/* adler32_avx2_p.h -- adler32 avx2 utility functions

 * Copyright (C) 2022 Adam Stylinski

 * For conditions of distribution and use, see copyright notice in zlib.h

 */

#ifndef ADLER32_AVX2_P_H_

#define ADLER32_AVX2_P_H_

#if defined(X86_AVX2) || defined(X86_AVX512VNNI)

/* 32 bit horizontal sum, adapted from Agner Fog's vector library. */

static inline uint32_t hsum256(__m256i x) {

    __m128i sum1  = _mm_add_epi32(_mm256_extracti128_si256(x, 1),

                                  _mm256_castsi256_si128(x));

    __m128i sum2  = _mm_add_epi32(sum1, _mm_unpackhi_epi64(sum1, sum1));

    __m128i sum3  = _mm_add_epi32(sum2, _mm_shuffle_epi32(sum2, 1));

    return (uint32_t)_mm_cvtsi128_si32(sum3);

}

adler32_avx2.c:hsum256

Line

Count

Source

12

796k

static inline uint32_t hsum256(__m256i x) {

13

796k

    __m128i sum1  = _mm_add_epi32(_mm256_extracti128_si256(x, 1),

14

796k

                                  _mm256_castsi256_si128(x));

15

796k

    __m128i sum2  = _mm_add_epi32(sum1, _mm_unpackhi_epi64(sum1, sum1));

16

    __m128i sum3  = _mm_add_epi32(sum2, _mm_shuffle_epi32(sum2, 1));

17

796k

    return (uint32_t)_mm_cvtsi128_si32(sum3);

18

796k

}

Unexecuted instantiation: adler32_avx512_vnni.c:hsum256

static inline uint32_t partial_hsum256(__m256i x) {

    /* We need a permutation vector to extract every other integer. The

     * rest are going to be zeros */

    const __m256i perm_vec = _mm256_setr_epi32(0, 2, 4, 6, 1, 1, 1, 1);

    __m256i non_zero = _mm256_permutevar8x32_epi32(x, perm_vec);

    __m128i non_zero_sse = _mm256_castsi256_si128(non_zero);

    __m128i sum2  = _mm_add_epi32(non_zero_sse,_mm_unpackhi_epi64(non_zero_sse, non_zero_sse));

    __m128i sum3  = _mm_add_epi32(sum2, _mm_shuffle_epi32(sum2, 1));

    return (uint32_t)_mm_cvtsi128_si32(sum3);

}

adler32_avx2.c:partial_hsum256

Line

Count

Source

20

796k

static inline uint32_t partial_hsum256(__m256i x) {

21

    /* We need a permutation vector to extract every other integer. The

22

     * rest are going to be zeros */

23

796k

    const __m256i perm_vec = _mm256_setr_epi32(0, 2, 4, 6, 1, 1, 1, 1);

24

796k

    __m256i non_zero = _mm256_permutevar8x32_epi32(x, perm_vec);

25

796k

    __m128i non_zero_sse = _mm256_castsi256_si128(non_zero);

26

796k

    __m128i sum2  = _mm_add_epi32(non_zero_sse,_mm_unpackhi_epi64(non_zero_sse, non_zero_sse));

27

    __m128i sum3  = _mm_add_epi32(sum2, _mm_shuffle_epi32(sum2, 1));

28

796k

    return (uint32_t)_mm_cvtsi128_si32(sum3);

29

796k

}

Unexecuted instantiation: adler32_avx512_vnni.c:partial_hsum256

#endif

#endif