There are so many performance problems in this code that it's kind of hard to know where to begin! Here's a few that jump out right away:
- Don't open file 2 for each line of file 1 and read through every line! If there are 1 million lines in file 1 and 500 thousand lines in file 2 then you'll read 500 billion lines from file 2! Instead read file 2 once and re-use the hash for each lookup.
- Don't re-sort all the keys from file 1 everytime you read a line from file 1. (UPDATE: Looking again I see that %fets is actually local to the while(). Why are you using a hash at all here? Why are you calling sort() when only one key is present?)
- You may not have enough memory to actually hold all of file 2 in memory at once. If you don't you'll run into swap, which will be slow no matter what you do. You can fix this by storing the hash in a database file via DB_File or something similar.