numr 0.5.1

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

//! CPU implementation of binary operations.

use crate::error::Result;
use crate::ops::BinaryOps;
use crate::runtime::cpu::{
    CpuClient, CpuRuntime,
    helpers::{BinaryOp, binary_op_impl, fused_add_mul_impl, fused_mul_add_impl},
};
use crate::tensor::Tensor;

/// BinaryOps implementation for CPU runtime.
impl BinaryOps<CpuRuntime> for CpuClient {
    fn add(&self, a: &Tensor<CpuRuntime>, b: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Add, a, b, "add")
    }

    fn sub(&self, a: &Tensor<CpuRuntime>, b: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Sub, a, b, "sub")
    }

    fn mul(&self, a: &Tensor<CpuRuntime>, b: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Mul, a, b, "mul")
    }

    fn div(&self, a: &Tensor<CpuRuntime>, b: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Div, a, b, "div")
    }

    fn pow(&self, a: &Tensor<CpuRuntime>, b: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Pow, a, b, "pow")
    }

    fn maximum(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Max, a, b, "maximum")
    }

    fn minimum(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Min, a, b, "minimum")
    }

    fn atan2(&self, y: &Tensor<CpuRuntime>, x: &Tensor<CpuRuntime>) -> Result<Tensor<CpuRuntime>> {
        binary_op_impl(self, BinaryOp::Atan2, y, x, "atan2")
    }

    fn fused_mul_add(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
        c: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        fused_mul_add_impl(self, a, b, c)
    }

    fn fused_add_mul(
        &self,
        a: &Tensor<CpuRuntime>,
        b: &Tensor<CpuRuntime>,
        c: &Tensor<CpuRuntime>,
    ) -> Result<Tensor<CpuRuntime>> {
        fused_add_mul_impl(self, a, b, c)
    }
}