github.com/grailbio/base@v0.0.11/compress/libdeflate/arm/matchfinder_impl.h (about)

     1  #ifndef GO_SRC_GITHUB_COM_GRAILBIO_BASE_COMPRESS_LIBDEFLATE_ARM_MATCHFINDER_IMPL_H_  // NOLINT(whitespace/line_length)
     2  #define GO_SRC_GITHUB_COM_GRAILBIO_BASE_COMPRESS_LIBDEFLATE_ARM_MATCHFINDER_IMPL_H_  // NOLINT(whitespace/line_length)
     3  /*
     4   * arm/matchfinder_impl.h - ARM implementations of matchfinder functions
     5   *
     6   * Copyright 2016 Eric Biggers
     7   *
     8   * Permission is hereby granted, free of charge, to any person
     9   * obtaining a copy of this software and associated documentation
    10   * files (the "Software"), to deal in the Software without
    11   * restriction, including without limitation the rights to use,
    12   * copy, modify, merge, publish, distribute, sublicense, and/or sell
    13   * copies of the Software, and to permit persons to whom the
    14   * Software is furnished to do so, subject to the following
    15   * conditions:
    16   *
    17   * The above copyright notice and this permission notice shall be
    18   * included in all copies or substantial portions of the Software.
    19   *
    20   * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
    21   * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES
    22   * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
    23   * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT
    24   * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY,
    25   * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
    26   * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
    27   * OTHER DEALINGS IN THE SOFTWARE.
    28   */
    29  
    30  #ifdef __ARM_NEON
    31  #  if MATCHFINDER_ALIGNMENT < 16
    32  #    undef MATCHFINDER_ALIGNMENT
    33  #    define MATCHFINDER_ALIGNMENT 16
    34  #  endif
    35  #  include <arm_neon.h>
    36  static forceinline bool
    37  matchfinder_init_neon(mf_pos_t *data, size_t size)
    38  {
    39  	int16x8_t v, *p;
    40  	size_t n;
    41  
    42  	if (size % (sizeof(int16x8_t) * 4) != 0)
    43  		return false;
    44  
    45  	STATIC_ASSERT(sizeof(mf_pos_t) == 2);
    46  	v = (int16x8_t) {
    47  		MATCHFINDER_INITVAL, MATCHFINDER_INITVAL, MATCHFINDER_INITVAL,
    48  		MATCHFINDER_INITVAL, MATCHFINDER_INITVAL, MATCHFINDER_INITVAL,
    49  		MATCHFINDER_INITVAL, MATCHFINDER_INITVAL,
    50  	};
    51  	p = (int16x8_t *)data;
    52  	n = size / (sizeof(int16x8_t) * 4);
    53  	do {
    54  		p[0] = v;
    55  		p[1] = v;
    56  		p[2] = v;
    57  		p[3] = v;
    58  		p += 4;
    59  	} while (--n);
    60  	return true;
    61  }
    62  #undef arch_matchfinder_init
    63  #define arch_matchfinder_init matchfinder_init_neon
    64  
    65  static forceinline bool
    66  matchfinder_rebase_neon(mf_pos_t *data, size_t size)
    67  {
    68  	int16x8_t v, *p;
    69  	size_t n;
    70  
    71  	if (size % (sizeof(int16x8_t) * 4) != 0)
    72  		return false;
    73  
    74  	STATIC_ASSERT(sizeof(mf_pos_t) == 2);
    75  	v = (int16x8_t) {
    76  		(u16)-MATCHFINDER_WINDOW_SIZE, (u16)-MATCHFINDER_WINDOW_SIZE,
    77  		(u16)-MATCHFINDER_WINDOW_SIZE, (u16)-MATCHFINDER_WINDOW_SIZE,
    78  		(u16)-MATCHFINDER_WINDOW_SIZE, (u16)-MATCHFINDER_WINDOW_SIZE,
    79  		(u16)-MATCHFINDER_WINDOW_SIZE, (u16)-MATCHFINDER_WINDOW_SIZE,
    80  	};
    81  	p = (int16x8_t *)data;
    82  	n = size / (sizeof(int16x8_t) * 4);
    83  	do {
    84  		p[0] = vqaddq_s16(p[0], v);
    85  		p[1] = vqaddq_s16(p[1], v);
    86  		p[2] = vqaddq_s16(p[2], v);
    87  		p[3] = vqaddq_s16(p[3], v);
    88  		p += 4;
    89  	} while (--n);
    90  	return true;
    91  }
    92  #undef arch_matchfinder_rebase
    93  #define arch_matchfinder_rebase matchfinder_rebase_neon
    94  
    95  #endif /* __ARM_NEON */
    96  
    97  #endif  // GO_SRC_GITHUB_COM_GRAILBIO_BASE_COMPRESS_LIBDEFLATE_ARM_MATCHFINDER_IMPL_H_  NOLINT(whitespace/line_length)