static BufferData *
loadFile (InputStream *stream)
{
const char *fileName;
Int32BigEndian magic;
/* Raw files have no magic number - so use the fileName extension */
fileName = _alutInputStreamGetFileName (stream);
if (fileName != NULL && hasSuffixIgnoringCase (fileName, ".raw"))
{
return loadRawFile (stream);
}
/* For other file formats, read the quasi-standard four byte magic number */
if (!_alutInputStreamReadInt32BE (stream, &magic))
{
return AL_FALSE;
}
/* Magic number 'RIFF' == Microsoft '.wav' format */
if (magic == 0x52494646)
{
return loadWavFile (stream);
}
/* Magic number '.snd' == Sun & Next's '.au' format */
if (magic == 0x2E736E64)
{
return loadAUFile (stream);
}
_alutSetError (ALUT_ERROR_UNSUPPORTED_FILE_TYPE);
return AL_FALSE;
}
void loadVolumeData(char *exec_path)
{
// load volume data
const char *path = sdkFindFilePath(volumeFilename, exec_path);
if (path == NULL)
{
fprintf(stderr, "Error unable to find 3D Volume file: '%s'\n", volumeFilename);
exit(EXIT_FAILURE);
}
size_t size = volumeSize.width*volumeSize.height*volumeSize.depth;
uchar *h_volume = loadRawFile(path, size);
initCuda(h_volume, volumeSize);
sdkCreateTimer(&timer);
free(h_volume);
}
////////////////////////////////////////////////////////////////////////////////
// Program main
////////////////////////////////////////////////////////////////////////////////
int
main( int argc, char** argv)
{
//start logs
shrSetLogFileName ("volumeRender.txt");
shrLog("%s Starting...\n\n", argv[0]);
if (cutCheckCmdLineFlag(argc, (const char **)argv, "qatest") ||
cutCheckCmdLineFlag(argc, (const char **)argv, "noprompt"))
{
g_bQAReadback = true;
fpsLimit = frameCheckNumber;
}
if (cutCheckCmdLineFlag(argc, (const char **)argv, "glverify"))
{
g_bQAGLVerify = true;
fpsLimit = frameCheckNumber;
}
if (g_bQAReadback) {
// use command-line specified CUDA device, otherwise use device with highest Gflops/s
if( cutCheckCmdLineFlag(argc, (const char**)argv, "device") ) {
cutilDeviceInit(argc, argv);
} else {
cudaSetDevice( cutGetMaxGflopsDeviceId() );
}
} else {
// First initialize OpenGL context, so we can properly set the GL for CUDA.
// This is necessary in order to achieve optimal performance with OpenGL/CUDA interop.
initGL( &argc, argv );
// use command-line specified CUDA device, otherwise use device with highest Gflops/s
if( cutCheckCmdLineFlag(argc, (const char**)argv, "device") ) {
cutilGLDeviceInit(argc, argv);
} else {
cudaGLSetGLDevice( cutGetMaxGflopsDeviceId() );
}
/*
int device;
struct cudaDeviceProp prop;
cudaGetDevice( &device );
cudaGetDeviceProperties( &prop, device );
if( !strncmp( "Tesla", prop.name, 5 ) ) {
shrLog("This sample needs a card capable of OpenGL and display.\n");
shrLog("Please choose a different device with the -device=x argument.\n");
cutilExit(argc, argv);
}
*/
}
// parse arguments
char *filename;
if (cutGetCmdLineArgumentstr( argc, (const char**) argv, "file", &filename)) {
volumeFilename = filename;
}
int n;
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "size", &n)) {
volumeSize.width = volumeSize.height = volumeSize.depth = n;
}
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "xsize", &n)) {
volumeSize.width = n;
}
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "ysize", &n)) {
volumeSize.height = n;
}
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "zsize", &n)) {
volumeSize.depth = n;
}
// load volume data
char* path = shrFindFilePath(volumeFilename, argv[0]);
if (path == 0) {
shrLog("Error finding file '%s'\n", volumeFilename);
exit(EXIT_FAILURE);
}
size_t size = volumeSize.width*volumeSize.height*volumeSize.depth*sizeof(VolumeType);
void *h_volume = loadRawFile(path, size);
initCuda(h_volume, volumeSize);
free(h_volume);
cutilCheckError( cutCreateTimer( &timer));
shrLog("Press '=' and '-' to change density\n"
" ']' and '[' to change brightness\n"
