0.2.0 release (#231)

* Added Spack config for Summit.

* Added allocator construction to ChaiBuffer and buffer construction to Array.

* Added half precision math functions.

* Updated RAJA to 0.13.0

* Added type conversion to ArrayView.

* Improved CUDA error macro messages.

* Added memcpy and memset functions.

* Use camp Platform for MemorySpace.

* Copyright update.

* Spack and release changes.

* Removed need for matrix output macro.

* Memcpy changes.

* Uncrustify changes.

* Removed constexpr from math functions.

* Changes for CUDA 11.

* Updated release date.

Author: corbett5

CMakeLists.txt

@@ -1,8 +1,15 @@
+###################################################################################################
+# Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
+# All rights reserved.
+# See the LICENSE file for details.
+# SPDX-License-Identifier: (BSD-3-Clause)
+###################################################################################################
+
 cmake_minimum_required( VERSION 3.9 )
 
 # Set version number
 set( LVARRAY_VERSION_MAJOR 0 )
-set( LVARRAY_VERSION_MINOR 1 )
+set( LVARRAY_VERSION_MINOR 2 )
 set( LVARRAY_VERSION_PATCHLEVEL 0 )
 
 # check if LvArray is build as a submodule or a separate project
@@ -32,33 +39,49 @@ if( NOT is_submodule )
     option( ENABLE_CHAI "Build with CHAI" OFF )
     option( ENABLE_CALIPER "Build with Caliper" OFF )
 
-    include( cmake/blt/SetupBLT.cmake )
+    if( NOT BLT_LOADED )
+        if( DEFINED BLT_SOURCE_DIR )
+            if( NOT EXISTS ${BLT_SOURCE_DIR}/SetupBLT.cmake )
+                message( FATAL_ERROR "Given BLT_SOURCE_DIR does not contain SetupBLT.cmake" )
+            endif()
+        else ()
+            set( BLT_SOURCE_DIR ${PROJECT_SOURCE_DIR}/cmake/blt CACHE PATH "" )
+
+            if( NOT EXISTS ${BLT_SOURCE_DIR}/SetupBLT.cmake )
+                message( FATAL_ERROR "The BLT submodule is not present. If in git repository run the following two commands:\n \
+                git submodule init\n \
+                git submodule update" )
+            endif ()
+        endif ()
+
+        include( ${BLT_SOURCE_DIR}/SetupBLT.cmake )
+    endif()
+
     include( cmake/CMakeBasics.cmake )
     include( cmake/SetupTPL.cmake )
 else()
+    if( NOT BLT_LOADED )
+        message( FATAL_ERROR "When using LvArray as a submodule you must have already loaded BLT." )
+    endif()
+
     include( cmake/CMakeBasics.cmake )
 endif()
 
-include(cmake/Macros.cmake)
-include(cmake/Config.cmake)
+include( cmake/Macros.cmake )
+include( cmake/Config.cmake )
 
 set( lvarray_dependencies dl )
 
-if( ENABLE_CHAI )
-    set( lvarray_dependencies ${lvarray_dependencies} chai umpire )
-endif()
+blt_list_append( TO lvarray_dependencies ELEMENTS camp RAJA )
 
-if( ENABLE_CUDA )
-    set( lvarray_dependencies ${lvarray_dependencies} cuda )
-endif()
+blt_list_append( TO lvarray_dependencies ELEMENTS umpire IF ENABLE_UMPIRE )
 
-if( ENABLE_RAJA )
-    set( lvarray_dependencies ${lvarray_dependencies} RAJA )
-endif()
+blt_list_append( TO lvarray_dependencies ELEMENTS chai IF ENABLE_CHAI )
+
+blt_list_append( TO lvarray_dependencies ELEMENTS cuda IF ENABLE_CUDA )
+
+blt_list_append( TO lvarray_dependencies ELEMENTS caliper IF ENABLE_CALIPER )
 
-if( ENABLE_CALIPER )
-    set( lvarray_dependencies ${lvarray_dependencies} caliper )
-endif()
 
 add_subdirectory( src )
 

RELEASE_NOTES.md

@@ -2,13 +2,33 @@ Version vxx.yy.zz -- Release date 20yy-mm-dd
 ============================================
 
 * New features:
-  * Added various tensorOps functions.
-  * Added a Python interface to most container classes.
+
+* API Changes:
+
+* Build changes/improvements:
+
+* Bug fixes:
+
+Version v0.2.0 -- Release date 2020-05-05
+============================================
+
+* New features:
+  * Various tensorOps functions.
+  * A Python library with interfaces to most container classes.
+  * Can create an `Array` that begins life on the device.
+  * Can create an `Array` with specific Umpire allocators.
+  * Math functions for CUDA's `__half` and `__half2`.
+  * Better CUDA error messages.
+  * Added in wrappers around Umpire's `memcpy` and `memset`.
 
 * API Changes:
   * You will now get a compilation error when performing certain undefined operations on rvalues, for example calling `toView` on an `Array` rvalue.
+  * Now use `camp::resources::Platform` for `MemorySpace`.
 
 * Build changes/improvements:
+  * config-build.py works with Python3.
+  * Support for object libraries via the CMake variable `LVARRAY_BUILD_OBJ_LIBS`.
+  * Support for RAJA 0.13.
 
