Just a exercise to count lines on text files using SSE

countlines.c

// gcc countlines.c -O2 -o countlines

#include <stdio.h>
#include <stdlib.h>
#include <stdint.h>
#include <unistd.h>
#include <fcntl.h>
#include <emmintrin.h>

#define BLOCK_SIZE 4096 * 2

__always_inline
int array_sum16(int8_t *array)
{
  return array[0]
    + array[1]
    + array[2]
    + array[3]
    + array[4]
    + array[5]
    + array[6]
    + array[7]
    + array[8]
    + array[9]
    + array[10]
    + array[11]
    + array[12]
    + array[13]
    + array[14]
    + array[15];
}

unsigned long long int count_block_lines(__m128i *block, ssize_t size)
{
  static int8_t array_result[sizeof (__m128i)];
  static int8_t array_lines[] = {'\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n', '\n'};

  __m128i result;
  __m128i vector_lines = _mm_load_si128((__m128i *) array_lines);

  unsigned int block_index;
  unsigned long long int nlines = 0;
  int block_count = size / sizeof (__m128i);

  for (block_index = 0; block_index < block_count; block_index++)
  {
    result = _mm_cmpeq_epi8(vector_lines, block[block_index]);
    _mm_store_si128((__m128i *) array_result, result);

    nlines += -array_sum16(array_result);
  }

  int remaining_content_size = size % sizeof (__m128i);
  if (remaining_content_size == 0)
  {
    return nlines;
  }

  _mm_store_si128((__m128i *) array_result, block[block_index]);

  for (int i = 0; i < remaining_content_size; i++)
  {
    nlines += (array_result[i] == '\n');
  }

  return nlines;
}

unsigned long long int count_file_lines(int fd)
{
  static __m128i block[BLOCK_SIZE];
  ssize_t buffer_size;
  unsigned long long int nlines = 0;

  while ((buffer_size = read(fd, block, BLOCK_SIZE)) > 0)
  {
    nlines += count_block_lines(block, buffer_size);
  }

  return nlines;
}

int main(int argc, char *argv[])
{
  if (argc < 2)
  {
    fputs("Usage: ./countlines <filepath>\n", stderr);
    return EXIT_SUCCESS;
  }

  int fd = open(argv[1], O_RDONLY);
  if (fd < 0) {
    perror("Error on open file");
    return EXIT_FAILURE;
  }

  printf("%llu\n", count_file_lines(fd));
  close(fd);

  return 0;
}