I was working on some code last week that, when I applied it to the full set of data, took about 1 hour and 10 minutes to run. Needless to say, this is quite slow, however I was using unoptimized code. I went for correctness first. Anyway, the first thing I noticed was that I could eliminate a good deal of array copying. I fixed that and it cut the time down to 23 minutes. This is still a lot slower than I wanted, so I went back to trying to optimize the code. After trying various things and researching some different things to optimize in Perl, I tried something completely and utterly stupid: I changed push to unshift. Now, I could do this since I was sorting the array later anyway, so ordering did not matter at this point.
Did it work? The code ran in 34 seconds. I must say, I was quite shocked by this. This leads me to my question: what's the difference between push and unshift in terms of how they work? I know they put data on different sides of the array. I'm just wondering why that made such a profound performance difference.