 * Bug fixes:
   * Fixed an indexing bug that produced a compilation error with GCC 10.

benchmarks/CMakeLists.txt

@@ -1,27 +1,34 @@
+###################################################################################################
+# Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
+# All rights reserved.
+# See the LICENSE file for details.
+# SPDX-License-Identifier: (BSD-3-Clause)
+###################################################################################################
+
 #
-# Specify list of tests
+# Specify list of benchmarks
 #
-set(benchmarkSources
-    benchmarkReduce.cpp
-    benchmarkInnerProduct.cpp
-    benchmarkOuterProduct.cpp
-    benchmarkMatrixVector.cpp
-    benchmarkMatrixMatrix.cpp
-    benchmarkArray1DR2TensorMultiplication.cpp
-    benchmarkSparsityGeneration.cpp
-    benchmarkArrayOfArraysReduce.cpp
-    benchmarkArrayOfArraysNodeToElementMapConstruction.cpp
-    benchmarkEigendecomposition.cpp
+set( benchmarkSources
+     benchmarkReduce.cpp
+     benchmarkInnerProduct.cpp
+     benchmarkOuterProduct.cpp
+     benchmarkMatrixVector.cpp
+     benchmarkMatrixMatrix.cpp
+     benchmarkArray1DR2TensorMultiplication.cpp
+     benchmarkSparsityGeneration.cpp
+     benchmarkArrayOfArraysReduce.cpp
+     benchmarkArrayOfArraysNodeToElementMapConstruction.cpp
+     benchmarkEigendecomposition.cpp
    )
 
-if (NOT ${ENABLE_BENCHMARKS})
+if( NOT ${ENABLE_BENCHMARKS} )
     message(FATAL_ERROR "Benchmarks not enabled!")
 endif()
 
 #
-# Add gtest C++ based tests
+# Add benchmarks
 #
-foreach(benchmark ${benchmarkSources})
+foreach( benchmark ${benchmarkSources} )
     get_filename_component( benchmark_name ${benchmark} NAME_WE )
     blt_add_executable( NAME ${benchmark_name}
                         SOURCES ${benchmark} ${benchmark_name}Kernels.cpp
@@ -33,10 +40,10 @@ foreach(benchmark ${benchmarkSources})
     # blt_add_target_compile_flags( TO ${benchmark_name}
     #                               FLAGS -fsave-optimization-record )
 
-    blt_add_benchmark(NAME ${benchmark_name}
-                      COMMAND ${benchmark_name})
+    blt_add_benchmark( NAME ${benchmark_name}
+                       COMMAND ${benchmark_name} )
 
-    install(TARGETS ${benchmark_name}
-            DESTINATION bin)
+    install( TARGETS ${benchmark_name}
+             DESTINATION bin )
 
 endforeach()

benchmarks/benchmarkArray1DR2TensorMultiplication.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)

benchmarks/benchmarkArray1DR2TensorMultiplicationKernels.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)
@@ -156,7 +156,7 @@ void ArrayOfR2TensorsNative< PERMUTATION >::
 RAJAViewKernel( RajaView< VALUE_TYPE const, PERMUTATION > const & a,
                 RajaView< VALUE_TYPE const, PERMUTATION > const & b,
                 RajaView< VALUE_TYPE, PERMUTATION > const & c )
-{ KERNEL( a.layout.sizes[ 0 ], a( i, j, l ), b( i, l, k ), c( i, j, k ) ); }
+{ KERNEL( getRAJAViewLayout( a ).sizes[ 0 ], a( i, j, l ), b( i, l, k ), c( i, j, k ) ); }
 
 template<>
 void ArrayOfR2TensorsNative< RAJA::PERM_IJK >::
@@ -245,7 +245,7 @@ void ArrayOfR2TensorsRAJA< PERMUTATION, POLICY >::
 RAJAViewKernel( RajaView< VALUE_TYPE const, PERMUTATION > const & a,
                 RajaView< VALUE_TYPE const, PERMUTATION > const & b,
                 RajaView< VALUE_TYPE, PERMUTATION > const & c )
-{ RAJA_KERNEL( a.layout.sizes[ 0 ], a( i, j, l ), b( i, l, k ), c( i, j, k ) ); }
+{ RAJA_KERNEL( getRAJAViewLayout( a ).sizes[ 0 ], a( i, j, l ), b( i, l, k ), c( i, j, k ) ); }
 
 template< typename POLICY >
 void pointerRajaHelper( RAJA::PERM_IJK,

benchmarks/benchmarkArray1DR2TensorMultiplicationKernels.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)
@@ -240,7 +240,7 @@ class ArrayOfR2TensorsRAJA : private ArrayOfR2TensorsNative< PERMUTATION >
   }
 
   ~ArrayOfR2TensorsRAJA()
-  { this->m_c.move( MemorySpace::CPU ); }
+  { this->m_c.move( MemorySpace::host ); }
 
   void fortranView() const
   {

benchmarks/benchmarkArrayOfArraysNodeToElementMapConstruction.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)

benchmarks/benchmarkArrayOfArraysNodeToElementMapConstructionKernels.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)

benchmarks/benchmarkArrayOfArraysNodeToElementMapConstructionKernels.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)

benchmarks/benchmarkArrayOfArraysReduce.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2020, Lawrence Livermore National Security, LLC and LvArray contributors.
+ * Copyright (c) 2021, Lawrence Livermore National Security, LLC and LvArray contributors.
  * All rights reserved.
  * See the LICENSE file for details.
  * SPDX-License-Identifier: (BSD-3-Clause)