urng 1.0.0

Universal Random Number Generator
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

use wrapn::{wrap, wu32};

use crate::rng::Rng;

/// SplitMix32 32-bit RNG implementation.
///
/// Fast 32-bit finalizer-based PRNG commonly used to seed other generators.
/// Uses a single 32-bit state word advanced by the golden-ratio constant.
///
/// # Example
/// ```
/// use urng::{Rng, SplitMix32};
///
/// let mut rng = SplitMix32::new(1);
/// let _ = rng.nextu();
/// ```
#[repr(C, align(64))]
#[derive(Debug, Clone, Copy)]
pub struct SplitMix32 {
    state: wu32,
}

const A: u64 = 0xFF51_AFD7_ED55_8CCD;
const B: u64 = 0xC4CE_B9FE_1A85_EC53;

impl SplitMix32 {
    /// Creates a new `SplitMix32` instance with the given seed.
    pub const fn new(seed: u32) -> Self {
        Self {
            state: wrap!(seed | 1),
        }
    }

    #[inline(always)]
    pub(crate) const fn nextu_const(&mut self) -> u32 {
        self.state.0.0 = self.state.0.0.wrapping_add(0x9E3779B9);

        let mut z = self.state.0.0 as u64;
        z = (z ^ (z >> 16)).wrapping_mul(A);
        z = (z ^ (z >> 16)).wrapping_mul(B);
        (z ^ (z >> 16)) as u32
    }
}

impl Rng for SplitMix32 {
    type Word = u32;

    #[inline]
    fn nextu(&mut self) -> Self::Word {
        self.state += 0x9E3779B9;

        let mut z = self.state.cast::<u64>();
        z = (z ^ (z >> 16)) * A;
        z = (z ^ (z >> 16)) * B;
        *(z ^ (z >> 16)).cast::<u32>()
    }
}

#[cfg(feature = "simd")]
pub use simd::*;

#[cfg(feature = "simd")]
pub mod simd {
    #[cfg(target_arch = "x86_64")]
    use std::arch::x86_64::*;

    #[cfg(target_arch = "x86_64")]
    use crate::{RngV, SplitMix32};

    pub const SPLITMIX32X16: usize = 16;
    pub const SPLITMIX32X16_PAR_CHUNK: usize = 8192;
    pub const SPLITMIX32_GAMMA: u32 = 0x9E37_79B9;

    /// 16-way SIMD implementation of SplitMix32 32-bit RNG.
    /// This implementation uses AVX-512F instructions to generate 16 random numbers in parallel.
    ///
    /// # Example
    /// ```no_run
    /// use urng::{RngV, SplitMix32x16};
    /// unsafe {
    ///     let mut rng = SplitMix32x16::new(1);
    ///     let _ = rng.nextuv();
    /// }
    /// ```
    #[cfg(target_arch = "x86_64")]
    #[repr(C, align(64))]
    pub struct SplitMix32x16 {
        pub(crate) state: __m512i,
    }

    #[cfg(target_arch = "x86_64")]
    impl SplitMix32x16 {
        /// Creates a new `SplitMix32x16` instance with the given seed.
        ///
        /// # Safety
        ///
        /// The caller must ensure the CPU supports the `avx512f` target feature.
        #[target_feature(enable = "avx512f")]
        pub unsafe fn new(seed: u32) -> Self {
            let mut sm = SplitMix32::new(seed);

            let mut s = [0u32; SPLITMIX32X16];
            for i in 0..SPLITMIX32X16 {
                s[i] = sm.nextu_const();
            }

            unsafe {
                Self {
                    state: _mm512_loadu_si512(s.as_ptr() as _),
                }
            }
        }

        /// Computes the SplitMix32 output for 16 lanes at once.
        ///
        /// # Safety
        ///
        /// The caller must ensure the CPU supports the `avx512f` target feature.
        #[target_feature(enable = "avx512f")]
        pub unsafe fn compute(state: __m512i) -> __m512i {
            let c1 = _mm512_set1_epi32(0x85EB_CA6Bu32 as i32);
            let c2 = _mm512_set1_epi32(0xC2B2_AE35u32 as i32);

            let mut z = state;
            z = _mm512_xor_si512(z, _mm512_srli_epi32(z, 16));
            z = _mm512_add_epi32(z, c1);
            z = _mm512_xor_si512(z, _mm512_srli_epi32(z, 13));
            z = _mm512_add_epi32(z, c2);
            _mm512_xor_si512(z, _mm512_srli_epi32(z, 16))
        }
    }

    #[cfg(target_arch = "x86_64")]
    impl RngV for SplitMix32x16 {
        type Word = __m512i;

        fn nextuv(&mut self) -> Self::Word {
            unsafe {
                let v = Self::compute(self.state);
                self.state = _mm512_add_epi32(
                    self.state,
                    _mm512_set1_epi32(SPLITMIX32_GAMMA.wrapping_mul(SPLITMIX32X16 as u32) as i32),
                );
                v
            }
        }
    }
}

#[cfg(test)]
mod tests {
    use super::*;

    crate::safe_test! { SplitMix32 }
}