include/malloc_unmanaged.hpp

@@ -6,42 +6,18 @@
 #define RAIISAFECUDA_MALLOC_UNMANAGED_HPP
 
 #include <concepts>
-#include <cstdint>
 #include <memory>
-
+#include <variant>
 #include <cuda_runtime_api.h>
 
+namespace safe_cuda::unmanaged {
+    using deviceMallocDestroyType = decltype(&cudaFree);
+    using hostMallocDestroyType = decltype(&cudaFreeHost);
 
-namespace safe_cuda {
+    template<typename D>
-    enum class allocType : std::uint8_t {
+    concept CudaArrayDestroyer = std::is_same_v<D, deviceMallocDestroyType> || std::is_same_v<D, hostMallocDestroyType>;
-        Unmanaged = 0,
-        Managed = 1,
-        Host = 2,
-    };
 
-    template<typename T, allocType alloc_type = allocType::Managed>
+    template<typename T, CudaArrayDestroyer D> requires std::integral<T> || std::floating_point<T>
-    struct destroyType {
-        void operator()(T *ptr) const noexcept {
-            (void) cudaFree(ptr);
-        }
-    };
-
-    template<typename T>
-    struct destroyType<T, allocType::Unmanaged> {
-        void operator()(T *ptr) const noexcept {
-            (void) cudaFree(ptr);
-        }
-    };
-
-    template<typename T>
-    struct destroyType<T, allocType::Host> {
-        void operator()(T *ptr) const noexcept {
-            (void) cudaFreeHost(ptr);
-        }
-    };
-
-    template<typename T, allocType alloc_type, typename D = destroyType<T, alloc_type> >
-        requires std::integral<T> || std::floating_point<T>
     using safePtrType = std::unique_ptr<T, D>;
 
     /**
@@ -49,26 +25,25 @@ namespace safe_cuda {
      *
      * It can allocate unmanaged memory on device and on Host for pinned memory.
      * \tparam T bare and built-in type.
-     * \tparam alloc_type Type of allocation: Managed (default), Unmanage, Host.
+     * \tparam D destroyer type. It determines if it allocates on device or pinned host.
      * \param byteDataSize
      * \return
      */
-    template<typename T, allocType alloc_type>
+    template<typename T, CudaArrayDestroyer D>
-    std::pair<safePtrType<T, alloc_type>, cudaError_t>
+    std::variant<safePtrType<T, D>, cudaError_t> cuda_malloc(const std::size_t byteDataSize) noexcept {
-    cuda_malloc(const std::size_t byteDataSize) noexcept {
         T *ptr_tmp = nullptr;
-        cudaError_t error = cudaSuccess;
+        if constexpr (std::is_same_v<D, deviceMallocDestroyType>) {
-        switch (alloc_type) {
+            const cudaError_t error = cudaMalloc(&ptr_tmp, byteDataSize);
-            case allocType::Unmanaged:
+            if (error != cudaSuccess) {
-                error = cudaMalloc(reinterpret_cast<void **>(&ptr_tmp), byteDataSize);
+                return error;
-                return { safePtrType<T, alloc_type>{ ptr_tmp, destroyType<T, alloc_type>{} }, error };
-            case allocType::Host:
-                error = cudaMallocHost(reinterpret_cast<void **>(&ptr_tmp), byteDataSize);
-                return { safePtrType<T, alloc_type>{ ptr_tmp, destroyType<T, alloc_type>{} }, error };
-            case allocType::Managed:
-                error = cudaMallocManaged(reinterpret_cast<void **>(&ptr_tmp), byteDataSize);
-                return { safePtrType<T, alloc_type>{ ptr_tmp, destroyType<T, alloc_type>{} }, error };
             }
+        } else if constexpr (std::is_same_v<D, hostMallocDestroyType>) {
+            const cudaError_t error = cudaMallocHost(&ptr_tmp, byteDataSize);
+            if (error != cudaSuccess) {
+                return error;
+            }
+        }
+        return safePtrType<T, D>{ ptr_tmp, cudaFree };
     }
 }
 

src/stream_related.cpp

@@ -3,7 +3,8 @@
 namespace safe_cuda {
     std::variant<std::unique_ptr<CUstream_st, streamDestroyType>, cudaError_t> create_stream() noexcept {
         cudaStream_t stream = nullptr;
-        if (const cudaError_t error = cudaStreamCreate(&stream); error != cudaSuccess) {
+        const cudaError_t error = cudaStreamCreate(&stream);
+        if (error != cudaSuccess) {
             return error;
         }
         return std::unique_ptr<CUstream_st, streamDestroyType>{ stream, cudaStreamDestroy };

tests/CMakeLists.txt

@@ -7,7 +7,7 @@ if (NOT Catch2_FOUND)
     fetchcontent_declare(
             Catch2
             GIT_REPOSITORY https://github.com/catchorg/Catch2
-            GIT_TAG v3.11.0
+            GIT_TAG v3.5.3
             EXCLUDE_FROM_ALL
             OVERRIDE_FIND_PACKAGE
     )
@@ -20,15 +20,7 @@ enable_testing()
 include(Catch)
 include(CatchAddTests)
 
-add_executable(tests
+add_executable(tests tests_stream.cpp)
-        tests_stream.cpp
+target_link_libraries(tests Catch2::Catch2WithMain)
-        tests_safe_allocation.cpp
-)
-target_link_libraries(tests Catch2::Catch2WithMain raiiSafeCuda)
-set_target_properties(tests PROPERTIES
-        CXX_STANDARD 20
-        CXX_EXTENSIONS OFF
-        INTERPROCEDURAL_OPTIMIZATION ON
-)
 
 catch_discover_tests(tests WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR})

tests/tests_safe_allocation.cpp

@@ -1,16 +0,0 @@
-//
-// Created by postaron on 13/12/2025.
-//
-
-#include <iostream>
-#include <catch2/catch_all.hpp>
-
-#include  "../include/malloc_unmanaged.hpp"
-
-TEST_CASE("Managed allocation", "[safe_alloc][0]") {
-    std::cout << "Safely allocates memory in CUDA" << std::endl;
-    const auto [safe_ptr, error] = safe_cuda::cuda_malloc<int, safe_cuda::allocType::Managed>(sizeof(int));
-    REQUIRE(safe_ptr != nullptr);
-    REQUIRE(error == cudaSuccess);
-    std::cout << "Safely deallocates memory in CUDA" << std::endl;
-}