Merge branch 'test' into develop

.github/workflows/continuous-integration.yaml

@@ -186,7 +186,7 @@ jobs:
           - os: windows-latest
             cc: ../llvm/clang
             cxx: ../llvm/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             blas: off
             getBlas: off
             fftw: off
@@ -197,7 +197,7 @@ jobs:
           - os: windows-latest
             cc: ../llvm/clang
             cxx: ../llvm/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             blas: on
             getBlas: on
             fftw: on
@@ -208,7 +208,7 @@ jobs:
           - os: windows-latest
             cc: ./llvm/clang
             cxx: ./llvm/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             blas: off
             getBlas: off
             fftw: off
@@ -219,7 +219,7 @@ jobs:
           - os: windows-latest
             cc: ./llvm/clang
             cxx: ./llvm/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             blas: on
             getBlas: on
             fftw: on
@@ -290,7 +290,7 @@ jobs:
         if: matrix.os == 'macos-latest'
         uses: maxim-lobanov/setup-xcode@v1.5.1
         with:
-            xcode-version: latest
+          xcode-version: latest
 
       # Tests must pass in both debug and release mode
       - name: Compile (Debug)
@@ -394,7 +394,7 @@ jobs:
         run: |
           sudo apt --fix-broken install
           sudo apt-get install python3-sphinx
-          
+
       - name: Configure Files
         run: |
           touch .is_local
@@ -428,14 +428,14 @@ jobs:
           - os: ubuntu-latest
             cc: ../../llvm/bin/clang
             cxx: ../../llvm/bin/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             smallArrays: ON
             pretty: "LibRapid_Ubuntu_Clang_OptimiseSmallArrays"
 
           - os: ubuntu-latest
             cc: ../../llvm/bin/clang
             cxx: ../../llvm/bin/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             smallArrays: OFF
             pretty: "LibRapid_Ubuntu_Clang"
 
@@ -454,14 +454,14 @@ jobs:
           - os: macos-latest
             cc: ../../llvm/bin/clang
             cxx: ../../llvm/bin/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             smallArrays: ON
             pretty: "LibRapid_MacOS_Clang_OptimiseSmallArrays"
 
           - os: macos-latest
             cc: ../../llvm/bin/clang
             cxx: ../../llvm/bin/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             smallArrays: OFF
             pretty: "LibRapid_MacOS_Clang"
 
@@ -480,14 +480,14 @@ jobs:
           - os: windows-latest
             cc: ../../llvm/clang
             cxx: ../../llvm/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             smallArrays: ON
             pretty: "LibRapid_Windows_Clang_OptimiseSmallArrays"
 
           - os: windows-latest
             cc: ../../llvm/clang
             cxx: ../../llvm/clang++
-            clangVer: "15.0"
+            clangVer: "16.0"
             smallArrays: OFF
             pretty: "LibRapid_Windows_Clang"
 
@@ -607,7 +607,7 @@ jobs:
         run: |
           cd IntermediateBenchmarkResults
           tree
-          
+
       - name: Upload Intermediate Benchmark Results
         uses: actions/upload-artifact@v3
         with:

.github/workflows/wheels.yaml

@@ -334,7 +334,7 @@ jobs:
           LIBRAPID_GET_FFTW: OFF
           CC: /usr/local/opt/llvm/bin/clang
           CXX: /usr/local/opt/llvm/bin/clang++
-          CIBW_ENVIRONMENT: CC=/usr/local/opt/llvm/bin/clang CXX=/usr/local/opt/llvm/bin/clang++
+          CIBW_ENVIRONMENT: CC=/usr/local/opt/llvm/bin/clang CXX=/usr/local/opt/llvm/bin/clang++ CMAKE_BUILD_PARALLEL_LEVEL=1
           CMAKE_C_COMPILER: /usr/local/opt/llvm/bin/clang
           CMAKE_CXX_COMPILER: /usr/local/opt/llvm/bin/clang++
 
