numrs2 0.3.3

A Rust implementation inspired by NumPy for numerical computing (NumRS2)
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
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273

//! Tests for GPU reduction operations
#![allow(clippy::result_large_err)]

#[cfg(feature = "gpu")]
mod gpu_reduction_tests {
    use numrs2::array::Array;
    use numrs2::error::Result;
    use numrs2::gpu::{self, GpuArray, GpuContext};

    fn create_gpu_context() -> Result<GpuContext> {
        // Create a simple runtime for the async GPU context creation
        let runtime = tokio::runtime::Builder::new_current_thread()
            .enable_all()
            .build()
            .map_err(|e| numrs2::error::NumRs2Error::RuntimeError(e.to_string()))?;

        runtime.block_on(GpuContext::new())
    }

    #[test]
    #[ignore = "GPU tests require proper GPU support"]
    fn test_sum_f32() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data
        let data = vec![1.0f32, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute sum on GPU
        let gpu_sum = gpu::sum_f32(&gpu_array)?;

        // Verify result
        let expected_sum: f32 = data.iter().sum();
        assert!(
            (gpu_sum - expected_sum).abs() < 1e-6,
            "Sum mismatch: expected {}, got {}",
            expected_sum,
            gpu_sum
        );

        Ok(())
    }

    #[test]
    #[ignore = "F64 GPU operations not supported on all hardware"]
    fn test_sum_f64() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data
        let data = vec![1.0f64, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute sum on GPU
        let gpu_sum = gpu::sum_f64(&gpu_array)?;

        // Verify result
        let expected_sum: f64 = data.iter().sum();
        assert!(
            (gpu_sum - expected_sum).abs() < 1e-12,
            "Sum mismatch: expected {}, got {}",
            expected_sum,
            gpu_sum
        );

        Ok(())
    }

    #[test]
    #[ignore = "GPU tests require proper GPU support"]
    fn test_mean_f32() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data
        let data = vec![1.0f32, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute mean on GPU
        let gpu_mean = gpu::mean_f32(&gpu_array)?;

        // Verify result
        let expected_mean = data.iter().sum::<f32>() / data.len() as f32;
        assert!(
            (gpu_mean - expected_mean).abs() < 1e-6,
            "Mean mismatch: expected {}, got {}",
            expected_mean,
            gpu_mean
        );

        Ok(())
    }

    #[test]
    #[ignore = "F64 GPU operations not supported on all hardware"]
    fn test_mean_f64() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data
        let data = vec![1.0f64, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute mean on GPU
        let gpu_mean = gpu::mean_f64(&gpu_array)?;

        // Verify result
        let expected_mean = data.iter().sum::<f64>() / data.len() as f64;
        assert!(
            (gpu_mean - expected_mean).abs() < 1e-12,
            "Mean mismatch: expected {}, got {}",
            expected_mean,
            gpu_mean
        );

        Ok(())
    }

    #[test]
    #[ignore = "GPU tests require proper GPU support"]
    fn test_max_f32() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data with various values including negative
        let data = vec![3.5f32, -2.0, 8.1, 1.0, -5.5, 7.9, 0.0, 4.2];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute max on GPU
        let gpu_max = gpu::max_f32(&gpu_array)?;

        // Verify result
        let expected_max = data.iter().cloned().fold(f32::NEG_INFINITY, f32::max);
        assert!(
            (gpu_max - expected_max).abs() < 1e-6,
            "Max mismatch: expected {}, got {}",
            expected_max,
            gpu_max
        );

        Ok(())
    }

    #[test]
    #[ignore = "F64 GPU operations not supported on all hardware"]
    fn test_max_f64() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data with various values including negative
        let data = vec![3.5f64, -2.0, 8.1, 1.0, -5.5, 7.9, 0.0, 4.2];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute max on GPU
        let gpu_max = gpu::max_f64(&gpu_array)?;

        // Verify result
        let expected_max = data.iter().cloned().fold(f64::NEG_INFINITY, f64::max);
        assert!(
            (gpu_max - expected_max).abs() < 1e-12,
            "Max mismatch: expected {}, got {}",
            expected_max,
            gpu_max
        );

        Ok(())
    }

    #[test]
    #[ignore = "GPU tests require proper GPU support"]
    fn test_min_f32() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data with various values including negative
        let data = vec![3.5f32, -2.0, 8.1, 1.0, -5.5, 7.9, 0.0, 4.2];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute min on GPU
        let gpu_min = gpu::min_f32(&gpu_array)?;

        // Verify result
        let expected_min = data.iter().cloned().fold(f32::INFINITY, f32::min);
        assert!(
            (gpu_min - expected_min).abs() < 1e-6,
            "Min mismatch: expected {}, got {}",
            expected_min,
            gpu_min
        );

        Ok(())
    }

    #[test]
    #[ignore = "F64 GPU operations not supported on all hardware"]
    fn test_min_f64() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test data with various values including negative
        let data = vec![3.5f64, -2.0, 8.1, 1.0, -5.5, 7.9, 0.0, 4.2];
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Compute min on GPU
        let gpu_min = gpu::min_f64(&gpu_array)?;

        // Verify result
        let expected_min = data.iter().cloned().fold(f64::INFINITY, f64::min);
        assert!(
            (gpu_min - expected_min).abs() < 1e-12,
            "Min mismatch: expected {}, got {}",
            expected_min,
            gpu_min
        );

        Ok(())
    }

    #[test]
    #[ignore = "GPU tests require proper GPU support"]
    fn test_large_array_reductions() -> Result<()> {
        let _context = create_gpu_context()?;

        // Test with a large array to ensure workgroup handling is correct
        let size = 100_000;
        let data: Vec<f32> = (0..size).map(|i| i as f32).collect();
        let cpu_array = Array::from_vec(data.clone());
        let gpu_array = GpuArray::from_array(&cpu_array)?;

        // Test sum
        let gpu_sum = gpu::sum_f32(&gpu_array)?;
        let expected_sum = (size * (size - 1)) as f32 / 2.0;
        assert!(
            (gpu_sum - expected_sum).abs() / expected_sum < 1e-5,
            "Large array sum mismatch: expected {}, got {}",
            expected_sum,
            gpu_sum
        );

        // Test mean
        let gpu_mean = gpu::mean_f32(&gpu_array)?;
        let expected_mean = (size - 1) as f32 / 2.0;
        assert!(
            (gpu_mean - expected_mean).abs() < 1e-3,
            "Large array mean mismatch: expected {}, got {}",
            expected_mean,
            gpu_mean
        );

        // Test max
        let gpu_max = gpu::max_f32(&gpu_array)?;
        let expected_max = (size - 1) as f32;
        assert!(
            (gpu_max - expected_max).abs() < 1e-6,
            "Large array max mismatch: expected {}, got {}",
            expected_max,
            gpu_max
        );

        // Test min
        let gpu_min = gpu::min_f32(&gpu_array)?;
        let expected_min = 0.0f32;
        assert!(
            (gpu_min - expected_min).abs() < 1e-6,
            "Large array min mismatch: expected {}, got {}",
            expected_min,
            gpu_min
        );

        Ok(())
    }
}