Had to wirte and optimize a C++ program that finds for 1000x1000x1000 grid points the 100 nearest points for each (with an additional factor to make it more complicated).
It had to run in under 18 minutes to pass. No matter what I did I couldn't get it fast enough. I tried kd-trees, caching of certain points, optimizing distamce calculations by ommiting any irrelevant factor, saving points' calculated squares etc etc. When Ibwas down to 20 minutes, I realized, that my makefile had an error and ignored the - O3 flag...
Well, it actually ran 5 minutes with -O3.

Why does it take so loooong!

Github Inc. (Feel good inc. parody)
=========================

Ha-ha-ha-ha-ha-ha-ha-ha-ha-ha-ha-ha-ha.
Github.
Fetch it, fetch it, fetch it, Github.
Fetch it, fetch it, fetch it, Github.
Fetch it, fetch it, fetch it, Github.
Fetch it, fetch it, fetch it, Github.

Fetch it, fetch it, fetch it, Github.
Fetch it, fetch it, fetch it, Github.

(change) Fetch it (change), Fetch it (change), Fetch it (change), Github
(change) fetch it (change), fetch it (change), fetch it (change), Github

Repos breaking down on pull request
Juniors have to go cause they don't know wack
So while you filling the commits and showing branch trees
You won't get paid cause it's all damn free

You set a new linter and a new phenomenal style
Hoping the new code will make you smile
But all you wanna have is a nice long sleep.
But your screams they'll keep you awake cause you don't get no sleep no.

git-blame, git-blame on this line
What the f*ck is wrong with that
Take it all and recompile
It is taking too lonnng

This code is better. This code is free
Let's clone this repo you and me.
git-blame, git-blame on this line
Is everybody in?

Laughing at the class past, fast CRUD
Testing them up for test cracks.
Star the repos at the start
It's my portfolio falling apart.

Shit, I'm forking in the code of this here.
Compile, breaking up this shit this y*er.
Watch me as I navigate.
Ha-ha-ha-ha-ha-ha.

Yo, this repo is Ghost Town
It's pulled down
With no clowns
You're in the sh*t
Gon' bite the dust
Can't nag with us
With no push
You kill the git

So don't stop, git it, git it, git it
Until you're the maintainers
And watch me criticize you now
Ha-ha-ha-ha-ha.

Break it, break it, break it, Github.
Break it, break it, break it, Github.
Break it, break it, break it, Github.
Break it, break it, break it, Github.

git-blame, git-blame on this line
What the f*ck is wrong with that
Take it all and recompile
It is taking too lonnng

This code is better. This code is free
Let's clone this repo you and me.
git-blame, git-blame on this line
Is everybody in?

Don't stop, shit it, git it.
See how your team updates it
Steady, watch me navigate
Aha-ha-ha-ha-ha.

Don't stop, shit it, git it.
Peep at updates and reconvert it
Steady, watch me git reset now
Aha-ha-ha-ha-ha.

Github.
Push it, push it, push it, Github.
Push it, push it, push it, Github.
Push it, push it, push it, Github.
Push it, push it, push it, Github.

When you take your fast but plain cgi scripts and convert them to fastcgi...
"no way the page parsed and analyzed 325 XML trees this fast" 😇

I just realized that I subconsciously believe more lines of code means slower code.

It's not intellectual. I understand that little lines of code often are just calling other code. That this is not how Big O works or does not replace benchmarking and that some data structure requires a lot of code for immense speed up. E.g: B-Trees with sizes at page size for big amounts of data read from a secondary storage location.

But still, when I see a function with just 3 to 5 lines, my inner monkey believes it must be fast.

Know your biases, I guess.