numr 0.5.2

High-performance numerical computing with multi-backend GPU acceleration (CPU/CUDA/WebGPU)
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

//! CPU implementation of logical operations.

use crate::dtype::DType;
use crate::error::{Error, Result};
use crate::ops::LogicalOps;
use crate::runtime::cpu::{CpuClient, CpuRuntime, helpers::ensure_contiguous, kernels};
use crate::tensor::Tensor;

impl LogicalOps<CpuRuntime> for CpuClient {
    fn logical_and(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        // Validate both tensors are U8 (boolean)
        if a.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: a.dtype(),
            });
        }
        if b.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: b.dtype(),
            });
        }

        // Validate same shape
        if a.shape() != b.shape() {
            return Err(Error::ShapeMismatch {
                expected: a.shape().to_vec(),
                got: b.shape().to_vec(),
            });
        }

        let a_contig = ensure_contiguous(a);
        let b_contig = ensure_contiguous(b);
        let out = Tensor::<CpuRuntime>::empty(a.shape(), DType::U8, &self.device);

        let a_ptr = a_contig.ptr() as *const u8;
        let b_ptr = b_contig.ptr() as *const u8;
        let out_ptr = out.ptr() as *mut u8;
        let numel = a.numel();

        unsafe {
            kernels::logical_and_kernel(a_ptr, b_ptr, out_ptr, numel);
        }

        Ok(out)
    }

    fn logical_or(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        // Validate both tensors are U8 (boolean)
        if a.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: a.dtype(),
            });
        }
        if b.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: b.dtype(),
            });
        }

        // Validate same shape
        if a.shape() != b.shape() {
            return Err(Error::ShapeMismatch {
                expected: a.shape().to_vec(),
                got: b.shape().to_vec(),
            });
        }

        let a_contig = ensure_contiguous(a);
        let b_contig = ensure_contiguous(b);
        let out = Tensor::<CpuRuntime>::empty(a.shape(), DType::U8, &self.device);

        let a_ptr = a_contig.ptr() as *const u8;
        let b_ptr = b_contig.ptr() as *const u8;
        let out_ptr = out.ptr() as *mut u8;
        let numel = a.numel();

        unsafe {
            kernels::logical_or_kernel(a_ptr, b_ptr, out_ptr, numel);
        }

        Ok(out)
    }

    fn logical_xor(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        // Validate both tensors are U8 (boolean)
        if a.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: a.dtype(),
            });
        }
        if b.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: b.dtype(),
            });
        }

        // Validate same shape
        if a.shape() != b.shape() {
            return Err(Error::ShapeMismatch {
                expected: a.shape().to_vec(),
                got: b.shape().to_vec(),
            });
        }

        let a_contig = ensure_contiguous(a);
        let b_contig = ensure_contiguous(b);
        let out = Tensor::<CpuRuntime>::empty(a.shape(), DType::U8, &self.device);

        let a_ptr = a_contig.ptr() as *const u8;
        let b_ptr = b_contig.ptr() as *const u8;
        let out_ptr = out.ptr() as *mut u8;
        let numel = a.numel();

        unsafe {
            kernels::logical_xor_kernel(a_ptr, b_ptr, out_ptr, numel);
        }

        Ok(out)
    }

    fn logical_not(&self, a: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        // Validate tensor is U8 (boolean)
        if a.dtype() != DType::U8 {
            return Err(Error::DTypeMismatch {
                lhs: DType::U8,
                rhs: a.dtype(),
            });
        }

        let a_contig = ensure_contiguous(a);
        let out = Tensor::<CpuRuntime>::empty(a.shape(), DType::U8, &self.device);

        let a_ptr = a_contig.ptr() as *const u8;
        let out_ptr = out.ptr() as *mut u8;
        let numel = a.numel();

        unsafe {
            kernels::logical_not_kernel(a_ptr, out_ptr, numel);
        }

        Ok(out)
    }
}