" ';' and ''' to modify transfer function offset\n"
" '.' and ',' to modify transfer function scale\n\n");
// calculate new grid size
gridSize = dim3(iDivUp(width, blockSize.x), iDivUp(height, blockSize.y));
if (g_bQAReadback) {
g_CheckRender = new CheckBackBuffer(width, height, 4, false);
g_CheckRender->setPixelFormat(GL_RGBA);
g_CheckRender->setExecPath(argv[0]);
g_CheckRender->EnableQAReadback(true);
uint *d_output;
cutilSafeCall(cudaMalloc((void**)&d_output, width*height*sizeof(uint)));
cutilSafeCall(cudaMemset(d_output, 0, width*height*sizeof(uint)));
float modelView[16] =
{
1.0f, 0.0f, 0.0f, 0.0f,
0.0f, 1.0f, 0.0f, 0.0f,
0.0f, 0.0f, 1.0f, 0.0f,
0.0f, 0.0f, 4.0f, 1.0f
};
invViewMatrix[0] = modelView[0]; invViewMatrix[1] = modelView[4]; invViewMatrix[2] = modelView[8]; invViewMatrix[3] = modelView[12];
invViewMatrix[4] = modelView[1]; invViewMatrix[5] = modelView[5]; invViewMatrix[6] = modelView[9]; invViewMatrix[7] = modelView[13];
invViewMatrix[8] = modelView[2]; invViewMatrix[9] = modelView[6]; invViewMatrix[10] = modelView[10]; invViewMatrix[11] = modelView[14];
// call CUDA kernel, writing results to PBO
copyInvViewMatrix(invViewMatrix, sizeof(float4)*3);
// Start timer 0 and process n loops on the GPU
int nIter = 10;
for (int i = -1; i < nIter; i++)
{
if( i == 0 ) {
cudaThreadSynchronize();
cutStartTimer(timer);
}
render_kernel(gridSize, blockSize, d_output, width, height, density, brightness, transferOffset, transferScale);
}
cudaThreadSynchronize();
cutStopTimer(timer);
// Get elapsed time and throughput, then log to sample and master logs
double dAvgTime = cutGetTimerValue(timer)/(nIter * 1000.0);
shrLogEx(LOGBOTH | MASTER, 0, "volumeRender, Throughput = %.4f MTexels/s, Time = %.5f s, Size = %u Texels, NumDevsUsed = %u, Workgroup = %u\n",
(1.0e-6 * width * height)/dAvgTime, dAvgTime, (width * height), 1, blockSize.x * blockSize.y);
cutilCheckMsg("Error: render_kernel() execution FAILED");
cutilSafeCall( cudaThreadSynchronize() );
cutilSafeCall( cudaMemcpy(g_CheckRender->imageData(), d_output, width*height*4, cudaMemcpyDeviceToHost) );
g_CheckRender->savePPM(sOriginal[g_Index], true, NULL);
if (!g_CheckRender->PPMvsPPM(sOriginal[g_Index], sReference[g_Index], MAX_EPSILON_ERROR, THRESHOLD)) {
shrLog("\nFAILED\n\n");
} else {
shrLog("\nPASSED\n\n");
}
cudaFree(d_output);
freeCudaBuffers();
if (g_CheckRender) {
delete g_CheckRender; g_CheckRender = NULL;
}
} else {
// This is the normal rendering path for VolumeRender
glutDisplayFunc(display);
glutKeyboardFunc(keyboard);
glutMouseFunc(mouse);
glutMotionFunc(motion);
glutReshapeFunc(reshape);
glutIdleFunc(idle);
initPixelBuffer();
if (g_bQAGLVerify) {
g_CheckRender = new CheckBackBuffer(width, height, 4);
g_CheckRender->setPixelFormat(GL_RGBA);
g_CheckRender->setExecPath(argv[0]);
g_CheckRender->EnableQAReadback(true);
}
atexit(cleanup);
glutMainLoop();
}
cudaThreadExit();
shrEXIT(argc, (const char**)argv);
}
void initData(int argc, char **argv)
{
// parse arguments
char *filename;
if (getCmdLineArgumentString(argc, (const char **) argv, "file", &filename))
{
volumeFilename = filename;
}
int n;
if (checkCmdLineFlag(argc, (const char **) argv, "size"))
{
n = getCmdLineArgumentInt(argc, (const char **) argv, "size");
volumeSize.width = volumeSize.height = volumeSize.depth = n;
}
if (checkCmdLineFlag(argc, (const char **) argv, "xsize"))
{
n = getCmdLineArgumentInt(argc, (const char **) argv, "xsize");
volumeSize.width = n;
}
if (checkCmdLineFlag(argc, (const char **) argv, "ysize"))
{
n = getCmdLineArgumentInt(argc, (const char **) argv, "ysize");
volumeSize.height = n;
}
if (checkCmdLineFlag(argc, (const char **) argv, "zsize"))
{
n = getCmdLineArgumentInt(argc, (const char **) argv, "zsize");
volumeSize.depth = n;
}
char *path = sdkFindFilePath(volumeFilename, argv[0]);
if (path == 0)
{
