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
194
195
196
197
198
199
200
use crate::{
    core_arch::{simd::*, simd_llvm::*, x86::*},
    mem::{self, transmute},
};

#[cfg(test)]
use stdarch_test::assert_instr;

/// Computes the absolute values of packed 32-bit integers in `a`.
///
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#avx512techs=AVX512F&expand=33,34,4990,33&text=_mm512_abs_epi32)
#[inline]
#[target_feature(enable = "avx512f")]
#[cfg_attr(test, assert_instr(vpabsd))]
pub unsafe fn _mm512_abs_epi32(a: __m512i) -> __m512i {
    let a = a.as_i32x16();
    // all-0 is a properly initialized i32x16
    let zero: i32x16 = mem::zeroed();
    let sub = simd_sub(zero, a);
    let cmp: i32x16 = simd_gt(a, zero);
    transmute(simd_select(cmp, a, sub))
}

/// Computes the absolute value of packed 32-bit integers in `a`, and store the
/// unsigned results in `dst` using writemask `k` (elements are copied from
/// `src` when the corresponding mask bit is not set).
///
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#avx512techs=AVX512F&expand=33,34,4990,33&text=_mm512_abs_epi32)
#[inline]
#[target_feature(enable = "avx512f")]
#[cfg_attr(test, assert_instr(vpabsd))]
pub unsafe fn _mm512_mask_abs_epi32(src: __m512i, k: __mmask16, a: __m512i) -> __m512i {
    let abs = _mm512_abs_epi32(a).as_i32x16();
    transmute(simd_select_bitmask(k, abs, src.as_i32x16()))
}

/// Computes the absolute value of packed 32-bit integers in `a`, and store the
/// unsigned results in `dst` using zeromask `k` (elements are zeroed out when
/// the corresponding mask bit is not set).
///
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#avx512techs=AVX512F&expand=33,34,4990,33,34,35,35&text=_mm512_maskz_abs_epi32)
#[inline]
#[target_feature(enable = "avx512f")]
#[cfg_attr(test, assert_instr(vpabsd))]
pub unsafe fn _mm512_maskz_abs_epi32(k: __mmask16, a: __m512i) -> __m512i {
    let abs = _mm512_abs_epi32(a).as_i32x16();
    let zero = _mm512_setzero_si512().as_i32x16();
    transmute(simd_select_bitmask(k, abs, zero))
}

/// Returns vector of type `__m512i` with all elements set to zero.
///
/// [Intel's documentation](https://software.intel.com/sites/landingpage/IntrinsicsGuide/#avx512techs=AVX512F&expand=33,34,4990&text=_mm512_setzero_si512)
#[inline]
#[target_feature(enable = "avx512f")]
#[cfg_attr(test, assert_instr(vxorps))]
pub unsafe fn _mm512_setzero_si512() -> __m512i {
    // All-0 is a properly initialized __m512i
    mem::zeroed()
}

/// Sets packed 32-bit integers in `dst` with the supplied values in reverse
/// order.
#[inline]
#[target_feature(enable = "avx512f")]
pub unsafe fn _mm512_setr_epi32(
    e15: i32,
    e14: i32,
    e13: i32,
    e12: i32,
    e11: i32,
    e10: i32,
    e9: i32,
    e8: i32,
    e7: i32,
    e6: i32,
    e5: i32,
    e4: i32,
    e3: i32,
    e2: i32,
    e1: i32,
    e0: i32,
) -> __m512i {
    let r = i32x16(
        e15, e14, e13, e12, e11, e10, e9, e8, e7, e6, e5, e4, e3, e2, e1, e0,
    );
    transmute(r)
}

/// Broadcast 64-bit integer `a` to all elements of `dst`.
#[inline]
#[target_feature(enable = "avx512f")]
pub unsafe fn _mm512_set1_epi64(a: i64) -> __m512i {
    transmute(i64x8::splat(a))
}

#[cfg(test)]
mod tests {
    use std;
    use stdarch_test::simd_test;

    use crate::core_arch::x86::*;

    #[simd_test(enable = "avx512f")]
    unsafe fn test_mm512_abs_epi32() {
        #[rustfmt::skip]
        let a = _mm512_setr_epi32(
            0, 1, -1, std::i32::MAX,
            std::i32::MIN, 100, -100, -32,
            0, 1, -1, std::i32::MAX,
            std::i32::MIN, 100, -100, -32,
        );
        let r = _mm512_abs_epi32(a);
        let e = _mm512_setr_epi32(
            0,
            1,
            1,
            std::i32::MAX,
            std::i32::MAX.wrapping_add(1),
            100,
            100,
            32,
            0,
            1,
            1,
            std::i32::MAX,
            std::i32::MAX.wrapping_add(1),
            100,
            100,
            32,
        );
        assert_eq_m512i(r, e);
    }

    #[simd_test(enable = "avx512f")]
    unsafe fn test_mm512_mask_abs_epi32() {
        #[rustfmt::skip]
        let a = _mm512_setr_epi32(
            0, 1, -1, std::i32::MAX,
            std::i32::MIN, 100, -100, -32,
            0, 1, -1, std::i32::MAX,
            std::i32::MIN, 100, -100, -32,
        );
        let r = _mm512_mask_abs_epi32(a, 0, a);
        assert_eq_m512i(r, a);
        let r = _mm512_mask_abs_epi32(a, 0b11111111, a);
        let e = _mm512_setr_epi32(
            0,
            1,
            1,
            std::i32::MAX,
            std::i32::MAX.wrapping_add(1),
            100,
            100,
            32,
            0,
            1,
            -1,
            std::i32::MAX,
            std::i32::MIN,
            100,
            -100,
            -32,
        );
        assert_eq_m512i(r, e);
    }

    #[simd_test(enable = "avx512f")]
    unsafe fn test_mm512_maskz_abs_epi32() {
        #[rustfmt::skip]
        let a = _mm512_setr_epi32(
            0, 1, -1, std::i32::MAX,
            std::i32::MIN, 100, -100, -32,
            0, 1, -1, std::i32::MAX,
            std::i32::MIN, 100, -100, -32,
        );
        let r = _mm512_maskz_abs_epi32(0, a);
        assert_eq_m512i(r, _mm512_setzero_si512());
        let r = _mm512_maskz_abs_epi32(0b11111111, a);
        let e = _mm512_setr_epi32(
            0,
            1,
            1,
            std::i32::MAX,
            std::i32::MAX.wrapping_add(1),
            100,
            100,
            32,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
            0,
        );
        assert_eq_m512i(r, e);
    }
}