pmat 3.15.0

PMAT - Zero-config AI context generation and code quality toolkit (CLI, MCP, HTTP)
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

    // ============ TRUENO-RAG-2-COSINE: SIMD Cosine Similarity Tests ============
    // RED Phase: These tests define the expected behavior of SIMD-accelerated cosine similarity

    /// Test that SIMD cosine similarity matches scalar implementation for identical vectors
    #[test]
    fn test_simd_cosine_similarity_identical_vectors() {
        let v1 = vec![1.0f32, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0];
        let v2 = vec![1.0f32, 2.0, 3.0, 4.0, 5.0, 6.0, 7.0, 8.0];

        let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

        assert!(
            (scalar - simd).abs() < 0.0001,
            "SIMD should match scalar for identical vectors: scalar={}, simd={}",
            scalar,
            simd
        );
        assert!(
            (simd - 1.0).abs() < 0.0001,
            "Identical vectors should have similarity 1.0"
        );
    }

    /// Test that SIMD cosine similarity matches scalar for orthogonal vectors
    #[test]
    fn test_simd_cosine_similarity_orthogonal_vectors() {
        let v1 = vec![1.0f32, 0.0, 0.0, 0.0];
        let v2 = vec![0.0f32, 1.0, 0.0, 0.0];

        let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

        assert!(
            (scalar - simd).abs() < 0.0001,
            "SIMD should match scalar for orthogonal vectors"
        );
        assert!(
            simd.abs() < 0.0001,
            "Orthogonal vectors should have similarity 0.0"
        );
    }

    /// Test that SIMD cosine similarity handles large vectors (OpenAI embedding dimension)
    #[test]
    fn test_simd_cosine_similarity_large_vectors() {
        // OpenAI ada-002 embedding dimension is 1536
        let v1: Vec<f32> = (0..1536).map(|i| (i as f32 * 0.001).sin()).collect();
        let v2: Vec<f32> = (0..1536).map(|i| (i as f32 * 0.001).cos()).collect();

        let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

        assert!(
            (scalar - simd).abs() < 0.001,
            "SIMD should match scalar for large vectors: scalar={}, simd={}",
            scalar,
            simd
        );
    }

    /// Test that SIMD handles non-SIMD-aligned vector sizes (not multiple of 4 or 8)
    #[test]
    fn test_simd_cosine_similarity_unaligned_sizes() {
        // Test various sizes that aren't multiples of SIMD lane width
        for size in [1, 3, 5, 7, 9, 13, 17, 33, 65, 129] {
            let v1: Vec<f32> = (0..size).map(|i| i as f32 * 0.1).collect();
            let v2: Vec<f32> = (0..size).map(|i| (i as f32 * 0.1).powi(2)).collect();

            let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
            let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

            assert!(
                (scalar - simd).abs() < 0.001,
                "SIMD should match scalar for size {}: scalar={}, simd={}",
                size,
                scalar,
                simd
            );
        }
    }

    /// Test SIMD handles empty vectors
    #[test]
    fn test_simd_cosine_similarity_empty_vectors() {
        let v1: Vec<f32> = vec![];
        let v2: Vec<f32> = vec![];

        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);
        assert!(
            simd == 0.0 || simd.is_nan(),
            "Empty vectors should return 0 or NaN"
        );
    }

    /// Test SIMD handles zero vectors
    #[test]
    fn test_simd_cosine_similarity_zero_vectors() {
        let v1 = vec![0.0f32, 0.0, 0.0, 0.0];
        let v2 = vec![1.0f32, 2.0, 3.0, 4.0];

        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);
        assert!(simd == 0.0, "Zero vector should result in 0.0 similarity");
    }

    /// Test SIMD handles negative values
    #[test]
    fn test_simd_cosine_similarity_opposite_vectors() {
        let v1 = vec![1.0f32, 2.0, 3.0, 4.0];
        let v2 = vec![-1.0f32, -2.0, -3.0, -4.0];

        let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

        assert!(
            (scalar - simd).abs() < 0.0001,
            "SIMD should match scalar for opposite vectors"
        );
        assert!(
            (simd + 1.0).abs() < 0.0001,
            "Opposite vectors should have similarity -1.0"
        );
    }

    /// Property test: SIMD and scalar should always produce equivalent results
    #[test]
    fn test_simd_scalar_equivalence_property() {
        use std::f32::consts::PI;

        // Generate 100 random test cases
        for seed in 0..100 {
            let size = 64 + (seed % 200); // Sizes from 64 to 263
            let v1: Vec<f32> = (0..size)
                .map(|i| ((i as f32 + seed as f32) * 0.1).sin())
                .collect();
            let v2: Vec<f32> = (0..size)
                .map(|i| ((i as f32 + seed as f32) * PI * 0.1).cos())
                .collect();

            let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
            let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

            assert!(
                (scalar - simd).abs() < 0.001,
                "Seed {}, size {}: scalar={}, simd={}, diff={}",
                seed,
                size,
                scalar,
                simd,
                (scalar - simd).abs()
            );
        }
    }

    /// Performance baseline test (documents expected speedup)
    #[test]
    fn test_simd_cosine_similarity_performance_baseline() {
        use std::time::Instant;

        // Large vectors to measure performance
        let v1: Vec<f32> = (0..10000).map(|i| (i as f32 * 0.0001).sin()).collect();
        let v2: Vec<f32> = (0..10000).map(|i| (i as f32 * 0.0001).cos()).collect();

        // Warmup
        let _ = TursoVectorDB::cosine_similarity(&v1, &v2);
        let _ = TursoVectorDB::cosine_similarity_simd(&v1, &v2);

        // Benchmark scalar
        let scalar_start = Instant::now();
        for _ in 0..1000 {
            let _ = TursoVectorDB::cosine_similarity(&v1, &v2);
        }
        let scalar_duration = scalar_start.elapsed();

        // Benchmark SIMD
        let simd_start = Instant::now();
        for _ in 0..1000 {
            let _ = TursoVectorDB::cosine_similarity_simd(&v1, &v2);
        }
        let simd_duration = simd_start.elapsed();

        // Document the speedup (soft assertion - just log it)
        let speedup = scalar_duration.as_nanos() as f64 / simd_duration.as_nanos() as f64;
        eprintln!(
            "SIMD cosine similarity speedup: {:.2}x (scalar: {:?}, simd: {:?})",
            speedup, scalar_duration, simd_duration
        );

        // Verify correctness
        let scalar = TursoVectorDB::cosine_similarity(&v1, &v2);
        let simd = TursoVectorDB::cosine_similarity_simd(&v1, &v2);
        assert!((scalar - simd).abs() < 0.001);
    }