@@ -357,7 +357,9 @@ jobs:
           LIBRAPID_GET_FFTW: OFF
           CC: ${RUNNER_TOOL_CACHE}/llvm/bin/clang
           CXX: ${RUNNER_TOOL_CACHE}/llvm/bin/clang++
-          CIBW_ENVIRONMENT: CC=${RUNNER_TOOL_CACHE}/llvm/bin/clang CXX=${RUNNER_TOOL_CACHE}/llvm/bin/clang++
+          CIBW_ENVIRONMENT: CC=${RUNNER_TOOL_CACHE}/llvm/bin/clang CXX=${RUNNER_TOOL_CACHE}/llvm/bin/clang++ CMAKE_BUILD_PARALLEL_LEVEL=1
+          CIBW_BEFORE_ALL: |
+            set CMAKE_BUILD_PARALLEL_LEVEL=1
           CMAKE_C_COMPILER: ${RUNNER_TOOL_CACHE}/llvm/bin/clang
           CMAKE_CXX_COMPILER: ${RUNNER_TOOL_CACHE}/llvm/bin/clang++
 
@@ -378,6 +380,7 @@ jobs:
           GITHUB_ACTIONS: ON
           LIBRAPID_GET_BLAS: ON
           LIBRAPID_GET_FFTW: OFF
+          CIBW_ENVIRONMENT: CMAKE_BUILD_PARALLEL_LEVEL=1
 
       - name: Store Artifacts
         uses: actions/upload-artifact@v3

CMakeLists.txt

@@ -114,6 +114,12 @@ if (${SKBUILD})
     )
     FetchContent_MakeAvailable(pybind11)
 
+    # Set CMAKE_BUILD_PARALLEL_LEVEL=1 to reduce memory usage
+    set(CMAKE_BUILD_PARALLEL_LEVEL 1)
+
+    # Set environment variable CMAKE_BUILD_PARALLEL_LEVEL=1 to reduce memory usage
+    set(ENV{CMAKE_BUILD_PARALLEL_LEVEL} 1)
+
     # Run auto-generation script
     # 1. Find the Python executable
     # 2. cd into 'librapid/bindings/generators'
@@ -157,7 +163,7 @@ if (${SKBUILD})
         set(LIBRAPID_USE_CUDA ON)
         set(LIBRAPID_USE_OPENCL ON)
         set(LIBRAPID_NATIVE_ARCH ON)
-    else()
+    else ()
         set(LIBRAPID_NO_ALWAYS_INLINE ON) # Reduce compile memory
     endif ()
 
@@ -181,6 +187,11 @@ if (${SKBUILD})
             ${PYTHON_SOURCES}
     )
 
+    # If using MSVC with GitHub, set /Zm512 to reduce memory usage
+    if (MSVC AND "$ENV{CI}")
+        target_compile_options(${module_name} PRIVATE /Zm512)
+    endif ()
+
     target_compile_definitions(${module_name} PUBLIC
             LIBRAPID_PYTHON
             WIN32_LEAN_AND_MEAN

Doxyfile

@@ -2310,6 +2310,7 @@ PREDEFINED             =      LIBRAPID_DOXYGEN=1\
                               LIBRAPID_NODISCARD=\
                               LIBRAPID_ALWAYS_INLINE=\
                               LIBRAPID_INLINE=\
+                              concept=\
 
 # If the MACRO_EXPANSION and EXPAND_ONLY_PREDEF tags are set to YES then this
 # tag can be used to specify a list of macro names that should be expanded. The

examples/example-cuda.cpp

@@ -4,7 +4,7 @@ namespace lrc = librapid;
 
 auto main() -> int {
 #if defined(LIBRAPID_HAS_CUDA)
-	auto cudaArray = lrc::Array<float, lrc::backend::CUDA>::fromData({{1, 2, 3}, {4, 5, 6}});
+	auto cudaArray = lrc::Array<float, lrc::backend::CUDA>({{1, 2, 3}, {4, 5, 6}});
 	fmt::print("CUDA Array:\n{}\n", cudaArray);
 
 	// Operations on CUDA arrays work exactly the same as on CPU arrays
@@ -26,13 +26,13 @@ auto main() -> int {
 	// Linear algebra operations also work
 	fmt::print("Transposed CUDA Array:\n{}\n", lrc::transpose(prod));
 
