Why my CUDA application isn't starting? - c++

Question

The below code compiles without any errors, but when I run it, it says "Application was not able to start correctly (0xc000007b). Click OK to close the application.".
#include <math.h>
#include <GL\glew.h>
#include <GL\glut.h>
#include <cuda_gl_interop.h>
#include <cuda_runtime.h>
#include <device_launch_parameters.h>
GLuint vbo;
struct cudaGraphicsResource* vbo_cuda;
unsigned int width, height;
float tim;
__global__ void createVertices(float4* positions, float tim,
unsigned int width, unsigned int height) {
unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;
unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;
float u = x / (float)width;
float v = y / (float)height;
u = u * 2.0f - 1.0f;
v = v * 2.0f - 1.0f;
// calculate simple sine wave pattern
float freq = 4.0f;
float w = sinf(u * freq + tim)
* cosf(v * freq + tim) * 0.5f;
positions[y * width + x] = make_float4(u, w, v, 1.0f);
}
void init(void) {
glClearColor(0, 0, 0, 0);
glShadeModel(GL_FLAT);
}
void reshape(int w, int h) {
glViewport(0, 0, (GLsizei)w, (GLsizei)h);
glMatrixMode(GL_PROJECTION);
glLoadIdentity();
gluPerspective(60, (GLfloat)w/(GLfloat)h, 1, 200);
}
void display() {
float4* positions;
cudaGraphicsMapResources(1, &vbo_cuda, 0);
size_t num_bytes;
cudaGraphicsResourceGetMappedPointer((void**)&positions,
&num_bytes,
vbo_cuda);
// execute kernel
dim3 dimBlock(16, 16, 1);
dim3 dimGrid(width / dimBlock.x, height / dimBlock.y, 1);
createVertices<<<dimGrid, dimBlock>>>(positions, tim,
width, height);
cudaGraphicsUnmapResources(1, &vbo_cuda, 0);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
glMatrixMode(GL_MODELVIEW);
glLoadIdentity();
// render from the vbo
glBindBuffer(GL_ARRAY_BUFFER, vbo);
glVertexPointer(4, GL_FLOAT, 0, 0);
glEnableClientState(GL_VERTEX_ARRAY);
glDrawArrays(GL_POINTS, 0, width * height);
glDisableClientState(GL_VERTEX_ARRAY);
glutSwapBuffers();
glutPostRedisplay();
}
void deleteVBO() {
cudaGraphicsUnregisterResource(vbo_cuda);
glDeleteBuffers(1, &vbo);
}
int main (int argc, char**argv) {
glutInit(&argc, argv);
glutInitDisplayMode(GLUT_RGBA | GLUT_DOUBLE);
glutInitWindowSize(500, 500);
glutInitWindowPosition(100, 100);
glutCreateWindow("Cuda OpenGL Interop");
init();
glutDisplayFunc(display);
glutReshapeFunc(reshape);
cudaGLSetGLDevice(0);
glGenBuffers(1, &vbo);
glBindBuffer(GL_ARRAY_BUFFER, vbo);
unsigned int size = width * height * 4 * sizeof(float);
glBufferData(GL_ARRAY_BUFFER, size, 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_ARRAY_BUFFER, 0);
cudaGLRegisterBufferObject(vbo);
glutMainLoop();
return 0;
}

Answer 1

The error is from Windows: your attempt is falling very short, as the executable you have produced is not valid for windows. It's possible you are using DEBUG DLL's with a RELEASE build. Or that you are mixing a 32 bit build with 64 bit DLL's, or many other odd combinations... (64 bit exe on 32 bit system?....)
Usually you can get more information regarding DLL problems looking in the Windows event viewer, but if you start running your application in the debugger (for sure with visual studio) you will get more information on your error.
If you can't understand what's wrong, you can try to find what is failing with http://www.dependencywalker.com/.

