Many know of the famous quote, "Premature optimization is the root of all evil," but most people do not know the full quote or understand the context in which optimization is considered evil. As with anything in programming optimization is evil, maybe. Stop using excuses for slow code, and start to think about the places and tools that you can use to optimize. Thankfully there are are many different tools like xhprof, Valgrind, and others to help us out and properly optimize our code for those times when we need to dig deep into our code.


Comments are closed.

Nicely lays out when optimization is worth the time!

Dana Luther at 09:31 on 12 Feb 2018

Loved this - a clear demonstration of how a little bit of optimization applied in the right places can make a tremendous difference. I particularly appreciated the information about the tools for finding out where that optimization is needed and am looking forward to exploring those and digging into a few of the projects that I work on with them.
The presentation was clear and engaging, quite a feat at the end of a long weekend when everyone is tired.

I enjoyed this quite a bit. I like the humor incorporated into the talk. A lot of the slides being pictures does make the slides a little less meaningful outside the context. Specifically the "Premature optimization is the root of all evil" slide. I'm definitely going to have to go and look that up outside of the context.

Overall, I really liked the talk, I'll definitely be talking about and trying out Xhprof at work! I'm also interested in looking up the Valgrind output for PHP. I've never even thought to try that before!

Sonny Savage at 15:03 on 13 Feb 2018

Great step-by-step in following the evidence to fix the problem. The "dumpster fire" slide made me laugh!

Omni Adams at 17:37 on 14 Feb 2018

-1 was hoping for more talk about micro-optimizations: ' vs " or echo vs print.

Seriously though, the things you talked about would be so useful for many of the developers I've worked with that want to just trust their gut on what needs to be optimized. They feel like part X is the slow part without measuring and throw caching or unroll loops or whatever at it, and they shave a quarter of a percent off their timing. If they had measured they might have seen that part Y was where the problem was and that a small fix there would cut request time in half.

Could have used animated gifs instead of those static ones.

Good talk and considerations to have regarding optimization.