trueno 0.16.4

High-performance SIMD compute library with GPU support for matrix operations
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

use super::super::super::super::*;
use crate::Backend;

// Basic construction tests
#[test]
fn test_from_slice() {
    let v = Vector::from_slice(&[1.0, 2.0, 3.0]);
    assert_eq!(v.as_slice(), &[1.0, 2.0, 3.0]);
    assert_eq!(v.len(), 3);
}

#[test]
fn test_from_slice_empty() {
    let v: Vector<f32> = Vector::from_slice(&[]);
    assert_eq!(v.len(), 0);
    assert!(v.is_empty());
}

#[test]
fn test_from_slice_single_element() {
    let v = Vector::from_slice(&[42.0]);
    assert_eq!(v.as_slice(), &[42.0]);
    assert_eq!(v.len(), 1);
}

#[test]
fn test_from_slice_with_backend() {
    let v = Vector::from_slice_with_backend(&[1.0, 2.0], Backend::Scalar);
    assert_eq!(v.backend(), Backend::Scalar);
}

#[test]
fn test_scalar_backend_operations() {
    // Test various operations with explicit Scalar backend to cover Backend::Scalar match arms
    let v1 = Vector::from_slice_with_backend(&[1.0, 2.0, 3.0, 4.0], Backend::Scalar);
    let v2 = Vector::from_slice_with_backend(&[4.0, 3.0, 2.0, 1.0], Backend::Scalar);

    // Test dot product (covers line 599)
    let dot = v1.dot(&v2).unwrap();
    assert_eq!(dot, 1.0 * 4.0 + 2.0 * 3.0 + 3.0 * 2.0 + 4.0 * 1.0); // = 20.0

    // Test sum (covers line 856)
    let sum = v1.sum().unwrap();
    assert_eq!(sum, 10.0);

    // Test max (covers line 661)
    let max = v1.max().unwrap();
    assert_eq!(max, 4.0);

    // Test min (covers line 709)
    let min = v1.min().unwrap();
    assert_eq!(min, 1.0);

    // Test argmax (covers line 757)
    let argmax = v1.argmax().unwrap();
    assert_eq!(argmax, 3);

    // Test argmin (covers line 805)
    let argmin = v1.argmin().unwrap();
    assert_eq!(argmin, 0);
}

#[test]
fn test_gpu_and_auto_backend_fallback() {
    // Test operations with GPU/Auto backend which fallback to scalar
    let v1 = Vector::from_slice_with_backend(&[1.0, 2.0, 3.0], Backend::GPU);
    let v2 = Vector::from_slice_with_backend(&[3.0, 2.0, 1.0], Backend::GPU);

    // These should all work (fallback to scalar)
    let dot = v1.dot(&v2).unwrap();
    assert_eq!(dot, 10.0);

    let sum = v1.sum().unwrap();
    assert_eq!(sum, 6.0);
}

#[test]
#[cfg(target_arch = "x86_64")]
fn test_avx512_backend_vector_ops() {
    // Test operations with explicit AVX-512 backend to cover Backend::AVX512 match arms
    if !is_x86_feature_detected!("avx512f") {
        return;
    }

    // Use large vectors to exercise SIMD paths
    let data1: Vec<f32> = (0..1024).map(|i| i as f32).collect();
    let data2: Vec<f32> = (0..1024).map(|i| (1024 - i) as f32).collect();

    let v1 = Vector::from_slice_with_backend(&data1, Backend::AVX512);
    let v2 = Vector::from_slice_with_backend(&data2, Backend::AVX512);

    // Test dot product (covers line 605)
    let dot = v1.dot(&v2).unwrap();
    let expected_dot: f32 = data1.iter().zip(data2.iter()).map(|(a, b)| a * b).sum();
    let dot_rel_err = (dot - expected_dot).abs() / expected_dot.abs().max(1.0);
    assert!(dot_rel_err < 0.01, "dot mismatch: {} vs {}", dot, expected_dot);