printf("Error finding file '%s'\n", volumeFilename);
exit(EXIT_FAILURE);
}
size_t size = volumeSize.width*volumeSize.height*volumeSize.depth*sizeof(VolumeType);
void *h_volume = loadRawFile(path, size);
FilterKernel_init();
Volume_init(&volumeOriginal,volumeSize, h_volume, 0);
free(h_volume);
Volume_init(&volumeFilter0, volumeSize, NULL, 1);
Volume_init(&volumeFilter1, volumeSize, NULL, 1);
VolumeRender_init();
VolumeRender_setPreIntegrated(preIntegrated);
VolumeRender_setVolume(&volumeOriginal);
sdkCreateTimer(&timer);
sdkCreateTimer(&animationTimer);
sdkStartTimer(&animationTimer);
// calculate new grid size
gridSize = dim3(iDivUp(width, blockSize.x), iDivUp(height, blockSize.y));
}
////////////////////////////////////////////////////////////////////////////////
// initialize marching cubes
////////////////////////////////////////////////////////////////////////////////
void
initMC(int argc, char** argv)
{
// parse command line arguments
int n;
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "grid", &n)) {
gridSizeLog2.x = gridSizeLog2.y = gridSizeLog2.z = n;
}
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "gridx", &n)) {
gridSizeLog2.x = n;
}
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "gridy", &n)) {
gridSizeLog2.y = n;
}
if (cutGetCmdLineArgumenti( argc, (const char**) argv, "gridz", &n)) {
gridSizeLog2.z = n;
}
char *filename;
if (cutGetCmdLineArgumentstr( argc, (const char**) argv, "file", &filename)) {
volumeFilename = filename;
}
gridSize = make_uint3(1<<gridSizeLog2.x, 1<<gridSizeLog2.y, 1<<gridSizeLog2.z);
gridSizeMask = make_uint3(gridSize.x-1, gridSize.y-1, gridSize.z-1);
gridSizeShift = make_uint3(0, gridSizeLog2.x, gridSizeLog2.x+gridSizeLog2.y);
numVoxels = gridSize.x*gridSize.y*gridSize.z;
voxelSize = make_float3(2.0f / gridSize.x, 2.0f / gridSize.y, 2.0f / gridSize.z);
maxVerts = gridSize.x*gridSize.y*100;
printf("grid: %d x %d x %d = %d voxels\n", gridSize.x, gridSize.y, gridSize.z, numVoxels);
printf("max verts = %d\n", maxVerts);
#if SAMPLE_VOLUME
// load volume data
char* path = cutFindFilePath(volumeFilename, argv[0]);
if (path == 0) {
fprintf(stderr, "Error finding file '%s'\n", volumeFilename);
cudaThreadExit();
exit(EXIT_FAILURE);
}
int size = gridSize.x*gridSize.y*gridSize.z*sizeof(uchar);
uchar *volume = loadRawFile(path, size);
cutilSafeCall(cudaMalloc((void**) &d_volume, size));
cutilSafeCall(cudaMemcpy(d_volume, volume, size, cudaMemcpyHostToDevice) );
free(volume);
bindVolumeTexture(d_volume);
#endif
if (g_bQAReadback) {
cudaMalloc((void **)&(d_pos), maxVerts*sizeof(float)*4);
cudaMalloc((void **)&(d_normal), maxVerts*sizeof(float)*4);
} else {
// create VBOs
createVBO(&posVbo, maxVerts*sizeof(float)*4);
// DEPRECATED: cutilSafeCall( cudaGLRegisterBufferObject(posVbo) );
cutilSafeCall(cudaGraphicsGLRegisterBuffer(&cuda_posvbo_resource, posVbo,
cudaGraphicsMapFlagsWriteDiscard));
createVBO(&normalVbo, maxVerts*sizeof(float)*4);
// DEPRECATED: cutilSafeCall(cudaGLRegisterBufferObject(normalVbo));
cutilSafeCall(cudaGraphicsGLRegisterBuffer(&cuda_normalvbo_resource, normalVbo,
cudaGraphicsMapFlagsWriteDiscard));
}
// allocate textures
allocateTextures( &d_edgeTable, &d_triTable, &d_numVertsTable );
// allocate device memory
unsigned int memSize = sizeof(uint) * numVoxels;
cutilSafeCall(cudaMalloc((void**) &d_voxelVerts, memSize));
cutilSafeCall(cudaMalloc((void**) &d_voxelVertsScan, memSize));
cutilSafeCall(cudaMalloc((void**) &d_voxelOccupied, memSize));
cutilSafeCall(cudaMalloc((void**) &d_voxelOccupiedScan, memSize));
cutilSafeCall(cudaMalloc((void**) &d_compVoxelArray, memSize));
// initialize CUDPP scan
CUDPPConfiguration config;
config.algorithm = CUDPP_SCAN;
config.datatype = CUDPP_UINT;
config.op = CUDPP_ADD;
config.options = CUDPP_OPTION_FORWARD | CUDPP_OPTION_EXCLUSIVE;
cudppPlan(&scanplan, config, numVoxels, 1, 0);
}
