Question

I currently have this kind of loop

while(1)
{
    generate_string(&buffer);

    for(int i = 0; i < filelines; i++)
    {
        if(strcmp(buffer,line[i]) == 0)
        {
           /*  do something  */
        }
    }
}

I have a file with a few million strings(which hopefully should be cut by half sometime soon), the number of all these strings is stored in filelines

line[i] is basically where the string itself is stored.

Currently, due to the comparison of these million strings, function generate_string(&buffer); is executed around 42 times per second.
Is there a faster way to do string comparison in C?

Answer 1

strcmp is usually optimized by all vendors. However, if you’re not satisfied with this you can try:

• Lookup Burst Tries
• Use a suffix tree for fast string comparison — see this article
• Depending on the size of strings in your application you can write a custom string comparator. E.g: GNU libc used to have this optimization for small strings where they tested strings smaller than five bytes as integers. MS cl also has some optimizations for small-strings (do look it up).

But more importantly make sure strcmp is your real bottleneck.