singe-npp 0.1.0-alpha.8

Safe Rust wrappers for NVIDIA Performance Primitives library (NPP).
Documentation 
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
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239

use super::*;

#[test]
fn fluent_image_pipeline_runs_scalar_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_host = [1u8, 2, 3, 4];
    let source_memory = DeviceMemory::from_slice(&source_host)?;
    let source = ImageView::<_, C1>::from_memory(&source_memory, Size::new(2, 2))?;
    let mut workspace = Workspace::create();

    let sum = ImagePipeline::from_view(&stream_context, &mut workspace, source).sum()?;
    let mean = ImagePipeline::from_view(&stream_context, &mut workspace, source).mean()?;
    let norm_inf = ImagePipeline::from_view(&stream_context, &mut workspace, source).norm_inf()?;
    let norm_l1 = ImagePipeline::from_view(&stream_context, &mut workspace, source).norm_l1()?;
    let norm_l2 = ImagePipeline::from_view(&stream_context, &mut workspace, source).norm_l2()?;

    stream.synchronize()?;

    assert_eq!(sum.len(), 1);
    assert_eq!(mean.len(), 1);
    assert_eq!(norm_inf.len(), 1);
    assert_eq!(norm_l1.len(), 1);
    assert_eq!(norm_l2.len(), 1);
    assert_eq!(sum.values().copy_to_host_vec()?, [10.0]);
    assert_eq!(mean.values().copy_to_host_vec()?, [2.5]);
    assert_eq!(norm_inf.values().copy_to_host_vec()?, [4.0]);
    assert_eq!(norm_l1.values().copy_to_host_vec()?, [10.0]);
    assert!((norm_l2.values().copy_to_host_vec()?[0] - 30.0f64.sqrt()).abs() < 1e-9);

    Ok(())
}

#[test]
fn fluent_image_pipeline_runs_per_channel_scalar_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_c3_host = [1u8, 2, 3, 4, 5, 6];
    let source_c4_host = [1u8, 2, 3, 4, 5, 6, 7, 8];
    let source_c3_memory = DeviceMemory::from_slice(&source_c3_host)?;
    let source_c4_memory = DeviceMemory::from_slice(&source_c4_host)?;
    let source_c3 = ImageView::<_, C3>::from_memory(&source_c3_memory, Size::new(2, 1))?;
    let source_c4 = ImageView::<_, C4>::from_memory(&source_c4_memory, Size::new(2, 1))?;
    let mut workspace = Workspace::create();

    let c3_sum = ImagePipeline::from_view(&stream_context, &mut workspace, source_c3).sum()?;
    let c3_mean = ImagePipeline::from_view(&stream_context, &mut workspace, source_c3).mean()?;
    let c3_norm_inf =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c3).norm_inf()?;
    let c4_sum = ImagePipeline::from_view(&stream_context, &mut workspace, source_c4).sum()?;
    let c4_norm_l1 =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c4).norm_l1()?;
    let c4_norm_l2 =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c4).norm_l2()?;

    stream.synchronize()?;

    assert_eq!(c3_sum.len(), 3);
    assert_eq!(c4_sum.len(), 4);
    assert_eq!(c3_sum.values().copy_to_host_vec()?, [5.0, 7.0, 9.0]);
    assert_eq!(c3_mean.values().copy_to_host_vec()?, [2.5, 3.5, 4.5]);
    assert_eq!(c3_norm_inf.values().copy_to_host_vec()?, [4.0, 5.0, 6.0]);
    assert_eq!(c4_sum.values().copy_to_host_vec()?, [6.0, 8.0, 10.0, 12.0]);
    assert_eq!(
        c4_norm_l1.values().copy_to_host_vec()?,
        [6.0, 8.0, 10.0, 12.0]
    );

    let c4_norm_l2_values = c4_norm_l2.values().copy_to_host_vec()?;
    assert!((c4_norm_l2_values[0] - 26.0f64.sqrt()).abs() < 1e-9);
    assert!((c4_norm_l2_values[1] - 40.0f64.sqrt()).abs() < 1e-9);
    assert!((c4_norm_l2_values[2] - 58.0f64.sqrt()).abs() < 1e-9);
    assert!((c4_norm_l2_values[3] - 80.0f64.sqrt()).abs() < 1e-9);

