hello cuda world

build.bat

@@ -0,0 +1,24 @@
+@echo off
+
+ctime -begin timeBuild.ctm
+
+set CommonCompilerFlags=/nologo /Zi /FC /Od
+@rem /WX /W4 /wd4201 /wd4100 /wd4189 /wd4244 /wd4127 /wd4456
+@rem
+@rem
+
+set cuda_root=D:/lib/cudatoolkit/lib/x64
+
+set CudaSources=../src/main.cu
+
+IF NOT EXIST .\build mkdir .\build
+pushd .\build
+nvcc %CudaSources% -o program.exe
+
+set LastError=%ERRORLEVEL%
+popd
+
+ctime -end timeBuild.ctm %LastError%
+IF NOT %LastError%==0 GOTO :end
+
+:end

src/editortest.cu

@@ -0,0 +1,12 @@
+
+__global__ void some_global_kernel() {
+
+   some_kernel<<<32, 32>>>(0,0);
+}
+
+
+__device__ void some_kernel(int a, int b)
+{
+
+  return;
+}

src/main.cu

@@ -0,0 +1,104 @@
+#include <stdio.h>
+
+
+typedef int S32;
+typedef unsigned int U32;
+typedef float F32;
+
+
+#define CUDA_CHECK(err) do { \
+  if (err != cudaSuccess) { \
+    fprintf(stderr, "CUDA ERROR: %s at %s:%d\n", \
+        cudaGetErrorString(err), __FILE__, __LINE__); \
+    exit(EXIT_FAILURE); \
+  } \
+} while (0)
+
+
+
+#define LOG printf
+
+
+
+#define NUM_BLOCKS   1
+#define NUM_THREADS  32
+
+
+
+
+
+__device__ void modify_array(U32 *arr, U32 N) 
+{
+  U32 i = threadIdx.x;
+  if(i < N)
+  {
+    arr[i] = i;
+  }
+
+  return;
+}
+
+
+__global__ void hello_from_device(U32 *arr, U32 N) 
+{
+
+  modify_array(arr, N);
+
+
+  return;
+}
+
+
+
+
+int main() 
+{
+
+  LOG("Hello from cpu\n");
+  
+  
+  U32 N = NUM_THREADS;
+  U32 arr_size = N * sizeof(U32);
+
+
+  U32 *arr = (U32 *)malloc(arr_size);
+  memset(arr, 0, arr_size);
+
+  cudaError_t cuErr;
+  U32 *d_arr = 0;
+
+
+
+  cuErr = cudaMalloc(&d_arr, arr_size);
+  CUDA_CHECK(cuErr);
+
+  cuErr = cudaMemcpy(d_arr, arr, arr_size, cudaMemcpyHostToDevice);
+  CUDA_CHECK(cuErr);
+
+  LOG("Array before CUDA \n");
+  for(U32 i = 0; i < N; i += 1) 
+  {
+    LOG("%i: %i \n", i, arr[i]);
+  }
+  LOG("\n");
+
+
+  hello_from_device<<<NUM_BLOCKS, NUM_THREADS>>>(d_arr, N);
+  
+  cuErr = cudaMemcpy(arr, d_arr, arr_size, cudaMemcpyDeviceToHost);
+  CUDA_CHECK(cuErr);
+
+  cuErr = cudaFree(d_arr);
+  CUDA_CHECK(cuErr);
+
+  LOG("Array after CUDA \n");
+  for(U32 i = 0; i < N; i += 1) 
+  {
+    LOG("%i \n", arr[i]);
+  }
+  LOG("\n");
+
+
+  return 0;
+}
+