Answer 2

1st error, which was the reason to start this thread, disappeared by installing the right glew32.dll libraries into the right folders.
2nd error, where the debugger stopped at glGenBuffers(1, vbo), was because I forgot about glewInit()
Below you can find the working application:
#include <math.h>
#include <GL\glew.h>
#include <GL\glut.h>
#include <cuda_gl_interop.h>
#include <cuda_runtime.h>
#include <device_launch_parameters.h>
GLuint vbo;
struct cudaGraphicsResource* vbo_cuda;
const unsigned int window_width = 512;
const unsigned int window_height = 512;
const unsigned int mesh_width = 256;
const unsigned int mesh_height = 256;
float tim = 0.0;
__global__ void createVertices(float4* positions, float tim,
unsigned int mesh_width, unsigned int mesh_height) {
unsigned int x = blockIdx.x * blockDim.x + threadIdx.x;
unsigned int y = blockIdx.y * blockDim.y + threadIdx.y;
float u = x / (float)mesh_width;
float v = y / (float)mesh_height;
u = u * 2.0f - 1.0f;
v = v * 2.0f - 1.0f;
// calculate simple sine wave pattern
float freq = 4.0f;
float w = sinf(u * freq + tim)
* cosf(v * freq + tim) * 0.5f;
positions[y * mesh_width + x] = make_float4(u, w, v, 1.0f);
}
void runCuda(GLuint vbo)
{
// map OpenGL buffer object for writing from CUDA
float4* positions;
cudaGraphicsMapResources(1, &vbo_cuda, 0);
size_t num_bytes;
cudaGraphicsResourceGetMappedPointer((void**)&positions,
&num_bytes,
vbo_cuda);
// execute kernel
dim3 dimBlock(16, 16, 1);
dim3 dimGrid(mesh_width / dimBlock.x, mesh_height / dimBlock.y, 1);
createVertices<<<dimGrid, dimBlock>>>(positions, tim,
mesh_width, mesh_height);
cudaGraphicsUnmapResources(1, &vbo_cuda, 0);
}
void init(void) {
glewInit();
glClearColor(0, 0, 0, 1);
glDisable(GL_DEPTH_TEST);
}
void reshape(int w, int h) {
// viewport
glViewport(0, 0, w, h);
// projection
glMatrixMode(GL_PROJECTION);
glLoadIdentity();
gluPerspective(60, (GLfloat)w/(GLfloat)h, 0.1, 10);
}
void createVBO(GLuint* vbo) {
// create buffer object
glGenBuffers(1, vbo);
glBindBuffer(GL_ARRAY_BUFFER, *vbo);
// initialize buffer object
unsigned int size = mesh_width * mesh_height * 4 * sizeof(float);
glBufferData(GL_ARRAY_BUFFER, size, 0, GL_DYNAMIC_DRAW);
glBindBuffer(GL_ARRAY_BUFFER, 0);
cudaGLRegisterBufferObject(*vbo);
}
void deleteVBO(GLuint* vbo) {
cudaGraphicsUnregisterResource(vbo_cuda);
glBindBuffer(1, *vbo);
glDeleteBuffers(1, vbo);
cudaGLUnregisterBufferObject(*vbo);
}
void display() {
// run CUDA kernel to generate vertex positions
runCuda(vbo);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
// set view matrix
glMatrixMode(GL_MODELVIEW);
glLoadIdentity();
// render from the vbo
glBindBuffer(GL_ARRAY_BUFFER, vbo);
glVertexPointer(4, GL_FLOAT, 0, 0);
glEnableClientState(GL_VERTEX_ARRAY);
glColor3f(1, 0, 0);
glDrawArrays(GL_POINTS, 0, mesh_width * mesh_height);
glDisableClientState(GL_VERTEX_ARRAY);
glutSwapBuffers();
glutPostRedisplay();
tim+=1;
}
void keyboard(unsigned char key, int x, int y)
{
switch(key) {
case(27) :
deleteVBO(&vbo);
exit(0);
}
}
int main (int argc, char**argv) {
glutInit(&argc, argv);
glutInitDisplayMode(GLUT_RGBA | GLUT_DOUBLE);
glutInitWindowSize(window_width, window_height);
glutInitWindowPosition(100, 100);
glutCreateWindow("Cuda GL interop");
init();
glutDisplayFunc(display);
glutKeyboardFunc(keyboard);
glutReshapeFunc(reshape);
// create VBO
createVBO(&vbo);
// run the cuda part
runCuda(vbo);
cudaGLSetGLDevice(0);
glutMainLoop();
return 0;
}