    Ok(())
}

#[test]
fn fluent_image_pipeline_runs_extremum_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_c1_host = [3u8, 1, 4, 2];
    let source_c3_host = [9u8, 2, 7, 4, 5, 6];
    let source_c4_host = [8u8, 2, 7, 4, 1, 6, 3, 5];
    let source_c1_memory = DeviceMemory::from_slice(&source_c1_host)?;
    let source_c3_memory = DeviceMemory::from_slice(&source_c3_host)?;
    let source_c4_memory = DeviceMemory::from_slice(&source_c4_host)?;
    let source_c1 = ImageView::<_, C1>::from_memory(&source_c1_memory, Size::new(2, 2))?;
    let source_c3 = ImageView::<_, C3>::from_memory(&source_c3_memory, Size::new(2, 1))?;
    let source_c4 = ImageView::<_, C4>::from_memory(&source_c4_memory, Size::new(2, 1))?;
    let mut workspace = Workspace::create();

    let c1_min = ImagePipeline::from_view(&stream_context, &mut workspace, source_c1).min()?;
    let c1_max = ImagePipeline::from_view(&stream_context, &mut workspace, source_c1).max()?;
    let c1_min_max =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c1).min_max()?;
    let c3_min_max =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c3).min_max()?;
    let c4_min = ImagePipeline::from_view(&stream_context, &mut workspace, source_c4).min()?;
    let c4_max = ImagePipeline::from_view(&stream_context, &mut workspace, source_c4).max()?;

    stream.synchronize()?;

    assert_eq!(c1_min.values().copy_to_host_vec()?, [1]);
    assert_eq!(c1_max.values().copy_to_host_vec()?, [4]);
    assert_eq!(c1_min_max.min.values().copy_to_host_vec()?, [1]);
    assert_eq!(c1_min_max.max.values().copy_to_host_vec()?, [4]);
    assert_eq!(c3_min_max.min.values().copy_to_host_vec()?, [4, 2, 6]);
    assert_eq!(c3_min_max.max.values().copy_to_host_vec()?, [9, 5, 7]);
    assert_eq!(c4_min.values().copy_to_host_vec()?, [1, 2, 3, 4]);
    assert_eq!(c4_max.values().copy_to_host_vec()?, [8, 6, 7, 5]);

    Ok(())
}

#[test]
fn fluent_image_pipeline_runs_every_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_host = [1u8, 9, 3, 7];
    let other_host = [2u8, 4, 6, 8];
    let source_memory = DeviceMemory::from_slice(&source_host)?;
    let other_memory = DeviceMemory::from_slice(&other_host)?;
    let source = ImageView::<_, C1>::from_memory(&source_memory, Size::new(2, 2))?;
    let other = ImageView::<_, C1>::from_memory(&other_memory, Size::new(2, 2))?;
    let mut workspace = Workspace::create();

    let max_sum = ImagePipeline::from_view(&stream_context, &mut workspace, source)
        .max_every(&other)?
        .sum()?;
    let min_sum = ImagePipeline::from_view(&stream_context, &mut workspace, source)
        .min_every(&other)?
        .sum()?;

    stream.synchronize()?;

    assert_eq!(max_sum.values().copy_to_host_vec()?, [25.0]);
    assert_eq!(min_sum.values().copy_to_host_vec()?, [15.0]);

    Ok(())
}

#[test]
fn fluent_image_pipeline_runs_indexed_extremum_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_c1_host = [3u8, 1, 4, 2];
    let source_c3_host = [9u8, 2, 7, 4, 5, 6];
    let source_c1_memory = DeviceMemory::from_slice(&source_c1_host)?;
    let source_c3_memory = DeviceMemory::from_slice(&source_c3_host)?;
    let source_c1 = ImageView::<_, C1>::from_memory(&source_c1_memory, Size::new(2, 2))?;
    let source_c3 = ImageView::<_, C3>::from_memory(&source_c3_memory, Size::new(2, 1))?;
    let mut workspace = Workspace::create();

    let c1_min =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c1).min_index()?;
    let c1_max =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c1).max_index()?;
    let c3_min =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c3).min_index()?;
    let c3_max =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c3).max_index()?;

    stream.synchronize()?;

    assert_eq!(c1_min.value.values().copy_to_host_vec()?, [1]);
    assert_eq!(c1_min.index_x.values().copy_to_host_vec()?, [1]);
    assert_eq!(c1_min.index_y.values().copy_to_host_vec()?, [0]);
    assert_eq!(c1_max.value.values().copy_to_host_vec()?, [4]);
    assert_eq!(c1_max.index_x.values().copy_to_host_vec()?, [0]);
    assert_eq!(c1_max.index_y.values().copy_to_host_vec()?, [1]);
    assert_eq!(c3_min.value.values().copy_to_host_vec()?, [4, 2, 6]);
    assert_eq!(c3_min.index_x.values().copy_to_host_vec()?, [1, 0, 1]);
    assert_eq!(c3_min.index_y.values().copy_to_host_vec()?, [0, 0, 0]);
    assert_eq!(c3_max.value.values().copy_to_host_vec()?, [9, 5, 7]);
    assert_eq!(c3_max.index_x.values().copy_to_host_vec()?, [0, 1, 0]);
    assert_eq!(c3_max.index_y.values().copy_to_host_vec()?, [0, 0, 0]);

