「CUDA-By Example」教科書の CUDA-OpenGL 相互運用例によって生成された画像を、画像に保存できるメモリ バッファーに保存しようとしています。
緑の「X」とオレンジがかった「X」の 2 つの画像をメモリ バッファに保存したいと考えています。OpenGL で pBuffer をレンダリングすると、出力例のように緑色の「X」イメージが表示されるはずですが、黒い画面が表示されます。正しい出力が得られない理由がわかりません。誰かが何が悪いのか教えてもらえますか?
A Memory buffer for multiple imagesからメモリ バッファのコードを取得しました
#include "book.h"
#include "cpu_bitmap.h"
#include "cuda.h"
#include <cuda_gl_interop.h>
PFNGLBINDBUFFERARBPROC glBindBuffer = NULL;
PFNGLDELETEBUFFERSARBPROC glDeleteBuffers = NULL;
PFNGLGENBUFFERSARBPROC glGenBuffers = NULL;
PFNGLBUFFERDATAARBPROC glBufferData = NULL;
#define DIM 512
#define IMAGESIZE_MAX (DIM*DIM) // MY CHANGE
GLuint bufferObj;
cudaGraphicsResource *resource;
// based on ripple code, but uses uchar4 which is the type of data
// graphic inter op uses. see screenshot - basic2.png
__global__ void kernel( uchar4 *ptr1)
{
// map from threadIdx/BlockIdx to pixel position
int x = threadIdx.x + blockIdx.x * blockDim.x;
int y = threadIdx.y + blockIdx.y * blockDim.y;
int offset = x + y * blockDim.x * gridDim.x ;
// now calculate the value at that position
float fx = x/(float)DIM - 0.5f;
float fy = y/(float)DIM - 0.5f;
unsigned char green = 128 + 127 * tan( abs(fx*100) - abs(fy*100) );
// accessing uchar4 vs unsigned char*
ptr1[offset].x = 0;
ptr1[offset].y = green;
ptr1[offset].z = 0;
ptr1[offset].w = 255;
}
// MY CODE
__global__ void kernel2( uchar4 *ptr2)
{
// map from threadIdx/BlockIdx to pixel position
int x = threadIdx.x + blockIdx.x * blockDim.x;
int y = threadIdx.y + blockIdx.y * blockDim.y;
int offset = x + y * blockDim.x * gridDim.x ;
// now calculate the value at that position
float fx = x/(float)DIM - 0.5f;
float fy = y/(float)DIM - 0.5f;
unsigned char green = 128 + 127 * tan( abs(fx*100) - abs(fy*100) );
// accessing uchar4 vs unsigned char*
ptr2[offset].x = 1000;
ptr2[offset].y = green;
ptr2[offset].z = 0;
ptr2[offset].w = 255;
}
__global__ void copy ( uchar4 *pBuffer, uchar4 *Ptr, uchar4 *Ptr2, size_t size, int a )
{
int x = threadIdx.x + blockIdx.x * blockDim.x;
int y = threadIdx.y + blockIdx.y * blockDim.y;
int idx = x + y * blockDim.x * gridDim.x ;
int bdx = idx;
if (a==1)
{
while ( idx < DIM*DIM)
{
pBuffer[idx] = Ptr[idx] ;
__syncthreads();
if (idx==DIM*DIM)
{
break;
}
}
}
if (a==2)
{
while ( (idx < DIM*DIM) && (bdx < DIM*DIM) )
{
uchar4 temp = Ptr2[bdx];
__syncthreads();
pBuffer[idx+4] = temp;
__syncthreads();
if ((idx==DIM*DIM) && (bdx==DIM*DIM))
{
break;
}
}
}
}
void key_func( unsigned char key, int x, int y )
{
switch (key)
{
case 27:
// clean up OpenGL and CUDA
( cudaGraphicsUnregisterResource( resource ) );
glBindBuffer( GL_PIXEL_UNPACK_BUFFER_ARB, 0 );
glDeleteBuffers( 1, &bufferObj );
exit(0);
}
}
void draw_func( void ) {
// we pass zero as the last parameter, because out bufferObj is now
// the source, and the field switches from being a pointer to a
// bitmap to now mean an offset into a bitmap object
glDrawPixels( DIM, DIM, GL_RGBA, GL_UNSIGNED_BYTE, 0 );
glutSwapBuffers();
}
int main( int argc, char **argv )
{
cudaDeviceProp prop;
int dev;
memset( &prop, 0, sizeof( cudaDeviceProp ) );
prop.major = 1;
prop.minor = 0;
( cudaChooseDevice( &dev, &prop ) );
// tell CUDA which dev we will be using for graphic interop
// from the programming guide: Interoperability with OpenGL
// requires that the CUDA device be specified by
// cudaGLSetGLDevice() before any other runtime calls.
