Well I setup my own little test, running just a bit of code in a for loop and printed the current system time at the beginnig and end of the loop; so time was only recorded in seconds and not down to miliseconds. I first tested the old code from a few days ago that got all the gpu info. I ran it in a for-loop from 1 to 10,000. It took 38 minutes!
Two days ago though I knew I could fix it up can make less system calls and take better use of perls pattern matching. I ran that in a for-loop from 1 to 100,000 can it only took 21 seconds!
Then I made a slight change to that, making use of the built in scaler variable and in a for-loop from 1 to 100,000 it tok ~1 second!
I couldn't believe the difference. The program wasn't slow to begian with, original only took ~2 seconds from command to GUI. I couldn't believe the speed difference though between the different styles.
Note: The 10,000 interations in the first test isn't a typo. After the last two test at 10,000 interations I could tell they were a lot faster, so I moved the interation up to 100,000 for the last two.
I'll try the articles speed test method later, as it looks like it'll take a bit longer to implement and try.
The article gave a good idea that I'll use....more hashes.