AutoHotkey Community

[Titan]

This function times the running length of each code snippet with very high precision. It more advanced than the standard benchmark as it calculates deviation and ranks code on performance.

qpc2(Code|Source [, "pre"|"inc"])
Code: AutoHotkey code, if this is a file its contents will be read and used
Options: use "pre" to include the code before the test (i.e. initializing variables) or "inc" will save the code at the end of the script (i.e. for deleting temporary files). Leave blank to test the code

result := qpc2([Iterations])
Output in TSV format


Example:

[Laszlo]

I'd give it a try, if you posted a link to download…

[Titan]

lol thanks.
_________________

RegExReplace("irc.freenode.net/autohotkey", "^(?=(.(?=[\0-r\[]*((?<=\.).))))(?:[c-\x73]{2,8}(\S))+((2)|\b[^\2-]){2}\D++$", "$u3$1$3$4$2")

[Laszlo]

It works, fine, but could you give a little more info:
- how does it work, what does it do?
- how many iterations does it perform and why that many?
- what is the Rank?

[Titan]

Like the previous version it repeats each given code to get the average length of time they take to complete. The number is calculated with the systems performance frequency, such that the integral of resulting execution times is loosely normalized with 1.8 seconds. Since this is random, depending on CPU load and thread interruption, an efficiency ratio (rank) is calculated to express deviation as a percentage. Standard deviation just tells you the range of variation; a value of <0.01 means the methods are almost identical performance wise and any could be used, but for anything greater than 0.03 you should use the method with the highest rank to increase the efficiency of your script.
_________________

RegExReplace("irc.freenode.net/autohotkey", "^(?=(.(?=[\0-r\[]*((?<=\.).))))(?:[c-\x73]{2,8}(\S))+((2)|\b[^\2-]){2}\D++$", "$u3$1$3$4$2")

[Titan]

In the latest update I fixed an error that caused some information to be deleted from the results.

Here is another example which I used to test my XPath function:

[toralf]

Dear Titan,

Because of the path functions we worked on I looked at your benchmark script. But it is kind of hard to decipher. I tried it with your XPath test script, but that doesn't give much infomation on what is tested either. Could you please add some comments to the benchmark code and give a clearer information on how it is supposed to be used. For me it doesn't make sense to use it if I do not know what is measured and what the values mean and how I have to interpret them.

Thanks. I think the script can be of great help. If it only would be easier to understand.
_________________
Ciao
toralf

[Titan]

Sure, I'll explain the example in my first post. It tests to see which method is the fastest to get an integer from a float. The first call sets a variable to 2.5 (using the "pre" option) which will be the subject of the test. The second call uses the floor function, then an integer divide and finally a SubStr method to exclude anything past the '.' character. The last call returns the results. The Rank column determines which method is best, in this case the SubStr method came out as 91% which means its superior to the other methods. It also had a relative time count of 32 which is much lower 91 or 216. If you want to know how the rank is calculated read my response to Laszlo.

This is what I used to compare our absolute path functions: