1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
/*
* Copyright 1993-2010 NVIDIA Corporation. All rights reserved.
*
* NVIDIA Corporation and its licensors retain all intellectual
* property and proprietary rights in and to this software and
* related documentation and any modifications thereto.
* Any use, reproduction, disclosure, or distribution of this
* software and related documentation without an express license
* agreement from NVIDIA Corporation is strictly prohibited.
*/
#if !defined CUDA_DISABLER
#include <math.h>
#include "TestResize.h"
template <class T>
TestResize<T>::TestResize(std::string testName_, NCVTestSourceProvider<T> &src_,
Ncv32u width_, Ncv32u height_, Ncv32u scaleFactor_, NcvBool bTextureCache_)
:
NCVTestProvider(testName_),
src(src_),
width(width_),
height(height_),
scaleFactor(scaleFactor_),
bTextureCache(bTextureCache_)
{
}
template <class T>
bool TestResize<T>::toString(std::ofstream &strOut)
{
strOut << "sizeof(T)=" << sizeof(T) << std::endl;
strOut << "width=" << width << std::endl;
strOut << "scaleFactor=" << scaleFactor << std::endl;
strOut << "bTextureCache=" << bTextureCache << std::endl;
return true;
}
template <class T>
bool TestResize<T>::init()
{
return true;
}
template <class T>
bool TestResize<T>::process()
{
NCVStatus ncvStat;
bool rcode = false;
Ncv32s smallWidth = this->width / this->scaleFactor;
Ncv32s smallHeight = this->height / this->scaleFactor;
if (smallWidth == 0 || smallHeight == 0)
{
return true;
}
NcvSize32u srcSize(this->width, this->height);
NCVMatrixAlloc<T> d_img(*this->allocatorGPU.get(), this->width, this->height);
ncvAssertReturn(d_img.isMemAllocated(), false);
NCVMatrixAlloc<T> h_img(*this->allocatorCPU.get(), this->width, this->height);
ncvAssertReturn(h_img.isMemAllocated(), false);
NCVMatrixAlloc<T> d_small(*this->allocatorGPU.get(), smallWidth, smallHeight);
ncvAssertReturn(d_small.isMemAllocated(), false);
NCVMatrixAlloc<T> h_small(*this->allocatorCPU.get(), smallWidth, smallHeight);
ncvAssertReturn(h_small.isMemAllocated(), false);
NCVMatrixAlloc<T> h_small_d(*this->allocatorCPU.get(), smallWidth, smallHeight);
ncvAssertReturn(h_small_d.isMemAllocated(), false);
NCV_SET_SKIP_COND(this->allocatorGPU.get()->isCounting());
NCV_SKIP_COND_BEGIN
ncvAssertReturn(this->src.fill(h_img), false);
NCV_SKIP_COND_END
ncvStat = h_img.copySolid(d_img, 0);
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
NCV_SKIP_COND_BEGIN
if (sizeof(T) == sizeof(Ncv32u))
{
ncvStat = nppiStDecimate_32u_C1R((Ncv32u *)d_img.ptr(), d_img.pitch(),
(Ncv32u *)d_small.ptr(), d_small.pitch(),
srcSize, this->scaleFactor,
this->bTextureCache);
}
else if (sizeof(T) == sizeof(Ncv64u))
{
ncvStat = nppiStDecimate_64u_C1R((Ncv64u *)d_img.ptr(), d_img.pitch(),
(Ncv64u *)d_small.ptr(), d_small.pitch(),
srcSize, this->scaleFactor,
this->bTextureCache);
}
else
{
ncvAssertPrintReturn(false, "Incorrect downsample test instance", false);
}
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
NCV_SKIP_COND_END
ncvStat = d_small.copySolid(h_small_d, 0);
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
NCV_SKIP_COND_BEGIN
if (sizeof(T) == sizeof(Ncv32u))
{
ncvStat = nppiStDecimate_32u_C1R_host((Ncv32u *)h_img.ptr(), h_img.pitch(),
(Ncv32u *)h_small.ptr(), h_small.pitch(),
srcSize, this->scaleFactor);
}
else if (sizeof(T) == sizeof(Ncv64u))
{
ncvStat = nppiStDecimate_64u_C1R_host((Ncv64u *)h_img.ptr(), h_img.pitch(),
(Ncv64u *)h_small.ptr(), h_small.pitch(),
srcSize, this->scaleFactor);
}
else
{
ncvAssertPrintReturn(false, "Incorrect downsample test instance", false);
}
ncvAssertReturn(ncvStat == NPPST_SUCCESS, false);
NCV_SKIP_COND_END
//bit-to-bit check
bool bLoopVirgin = true;
NCV_SKIP_COND_BEGIN
//const Ncv64f relEPS = 0.005;
for (Ncv32u i=0; bLoopVirgin && i < h_small.height(); i++)
{
for (Ncv32u j=0; bLoopVirgin && j < h_small.width(); j++)
{
if (h_small.ptr()[h_small.stride()*i+j] != h_small_d.ptr()[h_small_d.stride()*i+j])
{
bLoopVirgin = false;
}
}
}
NCV_SKIP_COND_END
if (bLoopVirgin)
{
rcode = true;
}
return rcode;
}
template <class T>
bool TestResize<T>::deinit()
{
return true;
}
template class TestResize<Ncv32u>;
template class TestResize<Ncv64u>;
#endif /* CUDA_DISABLER */