    // Test sum
    let sum = v1.sum().unwrap();
    let expected_sum: f32 = data1.iter().sum();
    let sum_rel_err = (sum - expected_sum).abs() / expected_sum.abs().max(1.0);
    assert!(sum_rel_err < 0.01, "sum mismatch: {} vs {}", sum, expected_sum);

    // Test max
    let max = v1.max().unwrap();
    assert_eq!(max, 1023.0);

    // Test min
    let min = v1.min().unwrap();
    assert_eq!(min, 0.0);

    // Test argmax
    let argmax = v1.argmax().unwrap();
    assert_eq!(argmax, 1023);

    // Test argmin
    let argmin = v1.argmin().unwrap();
    assert_eq!(argmin, 0);
}

#[test]
fn test_auto_backend_resolution() {
    let v = Vector::from_slice_with_backend(&[1.0], Backend::Auto);
    // Auto should be resolved to best available backend
    let expected_backend = crate::select_best_available_backend();
    assert_eq!(v.backend(), expected_backend);

    // Verify it's not still Backend::Auto after resolution
    assert_ne!(v.backend(), Backend::Auto);

    // On x86_64, should be a SIMD backend (not Scalar)
    #[cfg(target_arch = "x86_64")]
    {
        assert_ne!(v.backend(), Backend::Scalar);
        assert!(matches!(
            v.backend(),
            Backend::SSE2 | Backend::AVX | Backend::AVX2 | Backend::AVX512
        ));
    }
}

#[test]
fn test_with_alignment_valid() {
    let v = Vector::with_alignment(100, Backend::SSE2, 16).unwrap();
    assert_eq!(v.len(), 100);
    assert_eq!(v.backend(), Backend::SSE2);
}

#[test]
fn test_with_alignment_power_of_two() {
    // Test various power-of-2 alignments
    assert!(Vector::with_alignment(10, Backend::Scalar, 1).is_ok());
    assert!(Vector::with_alignment(10, Backend::Scalar, 2).is_ok());
    assert!(Vector::with_alignment(10, Backend::Scalar, 4).is_ok());
    assert!(Vector::with_alignment(10, Backend::Scalar, 8).is_ok());
    assert!(Vector::with_alignment(10, Backend::Scalar, 16).is_ok());
    assert!(Vector::with_alignment(10, Backend::Scalar, 32).is_ok());
    assert!(Vector::with_alignment(10, Backend::Scalar, 64).is_ok());
}

#[test]
fn test_with_alignment_invalid_zero() {
    let result = Vector::with_alignment(100, Backend::Scalar, 0);
    assert!(result.is_err());
    match result {
        Err(TruenoError::InvalidInput(msg)) => {
            assert!(msg.contains("power of 2"));
            assert!(msg.contains('0'));
        }
        _ => panic!("Expected InvalidInput error for zero alignment"),
    }
}

#[test]
fn test_with_alignment_invalid_not_power_of_two() {
    // Test various non-power-of-2 values
    for alignment in &[3, 5, 6, 7, 9, 10, 12, 15, 17, 20, 24, 31, 33] {
        let result = Vector::with_alignment(100, Backend::Scalar, *alignment);
        assert!(result.is_err(), "Alignment {} should be invalid", alignment);
        match result {
            Err(TruenoError::InvalidInput(msg)) => {
                assert!(msg.contains("power of 2"), "Error message should mention power of 2");
                assert!(
                    msg.contains(&alignment.to_string()),
                    "Error message should mention the invalid alignment"
                );
            }
            _ => panic!("Expected InvalidInput error for non-power-of-2 alignment {}", alignment),
        }
    }
}

#[test]
fn test_with_alignment_auto_backend_resolution() {
    let v = Vector::with_alignment(100, Backend::Auto, 16).unwrap();
    // Backend::Auto should be resolved to best available backend
    assert_ne!(v.backend(), Backend::Auto);
}