generic-library/opencv
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

fixes for gpu module:

Browse Source 
- fixed printCudaDeviceInfo for new CC
- fixed some compilation errors and warnings
- removed unset command from CMake script
- removed unused std imports
This commit is contained in:
Vladislav Vinogradov
2013-01-23 13:59:14 +04:00
parent b7e6b5af1b
commit ae6266e101
32 changed files with 232 additions and 144 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
modules/gpu/src/cuda/ccomponetns.cu
View File

@@ -497,6 +497,7 @@ namespace cv { namespace gpu { namespace device
void labelComponents(const PtrStepSzb& edges, PtrStepSzi comps, int flags, cudaStream_t stream)
{
(void) flags;
dim3 block(CTA_SIZE_X, CTA_SIZE_Y);
dim3 grid(divUp(edges.cols, TILE_COLS), divUp(edges.rows, TILE_ROWS));
@@ -529,4 +530,4 @@ namespace cv { namespace gpu { namespace device
}
} } }
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

3
modules/gpu/src/cuda/gftt.cu
View File

@@ -47,6 +47,7 @@
#if !defined CUDA_DISABLER
#include <thrust/device_ptr.h>
#include <thrust/sort.h>
#include "opencv2/gpu/device/common.hpp"
@@ -148,4 +149,4 @@ namespace cv { namespace gpu { namespace device
}}}
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

4
modules/gpu/src/cuda/hough.cu
View File

@@ -42,7 +42,9 @@
#if !defined CUDA_DISABLER
#include <thrust/device_ptr.h>
#include <thrust/sort.h>
#include "opencv2/gpu/device/common.hpp"
#include "opencv2/gpu/device/emulation.hpp"
#include "opencv2/gpu/device/vec_math.hpp"
@@ -1509,4 +1511,4 @@ namespace cv { namespace gpu { namespace device
}}}
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

2
modules/gpu/src/cuda/lbp.cu
View File

@@ -295,7 +295,7 @@ namespace cv { namespace gpu { namespace device
int grid = divUp(workAmount, block);
cudaFuncSetCacheConfig(lbp_cascade, cudaFuncCachePreferL1);
Cascade cascade((Stage*)mstages.ptr(), nstages, (ClNode*)mnodes.ptr(), mleaves.ptr(), msubsets.ptr(), (uchar4*)mfeatures.ptr(), subsetSize);
lbp_cascade<<<grid, block>>>(cascade, frameW, frameH, windowW, windowH, initialScale, factor, workAmount, integral.ptr(), integral.step / sizeof(int), objects, classified);
lbp_cascade<<<grid, block>>>(cascade, frameW, frameH, windowW, windowH, initialScale, factor, workAmount, integral.ptr(), (int)integral.step / sizeof(int), objects, classified);
}
}
}}}

8
modules/gpu/src/cuda/mathfunc.cu
View File

@@ -76,7 +76,7 @@ namespace cv { namespace gpu { namespace device
static __device__ __forceinline__ void calc(int x, int y, float x_data, float y_data, float* dst, size_t dst_step, float scale)
{
float angle = ::atan2f(y_data, x_data);
angle += (angle < 0) * 2.0 * CV_PI;
angle += (angle < 0) * 2.0f * CV_PI_F;
dst[y * dst_step + x] = scale * angle;
}
};
@@ -140,7 +140,7 @@ namespace cv { namespace gpu { namespace device
grid.x = divUp(x.cols, threads.x);
grid.y = divUp(x.rows, threads.y);
const float scale = angleInDegrees ? (float)(180.0f / CV_PI) : 1.f;
const float scale = angleInDegrees ? (180.0f / CV_PI_F) : 1.f;
cartToPolar<Mag, Angle><<<grid, threads, 0, stream>>>(
x.data, x.step/x.elemSize(), y.data, y.step/y.elemSize(),
@@ -190,7 +190,7 @@ namespace cv { namespace gpu { namespace device
grid.x = divUp(mag.cols, threads.x);
grid.y = divUp(mag.rows, threads.y);
const float scale = angleInDegrees ? (float)(CV_PI / 180.0f) : 1.0f;
const float scale = angleInDegrees ? (CV_PI_F / 180.0f) : 1.0f;
polarToCart<Mag><<<grid, threads, 0, stream>>>(mag.data, mag.step/mag.elemSize(),
angle.data, angle.step/angle.elemSize(), scale, x.data, x.step/x.elemSize(), y.data, y.step/y.elemSize(), mag.cols, mag.rows);
@@ -214,4 +214,4 @@ namespace cv { namespace gpu { namespace device
} // namespace mathfunc
}}} // namespace cv { namespace gpu { namespace device
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

