cfavml 0.2.0

CF's Accelerated Vector Math Library providing SIMD optimzied routines for vector operations
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

//! Safe but somewhat low-level variants of the aggregation operations in CFAVML.
//!
//! In general, I would recommend using the higher level generic functions api which provides
//! some syntax sugar over these traits.

use crate::danger::export_agg_ops;

/// Various aggregation operations on a single vector.
pub trait AggOps: Sized {
    /// Performs a horizontal sum of all elements in `a` returning the result.
    ///
    /// ### Pseudocode
    ///
    /// ```ignore
    /// result = 0
    ///
    /// for i in range(dims):
    ///     result += a[i]
    ///
    /// return result
    /// ```
    ///
    /// ### Panics
    ///
    /// This function will panic if vectors `a` does not match size `dims`.
    fn sum(dims: usize, a: &[Self]) -> Self;
}

macro_rules! agg_ops {
    ($t:ty) => {
        impl AggOps for $t {
            fn sum(dims: usize, a: &[Self]) -> Self {
                assert_eq!(a.len(), dims, "Input vector `a` does not match size `dims`");

                unsafe {
                    crate::dispatch!(
                        avx512 = export_agg_ops::generic_avx512_sum,
                        avx2 = export_agg_ops::generic_avx2_sum,
                        neon = export_agg_ops::generic_neon_sum,
                        fallback = export_agg_ops::generic_fallback_sum,
                        args = (dims, a)
                    )
                }
            }
        }
    };
}

agg_ops!(f32);
agg_ops!(f64);
agg_ops!(i8);
agg_ops!(i16);
agg_ops!(i32);
agg_ops!(i64);
agg_ops!(u8);
agg_ops!(u16);
agg_ops!(u32);
agg_ops!(u64);