( cudaGLSetGLDevice( dev ) );
// these GLUT calls need to be made before the other OpenGL
// calls, else we get a seg fault
glutInit( &argc, argv );
glutInitDisplayMode( GLUT_DOUBLE | GLUT_RGBA );
glutInitWindowSize( DIM, DIM );
glutCreateWindow( "bitmap" );
glBindBuffer = (PFNGLBINDBUFFERARBPROC)GET_PROC_ADDRESS("glBindBuffer");
glDeleteBuffers = (PFNGLDELETEBUFFERSARBPROC)GET_PROC_ADDRESS("glDeleteBuffers");
glGenBuffers = (PFNGLGENBUFFERSARBPROC)GET_PROC_ADDRESS("glGenBuffers");
glBufferData = (PFNGLBUFFERDATAARBPROC)GET_PROC_ADDRESS("glBufferData");
// the first three are standard OpenGL, the 4th is the CUDA reg
// of the bitmap these calls exist starting in OpenGL 1.5
glGenBuffers( 1, &bufferObj );
glBindBuffer( GL_PIXEL_UNPACK_BUFFER_ARB, bufferObj );
glBufferData( GL_PIXEL_UNPACK_BUFFER_ARB, DIM * DIM * 4 ,NULL, GL_DYNAMIC_DRAW_ARB );
// REGISTER THE GL BufferObj and CUDA Resource
( cudaGraphicsGLRegisterBuffer( &resource, bufferObj, cudaGraphicsMapFlagsNone ) );
// do work with the memory dst being on the GPU, gotten via mapping
HANDLE_ERROR( cudaGraphicsMapResources( 1, &resource, NULL ) );
// MY MODIFIED CODE
uchar4 *devPtr;
size_t size;
size_t sizeTotal = 0;
cudaMalloc ( (uchar4 **)&devPtr, size);
uchar4 *devPtr2;
cudaMalloc ( (uchar4 **)&devPtr2, size);
uchar4 *pBuffer;
(cudaMalloc ( (uchar4 **)&pBuffer, size));
uchar4 *pBufferCurrent;
(cudaMalloc ( (uchar4 **)&pBufferCurrent, size));
uchar4 *pBufferImage;
(cudaMalloc ( (uchar4 **)&pBufferImage, size));
// REGISTER THE C BUFFER and CUDA Resource
HANDLE_ERROR(
cudaGraphicsResourceGetMappedPointer( (void**)&pBufferImage,
&size,
resource) );
dim3 grids(DIM/16,DIM/16);
dim3 threads(16,16);
kernel<<<grids,threads>>>(devPtr);
kernel2<<<grids,threads>>>(devPtr2);
int a = 1;
do
{
if (a==1)
{
copy<<< grids, threads>>>(pBufferImage, devPtr, devPtr2, size, a);
}
if(a==2)
{
copy<<< grids, threads>>>(pBufferImage, devPtr, devPtr2, size, a);
}
a++;
} while (a<=2);
cudaGraphicsUnmapResources( 1, &resource, NULL ) );
// set up GLUT and kick off main loop
glutKeyboardFunc( key_func );
glutDisplayFunc( draw_func );
glutMainLoop();
}