22
modules/gpu/src/cuda/optical_flow.cu
View File

@@ -164,40 +164,40 @@ namespace cv { namespace gpu { namespace device
r = ::fmin(r, 2.5f);
v[1].x = arrow_x + r * ::cosf(theta - CV_PI / 2.0f);
v[1].y = arrow_y + r * ::sinf(theta - CV_PI / 2.0f);
v[1].x = arrow_x + r * ::cosf(theta - CV_PI_F / 2.0f);
v[1].y = arrow_y + r * ::sinf(theta - CV_PI_F / 2.0f);
v[4].x = arrow_x + r * ::cosf(theta + CV_PI / 2.0f);
v[4].y = arrow_y + r * ::sinf(theta + CV_PI / 2.0f);
v[4].x = arrow_x + r * ::cosf(theta + CV_PI_F / 2.0f);
v[4].y = arrow_y + r * ::sinf(theta + CV_PI_F / 2.0f);
int indx = (y * u_avg.cols + x) * NUM_VERTS_PER_ARROW * 3;
color_data[indx] = (theta - CV_PI) / CV_PI * 180.0f;
color_data[indx] = (theta - CV_PI_F) / CV_PI_F * 180.0f;
vertex_data[indx++] = v[0].x * xscale;
vertex_data[indx++] = v[0].y * yscale;
vertex_data[indx++] = v[0].z;
color_data[indx] = (theta - CV_PI) / CV_PI * 180.0f;
color_data[indx] = (theta - CV_PI_F) / CV_PI_F * 180.0f;
vertex_data[indx++] = v[1].x * xscale;
vertex_data[indx++] = v[1].y * yscale;
vertex_data[indx++] = v[1].z;
color_data[indx] = (theta - CV_PI) / CV_PI * 180.0f;
color_data[indx] = (theta - CV_PI_F) / CV_PI_F * 180.0f;
vertex_data[indx++] = v[2].x * xscale;
vertex_data[indx++] = v[2].y * yscale;
vertex_data[indx++] = v[2].z;
color_data[indx] = (theta - CV_PI) / CV_PI * 180.0f;
color_data[indx] = (theta - CV_PI_F) / CV_PI_F * 180.0f;
vertex_data[indx++] = v[3].x * xscale;
vertex_data[indx++] = v[3].y * yscale;
vertex_data[indx++] = v[3].z;
color_data[indx] = (theta - CV_PI) / CV_PI * 180.0f;
color_data[indx] = (theta - CV_PI_F) / CV_PI_F * 180.0f;
vertex_data[indx++] = v[4].x * xscale;
vertex_data[indx++] = v[4].y * yscale;
vertex_data[indx++] = v[4].z;
color_data[indx] = (theta - CV_PI) / CV_PI * 180.0f;
color_data[indx] = (theta - CV_PI_F) / CV_PI_F * 180.0f;
vertex_data[indx++] = v[5].x * xscale;
vertex_data[indx++] = v[5].y * yscale;
vertex_data[indx++] = v[5].z;
@@ -217,4 +217,4 @@ namespace cv { namespace gpu { namespace device
}
}}}
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

3
modules/gpu/src/cuda/optical_flow_farneback.cu
View File

@@ -42,7 +42,6 @@
#if !defined CUDA_DISABLER
#include <stdio.h>
#include "internal_shared.hpp"
#include "opencv2/gpu/device/common.hpp"
#include "opencv2/gpu/device/border_interpolate.hpp"
@@ -57,8 +56,6 @@
#define BORDER_SIZE 5
#define MAX_KSIZE_HALF 100
using namespace std;
namespace cv { namespace gpu { namespace device { namespace optflow_farneback
{
__constant__ float c_g[8];

4
modules/gpu/src/cuda/pyrlk.cu
View File

@@ -267,7 +267,7 @@ namespace cv { namespace gpu { namespace device
}
__device__ __forceinline__ float4 abs_(const float4& a)
{
return fabs(a);
return abs(a);
}
template <int cn, int PATCH_X, int PATCH_Y, bool calcErr>
@@ -681,4 +681,4 @@ namespace cv { namespace gpu { namespace device
}
}}}
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

2
modules/gpu/src/cuda/split_merge.cu
View File

@@ -508,4 +508,4 @@ namespace cv { namespace gpu { namespace device
}}} // namespace cv { namespace gpu { namespace device
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

2
modules/gpu/src/cuda/stereobp.cu
View File

@@ -454,7 +454,7 @@ namespace cv { namespace gpu { namespace device
grid.x = divUp(cols, threads.x << 1);
grid.y = divUp(rows, threads.y);
int elem_step = u.step/sizeof(T);
int elem_step = (int)(u.step / sizeof(T));
for(int t = 0; t < iters; ++t)
{

4
modules/gpu/src/cuda/surf.cu
View File

@@ -638,7 +638,7 @@ namespace cv { namespace gpu { namespace device
kp_dir *= 180.0f / CV_PI_F;
kp_dir = 360.0f - kp_dir;
if (abs(kp_dir - 360.f) < FLT_EPSILON)
if (::fabsf(kp_dir - 360.f) < FLT_EPSILON)
kp_dir = 0.f;
featureDir[blockIdx.x] = kp_dir;
@@ -1003,4 +1003,4 @@ namespace cv { namespace gpu { namespace device
}}} // namespace cv { namespace gpu { namespace device
#endif /* CUDA_DISABLER */
#endif /* CUDA_DISABLER */

2
modules/gpu/src/cuda/texture_binder.hpp
View File

@@ -85,7 +85,7 @@ namespace cv
namespace device
{
using pcl::gpu::TextureBinder;
using cv::gpu::TextureBinder;
}
}