-	auto vector = lrc::Array<float, lrc::backend::CUDA>::fromData({{1, 2, 3}});
+	auto vector = lrc::Array<float, lrc::backend::CUDA>({{1, 2, 3}});
 	fmt::print("Array: \n{}\n", cudaArray);
 	fmt::print("Vector: \n{}\n", vector);
 	fmt::print("Matrix dot Vector^T:\n{}\n", lrc::dot(cudaArray, lrc::transpose(vector)));
 #else
 	fmt::print("CUDA not enabled in this build of librapid\n");
-	fmt::print("Check the documentation for more information on enabling OpenCL\n");
+	fmt::print("Check the documentation for more information on enabling CUDA\n");
 	fmt::print("https://librapid.readthedocs.io/en/latest/cmakeIntegration.html#librapid-use-cuda\n");
 #endif // LIBRAPID_HAS_CUDA
 

examples/example-opencl.cpp

@@ -9,7 +9,7 @@ auto main() -> int {
 	// (You can pass (true, true) to select the device manually)
 	lrc::configureOpenCL(true);
 
-	auto openclArray = lrc::Array<float, lrc::backend::OpenCL>::fromData({{1, 2, 3}, {4, 5, 6}});
+	auto openclArray = lrc::Array<float, lrc::backend::OpenCL>({{1, 2, 3}, {4, 5, 6}});
 	fmt::print("OpenCL Array:\n{}\n", openclArray);
 
 	// Operations on OpenCL arrays work exactly the same as on CPU arrays
@@ -31,7 +31,7 @@ auto main() -> int {
 	// Linear algebra operations also work
 	fmt::print("Transposed OpenCL Array:\n{}\n", lrc::transpose(prod));
 
-	auto vector = lrc::Array<float, lrc::backend::OpenCL>::fromData({{1, 2, 3}});
+	auto vector = lrc::Array<float, lrc::backend::OpenCL>({{1, 2, 3}});
 	fmt::print("Array: \n{}\n", openclArray);
 	fmt::print("Vector: \n{}\n", vector);
 	fmt::print("Matrix dot Vector^T:\n{}\n", lrc::dot(openclArray, lrc::transpose(vector)));

librapid/__init__.py

@@ -103,56 +103,23 @@ def Array(*args, **kwargs):
         backend = CPU
 
     if backend == CPU:
-        cpuTypeList = {
-            float32: _librapid.ArrayFloatCPU,
-            float64: _librapid.ArrayDoubleCPU,
-            int32: _librapid.ArrayInt32CPU,
-            int64: _librapid.ArrayInt64CPU,
-            uint32: _librapid.ArrayUInt32CPU,
-            uint64: _librapid.ArrayUInt64CPU,
-            # Complex32: _librapid.ArrayComplexFloatCPU,
-            # Complex64: _librapid.ArrayComplexDoubleCPU,
-        }
-
-        arrayType = cpuTypeList.get(dtype, None)
+        arrayType = __cpuTypeList.get(dtype, None)
     elif backend == OpenCL:
         if not _librapid.hasOpenCL():
             raise RuntimeError("OpenCL is not supported in this build "
                                "of librapid. Ensure OpenCL is installed "
                                "on your system and reinstall librapid "
                                "from source.")
 
-        openclTypeList = {
-            float32: _librapid.ArrayFloatOpenCL,
-            float64: _librapid.ArrayDoubleOpenCL,
-            int32: _librapid.ArrayInt32OpenCL,
-            int64: _librapid.ArrayInt64OpenCL,
-            uint32: _librapid.ArrayUInt32OpenCL,
-            uint64: _librapid.ArrayUInt64OpenCL,
-            # Complex32: _librapid.ArrayComplexFloatOpenCL,
-            # Complex64: _librapid.ArrayComplexDoubleOpenCL,
-        }
-
-        arrayType = openclTypeList.get(dtype, None)
+        arrayType = __openclTypeList.get(dtype, None)
     elif backend == CUDA:
         if not _librapid.hasCUDA():
             raise RuntimeError("CUDA is not supported in this build "
                                "of librapid. Ensure CUDA is installed "
                                "on your system and reinstall librapid "
                                "from source.")
 
-        cudaTypeList = {
-            float32: _librapid.ArrayFloatCUDA,
-            float64: _librapid.ArrayDoubleCUDA,
-            int32: _librapid.ArrayInt32CUDA,
-            int64: _librapid.ArrayInt64CUDA,
-            uint32: _librapid.ArrayUInt32CUDA,
-            uint64: _librapid.ArrayUInt64CUDA,
-            # Complex32: _librapid.ArrayComplexFloatCUDA,
-            # Complex64: _librapid.ArrayComplexDoubleCUDA,
-        }
-
-        arrayType = cudaTypeList.get(dtype, None)
+        arrayType = __cudaTypeList.get(dtype, None)
     else:
         raise ValueError(f"Unknown backend {backend}")
 
