stotko · stotko · Nov 21, 2019 · Nov 21, 2019 · Nov 21, 2019
diff --git a/examples/CMakeLists.txt b/examples/CMakeLists.txt
@@ -1,24 +1,20 @@
+
+# Input parameters:
+# - File name of the example without file extension
+# - File extension of the example
 macro(stdgpu_detail_add_example)
     set(STDGPU_EXAMPLES_NAME "${ARGV0}")
     add_executable(${STDGPU_EXAMPLES_NAME} "${STDGPU_EXAMPLES_NAME}.${ARGV1}")
     target_link_libraries(${STDGPU_EXAMPLES_NAME} PRIVATE stdgpu::stdgpu)
 endmacro()
 
-macro(stdgpu_add_example_cu)
-    stdgpu_detail_add_example(${ARGV0} "cu")
-endmacro()
-
 macro(stdgpu_add_example_cpp)
     stdgpu_detail_add_example(${ARGV0} "cpp")
 endmacro()
 
 
-stdgpu_add_example_cu(bitset)
-stdgpu_add_example_cu(container_iterator)
-stdgpu_add_example_cu(container_kernel)
 stdgpu_add_example_cpp(contract)
-stdgpu_add_example_cu(createAndDestroyDeviceArray)
-stdgpu_add_example_cu(createAndDestroyDeviceObject)
-stdgpu_add_example_cu(mutex_array)
-stdgpu_add_example_cu(thrust_interoperability)
-stdgpu_add_example_cu(thrust_towards_ranges)
+stdgpu_add_example_cpp(createAndDestroyDeviceArray)
+stdgpu_add_example_cpp(createAndDestroyDeviceObject)
+
+add_subdirectory(cuda)
diff --git a/examples/createAndDestroyDeviceArray.cu → examples/createAndDestroyDeviceArray.cpp b/examples/createAndDestroyDeviceArray.cu → examples/createAndDestroyDeviceArray.cpp
diff --git a/examples/createAndDestroyDeviceObject.cu → examples/createAndDestroyDeviceObject.cpp b/examples/createAndDestroyDeviceObject.cu → examples/createAndDestroyDeviceObject.cpp
@@ -13,7 +13,7 @@
  *  limitations under the License.
  */
 
-#include <stdgpu/memory.h>          // createDeviceArray, destroyDeviceArray
+#include <stdgpu/memory.h>          // createDeviceArray, destroyDeviceArray, createHostArray, destroyHostArray
 #include <stdgpu/platform.h>        // STDGPU_HOST_DEVICE
 
 
@@ -44,6 +44,27 @@ class Image
             device_object._height = 0;
         }
 
+        static Image
+        createHostObject(const stdgpu::index_t width,
+                         const stdgpu::index_t height)
+        {
+            Image result;
+
+            result._values = createHostArray<std::uint8_t>(width * height);
+            result._width = width;
+            result._height = height;
+
+            return result;
+        }
+
+        static void
+        destroyHostObject(Image& host_object)
+        {
+            destroyHostArray<std::uint8_t>(host_object._values);
+            host_object._width = 0;
+            host_object._height = 0;
+        }
+
         // Further (static) member functions ...
 
         STDGPU_HOST_DEVICE std::uint8_t&
@@ -72,32 +93,29 @@ class Image
 };
 
 
-__global__ void
-fill_image(Image d_image)
+void
+fill_image(Image image)
 {
-    stdgpu::index_t i = blockIdx.x * blockDim.x + threadIdx.x;
-    stdgpu::index_t j = blockIdx.y * blockDim.y + threadIdx.y;
-
-    if (i >= d_image.width() || j >= d_image.height()) return;
-
-    std::uint8_t value = (i * i + j * j) % (1 << 8);
+    for (stdgpu::index_t i = 0; i < image.width(); ++i)
+    {
+        for (stdgpu::index_t j = 0; j < image.height(); ++j)
+        {
+            std::uint8_t value = (i * i + j * j) % (1 << 8);
 
-    d_image(i, j) = value;
+            image(i, j) = value;
+        }
+    }
 }
 
 
 int
 main()
 {
-    Image d_image = Image::createDeviceObject(1920, 1080);
-
-    dim3 threads(32, 8);
-    dim3 blocks((d_image.width() + threads.x - 1) / threads.x, (d_image.height() + threads.y - 1) / threads.y);
+    Image image = Image::createHostObject(1920, 1080);
 
-    fill_image<<< blocks, threads >>>(d_image);
-    cudaDeviceSynchronize();
+    fill_image(image);
 
-    Image::destroyDeviceObject(d_image);
+    Image::destroyHostObject(image);
 }
 
 
diff --git a/examples/cuda/CMakeLists.txt b/examples/cuda/CMakeLists.txt
@@ -0,0 +1,11 @@
+
+macro(stdgpu_add_example_cu)
+    stdgpu_detail_add_example(${ARGV0} "cu")
+endmacro()
+
+stdgpu_add_example_cu(bitset)
+stdgpu_add_example_cu(container_iterator)
+stdgpu_add_example_cu(container_kernel)
+stdgpu_add_example_cu(mutex_array)
+stdgpu_add_example_cu(thrust_interoperability)
+stdgpu_add_example_cu(thrust_towards_ranges)
diff --git a/examples/bitset.cu → examples/cuda/bitset.cu b/examples/bitset.cu → examples/cuda/bitset.cu
diff --git a/examples/container_iterator.cu → examples/cuda/container_iterator.cu b/examples/container_iterator.cu → examples/cuda/container_iterator.cu
diff --git a/examples/container_kernel.cu → examples/cuda/container_kernel.cu b/examples/container_kernel.cu → examples/cuda/container_kernel.cu
diff --git a/examples/mutex_array.cu → examples/cuda/mutex_array.cu b/examples/mutex_array.cu → examples/cuda/mutex_array.cu
diff --git a/examples/thrust_interoperability.cu → examples/cuda/thrust_interoperability.cu b/examples/thrust_interoperability.cu → examples/cuda/thrust_interoperability.cu
@@ -13,6 +13,7 @@
  *  limitations under the License.
  */
 
+#include <iostream>
 #include <thrust/reduce.h>
 #include <thrust/sequence.h>
 #include <thrust/transform.h>

diff --git a/examples/thrust_towards_ranges.cu → examples/cuda/thrust_towards_ranges.cu b/examples/thrust_towards_ranges.cu → examples/cuda/thrust_towards_ranges.cu
@@ -13,6 +13,7 @@
  *  limitations under the License.
  */
 
+#include <iostream>
 #include <thrust/for_each.h>
 #include <thrust/sequence.h>
 #include <thrust/transform.h>