Reordering bytes in memory, then writing to file

Question

I have a block of data in memory starting at memory_ptr. What this program does is byte reverse every qword and then write that to a file. For example, 18171615141312112827262524232221 will be written as 11121314151617182122232425262728 and so on.

I want this to run as fast as possible, so I've stored the flipped bytes in memory, and then write it to the file once its at 8KB to reduce the # of calls to fwrite(). However, I feel like there is a simpler and faster way of doing this rather than increasing the size of malloc to further reduce calls to fwrite. Any ideas on how to speed this up?

#define htonll(x) (((uint64_t)htonl((x) & 0xFFFFFFFF) << 32) | htonl((x) >> 32))

uint64_t data;
int total_chunks = 1000;
int chunk_size = 8192;
char *tmp = malloc(8192);
FILE *fp = fopen("data.bin","wb");

for (int i = 0; i < total_chunks; i++) {
    for (int j = 0; j < chunk_size; j+=8) {
        memcpy(&data, memory_ptr, 8);
        data = htonll(data);
        memcpy(tmp+j, &data, 8);
        memory_ptr+=8;
    }
    fwrite(tmp, 8192, 1, fp);
    memset(tmp,0,8192);
}