For a 6.7GB file, takes 2.2s. Compiled with clang -Ofast -Wall -Weverything -pedantic -std=c11 -fopenmp -flto=thin main.c.

histogram.c

#include <stdio.h>
#include <stdlib.h>
#include <omp.h>
#include <stdint.h>

#define BUFF_SIZE 1024*1024*16
#define UCHAR_MAX 255

int main(void) {
    FILE *f = fopen("/home/connorbaker/storage/large_test_file-02", "r");

    if (NULL == f) {
        printf("Well damn. File not found.\n");
        exit(1);
    }

    size_t *map = (size_t *) calloc(UCHAR_MAX + 1, sizeof(size_t));
    char *arr = (char *) calloc(BUFF_SIZE, sizeof(char));
    size_t read;

    size_t iterations = 0;

    while ((read = fread(arr, sizeof(char), BUFF_SIZE - 1, f)) != 0) {
        #pragma omp parallel
        {
            size_t local_map[UCHAR_MAX + 1] = {0};

            #pragma omp for
            for (size_t i = 0; i < read; i++) {
                local_map[(int) arr[i]]++;
            }

            #pragma omp critical
            for (int i = 0; i <= UCHAR_MAX; i++) {
                map[i] += local_map[i];
            }
        }
        iterations++;
    }

    for (int i = 0; i <= UCHAR_MAX; i++) {
        if (map[i] != 0) {
            printf("%c = %lu\n", (char) i, map[i]);
        }
    }

    printf("%zu iterations.\n", iterations);


    fclose(f);
    free(map);
    free(arr);

    return 0;
}