${{ matrix.cuda }}:
+            ccache:ubuntu:${{ matrix.compiler }}:${{ matrix.cuda }}:${{ matrix.openmp }}:${{ github.ref }}
+            ccache:ubuntu:${{ matrix.compiler }}:${{ matrix.cuda }}:${{ matrix.openmp }}:
 
       - name: create empty libraries
         # This is to work around a bug in nvlink.
@@ -127,6 +136,7 @@ jobs:
                   -DCMAKE_Fortran_COMPILER_LAUNCHER="ccache" \
                   -DBLA_VENDOR="OpenBLAS" \
                   ${{ matrix.cuda-cmake-flags }} \
+                  ${{ matrix.openmp-cmake-flags }} \
                   ..
             echo "::endgroup::"
             echo "::group::Build $lib"

CHOLMOD/SuiteSparse_metis/GKlib/random.c

@@ -71,8 +71,8 @@ GK_MKRANDOM(gk_z,   size_t, ssize_t)
 #elif defined ( HAVE_KEYWORD__THREAD )
 
     // gcc and many other compilers support the __thread keyword
-    __thread static uint64_t mt[NN];
-    __thread static int mti=NN+1;
+    static __thread uint64_t mt[NN];
+    static __thread int mti=NN+1;
 
 #elif defined ( HAVE_KEYWORD__DECLSPEC_THREAD )
 

ChangeLog

@@ -1,5 +1,9 @@
-Sept 18, 2023: version 7.2.1
+Oct 7, 2023: version 7.2.1
 
+    * GraphBLAS 8.2.1: bug fix to GrB_mxm; incorrect handling of typecasting
+    * cross-compiler support: replace check_c_source_runs with _compiles,
+        for GraphBLAS and SuiteSparse_config, and remove check for
+        getenv("HOME").
     * cmake update: add "None" build type, from Antonio Rojas, for Arch Linux,
         to all *Config.cmake files for all packages except CSparse (CXSparse
         is built instead, and CSparse does not have CSparseConfig.cmake file)

GraphBLAS/CMakeLists.txt

@@ -384,9 +384,10 @@ else ( )
     "(Linux, conda-forge, spack, brew, vcpkg, etc). " )
     # check for __thread, __declspec(thread), and _Thread_local keywords
     include ( SuiteSparse__thread )
-    if ( NOT HAVE_KEYWORD__DECLSPEC_THREAD AND NOT HAVE_KEYWORD__THREAD )
-        message  ( WARNING "The C compiler supports neither the __thread "
-        "nor the __declspec(thread) keyword for thread-local-storage; "
+    if ( NOT HAVE_KEYWORD__DECLSPEC_THREAD AND NOT HAVE_KEYWORD__THREAD
+        AND NOT HAVE_KEYWORD__THREAD_LOCAL )
+        message  ( WARNING
+        "The C compiler does not support thread-local-storage; "
         "GxB_Context_engage will return GrB_NOT_IMPLEMENTED." )
     endif ( )
 endif ( )

GraphBLAS/Config/README.md.in

@@ -225,10 +225,9 @@ Thanks for packaging SuiteSparse!  Here are some suggestions:
 
 To cite this package, please use the following:
 
-    T. Davis, Algorithm 10xx: SuiteSparse:GraphBLAS: parallel graph
-    algorithms in the language of sparse linear algebra, ACM Trans on
-    Mathematical Software, just acceped, 2023.  See the pdf in
-    https://github.com/DrTimothyAldenDavis/GraphBLAS/tree/stable/Doc .
+    T. A. Davis. Algorithm 1037: SuiteSparse:GraphBLAS: Parallel Graph
+    Algorithms in the Language of Sparse Linear Algebra. ACM Trans. Math.
+    Softw. 49, 3, Article 28 (September 2023), 30 pages.
     https://doi.org/10.1145/3577195
 
     T. Davis, Algorithm 1000: SuiteSparse:GraphBLAS: graph algorithms in

GraphBLAS/Doc/ChangeLog

@@ -1,5 +1,10 @@
-Sept 18, 2023: version 8.2.1
+Oct 7, 2023: version 8.2.1
 
+    * (49) bug fix: GrB_mxm saxpy4 and saxpy5 had incorrectly handling of
+        typecasting in v8.0.0 to v8.2.0 (caught by Erik Welch)
+    * cross-compiler support: replace check_c_source_runs with _compiles,
+        for GraphBLAS and SuiteSparse_config, and remove check for
+        getenv("HOME").
     * cmake update: add "None" build type, from Antonio Rojas, for Arch Linux
 
 Sept 8, 2023: version 8.2.0

GraphBLAS/Doc/GraphBLAS_UserGuide.bib

@@ -162,15 +162,26 @@ @article{Davis19
 keywords = {sparse matrices, GraphBLAS, Graph algorithms}
 }
 
-@article{Davis22,
+@article{Davis23,
 author = {Davis, Timothy A.},
-title = {Algorithm 10xx: {SuiteSparse:GraphBLAS}: Parallel Graph Algorithms in the Language of Sparse Linear Algebra},
-journal = {ACM Trans. Math. Softw.},
+title = {Algorithm 1037: SuiteSparse:GraphBLAS: Parallel Graph Algorithms in the Language of Sparse Linear Algebra},
 year = {2023},
-note = {(to appear)}
+issue_date = {September 2023},
+publisher = {Association for Computing Machinery},
+address = {New York, NY, USA},
+volume = {49},
+number = {3},
+issn = {0098-3500},
+url = {https://doi.org/10.1145/3577195},
+doi = {10.1145/3577195},
+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 = {sep},
+articleno = {28},
+numpages = {30},
+keywords = {Graph algorithms, GraphBLAS, sparse matrices}
 }
 
-
 @book{Higham,
         author={Higham, N.},
         title={Accuracy and Stability of Numerical Algorithms},

GraphBLAS/Doc/GraphBLAS_UserGuide.tex

@@ -116,7 +116,7 @@ \section{Introduction} %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
 GraphBLAS library.
 
 For more details on SuiteSparse:GraphBLAS, and its use in LAGraph, see
-\cite{Davis19,Davis22,Davis18b,DavisAznavehKolodziej19,Davis20,Mattson19}.
+\cite{Davis19,Davis23,Davis18b,DavisAznavehKolodziej19,Davis20,Mattson19}.
 
 A full and precise definition of the GraphBLAS specification is provided in
 {\em The GraphBLAS C API Specification} by {Ayd\i n Bulu\c{c}, Timothy Mattson,
@@ -16448,9 +16448,14 @@ \section{Release Notes}
 
 \begin{itemize}
 
-\item Sept 18, 2023: version 8.2.1
+\item Oct 7, 2023: version 8.2.1
 
     \begin{itemize}
+    \item (49) bug fix: \verb'GrB_mxm' saxpy4 and saxpy5 had incorrectly handling of
+        typecasting in v8.0.0 to v8.2.0 (caught by Erik Welch)
+    \item cross-compiler support: replace \verb'check_c_source_runs' with \verb'_compiles'
+        for GraphBLAS and SuiteSparse\_config, and remove check for
+        \verb'getenv("HOME")'.
     \item cmake update: add "None" build type, from Antonio Rojas, for Arch Linux
     \end{itemize}
 

GraphBLAS/Doc/GraphBLAS_version.tex

@@ -1,5 +1,5 @@
 % version of SuiteSparse:GraphBLAS
 \date{VERSION
 8.2.1,
-Sept 18, 2023}
+Oct 7, 2023}
 

GraphBLAS/Include/GraphBLAS.h

@@ -220,7 +220,7 @@
 
 // The version of this implementation, and the GraphBLAS API version:
 #define GxB_IMPLEMENTATION_NAME "SuiteSparse:GraphBLAS"
-#define GxB_IMPLEMENTATION_DATE "Sept 18, 2023"
+#define GxB_IMPLEMENTATION_DATE "Oct 7, 2023"
 #define GxB_IMPLEMENTATION_MAJOR 8
 #define GxB_IMPLEMENTATION_MINOR 2
 #define GxB_IMPLEMENTATION_SUB   1