@@ -167,58 +134,6 @@ def Array(*args, **kwargs):
         raise RuntimeError("Unknown error")
 
 
-def isArray(obj):
-    """
-    Checks if an object is an Array.
-
-    Parameters
-    ----------
-    obj : object
-        The object to check.
-
-    Returns
-    -------
-    bool
-        True if the object is an Array, False otherwise.
-    """
-
-    if type(obj) in [
-        _librapid.ArrayFloatCPU,
-        _librapid.ArrayDoubleCPU,
-        _librapid.ArrayInt32CPU,
-        _librapid.ArrayInt64CPU,
-        _librapid.ArrayUInt32CPU,
-        _librapid.ArrayUInt64CPU,
-        # _librapid.ArrayComplexFloatCPU,
-        # _librapid.ArrayComplexDoubleCPU
-    ]:
-        return True
-
-    if _librapid.hasOpenCL() and type(obj) in [
-        _librapid.ArrayFloatOpenCL,
-        _librapid.ArrayDoubleOpenCL,
-        _librapid.ArrayInt32OpenCL,
-        _librapid.ArrayInt64OpenCL,
-        _librapid.ArrayUInt32OpenCL,
-        _librapid.ArrayUInt64OpenCL,
-        # _librapid.ArrayComplexFloatOpenCL,
-        # _librapid.ArrayComplexDoubleOpenCL
-    ]:
-        return True
-
-    if _librapid.hasCUDA() and type(obj) in [
-        _librapid.ArrayFloatCUDA,
-        _librapid.ArrayDoubleCUDA,
-        _librapid.ArrayInt32CUDA,
-        _librapid.ArrayInt64CUDA,
-        _librapid.ArrayUInt32CUDA,
-        _librapid.ArrayUInt64CUDA,
-        # _librapid.ArrayComplexFloatCUDA,
-       #  _librapid.ArrayComplexDoubleCUDA
-    ]:
-        return True
-
-
 def hasOpenCL():
     """
     Checks if OpenCL is supported.
@@ -320,7 +235,6 @@ def getSeed():
 exp2 = _librapid.exp2
 exp10 = _librapid.exp10
 
-
 float32 = DataType("float32", 4)
 float64 = DataType("float64", 8)
 int32 = DataType("int32", 4)
@@ -330,6 +244,45 @@ def getSeed():
 # Complex32 = DataType("Complex32", 8)
 # Complex64 = DataType("Complex64", 16)
 
+__cpuTypeList = {
+    int32: _librapid.ArrayInt32CPU,
+    int64: _librapid.ArrayInt64CPU,
+    uint32: _librapid.ArrayUInt32CPU,
+    uint64: _librapid.ArrayUInt64CPU,
+    float32: _librapid.ArrayFloatCPU,
+    float64: _librapid.ArrayDoubleCPU,
+    # Complex32: _librapid.ArrayComplexFloatCPU,
+    # Complex64: _librapid.ArrayComplexDoubleCPU,
+}
+
+if _librapid.hasOpenCL():
+    __openclTypeList = {
+        int32: _librapid.ArrayInt32OpenCL,
+        int64: _librapid.ArrayInt64OpenCL,
+        uint32: _librapid.ArrayUInt32OpenCL,
+        uint64: _librapid.ArrayUInt64OpenCL,
+        float32: _librapid.ArrayFloatOpenCL,
+        float64: _librapid.ArrayDoubleOpenCL,
+        # Complex32: _librapid.ArrayComplexFloatOpenCL,
+        # Complex64: _librapid.ArrayComplexDoubleOpenCL,
+    }
+else:
+    __openclTypeList = None
+
+if _librapid.hasCUDA():
+    __cudaTypeList = {
+        int32: _librapid.ArrayInt32CUDA,
+        int64: _librapid.ArrayInt64CUDA,
+        uint32: _librapid.ArrayUInt32CUDA,
+        uint64: _librapid.ArrayUInt64CUDA,
+        float32: _librapid.ArrayFloatCUDA,
+        float64: _librapid.ArrayDoubleCUDA,
+        # Complex32: _librapid.ArrayComplexFloatCUDA,
+        # Complex64: _librapid.ArrayComplexDoubleCUDA,
+    }
+else:
+    __cudaTypeList = None
+
 CPU = Backend("CPU")
 OpenCL = Backend("OpenCL")
 CUDA = Backend("CUDA")