    Ok(())
}

#[test]
fn fluent_image_pipeline_runs_paired_indexed_extremum_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_c1_host = [3u8, 1, 4, 2];
    let source_c3_host = [9u8, 2, 7, 4, 5, 6];
    let source_c1_memory = DeviceMemory::from_slice(&source_c1_host)?;
    let source_c3_memory = DeviceMemory::from_slice(&source_c3_host)?;
    let source_c1 = ImageView::<_, C1>::from_memory(&source_c1_memory, Size::new(2, 2))?;
    let source_c3 = ImageView::<_, C3>::from_memory(&source_c3_memory, Size::new(2, 1))?;
    let mut workspace = Workspace::create();

    let c1 =
        ImagePipeline::from_view(&stream_context, &mut workspace, source_c1).min_max_index()?;
    let c3_channel_2 = ImagePipeline::from_view(&stream_context, &mut workspace, source_c3)
        .min_max_index_channel(2)?;

    stream.synchronize()?;

    assert_eq!(c1.min.value.values().copy_to_host_vec()?, [1]);
    assert_eq!(c1.min.index_x.values().copy_to_host_vec()?, [1]);
    assert_eq!(c1.min.index_y.values().copy_to_host_vec()?, [0]);
    assert_eq!(c1.max.value.values().copy_to_host_vec()?, [4]);
    assert_eq!(c1.max.index_x.values().copy_to_host_vec()?, [0]);
    assert_eq!(c1.max.index_y.values().copy_to_host_vec()?, [1]);
    assert_eq!(c3_channel_2.min.value.values().copy_to_host_vec()?, [6]);
    assert_eq!(c3_channel_2.min.index_x.values().copy_to_host_vec()?, [1]);
    assert_eq!(c3_channel_2.min.index_y.values().copy_to_host_vec()?, [0]);
    assert_eq!(c3_channel_2.max.value.values().copy_to_host_vec()?, [7]);
    assert_eq!(c3_channel_2.max.index_x.values().copy_to_host_vec()?, [0]);
    assert_eq!(c3_channel_2.max.index_y.values().copy_to_host_vec()?, [0]);

    Ok(())
}

#[test]
fn fluent_image_pipeline_runs_count_in_range_statistics() -> Result<()> {
    let (stream, stream_context) = create_stream_context()?;
    let source_c1_host = [1.0f32, 2.0, 3.0, 4.0];
    let source_c3_host = [1u8, 2, 3, 4, 5, 6];
    let source_ac4_host = [1u8, 2, 3, 99, 4, 5, 6, 99];
    let source_c1_memory = DeviceMemory::from_slice(&source_c1_host)?;
    let source_c3_memory = DeviceMemory::from_slice(&source_c3_host)?;
    let source_ac4_memory = DeviceMemory::from_slice(&source_ac4_host)?;
    let source_c1 = ImageView::<_, C1>::from_memory(&source_c1_memory, Size::new(2, 2))?;
    let source_c3 = ImageView::<_, C3>::from_memory(&source_c3_memory, Size::new(2, 1))?;
    let source_ac4 = ImageView::<_, AC4>::from_memory(&source_ac4_memory, Size::new(2, 1))?;
    let mut workspace = Workspace::create();

    let c1_count = ImagePipeline::from_view(&stream_context, &mut workspace, source_c1)
        .count_in_range(1.5, 3.5)?;
    let c3_count = ImagePipeline::from_view(&stream_context, &mut workspace, source_c3)
        .count_in_range([0, 0, 0], [10, 10, 10])?;
    let ac4_count = ImagePipeline::from_view(&stream_context, &mut workspace, source_ac4)
        .count_in_range([0, 0, 0], [10, 10, 10])?;

    stream.synchronize()?;

    assert_eq!(c1_count.values().copy_to_host_vec()?, [2]);
    assert_eq!(c3_count.values().copy_to_host_vec()?, [2, 2, 2]);
    assert_eq!(ac4_count.values().copy_to_host_vec()?, [2, 2, 2]);

    Ok(())
}