I'll bite. I write CakePHP apps for my clients, and I've heard this "cake is slow" meme before. Never been a problem for us, and we've shipped a lot of apps.

In almost all cases, the benchmark was done with no caching. I don't know who launches web apps with no caching, but they shouldn't be doing benchmarks.

I don't concede that Cake is slow for any meaningful definition, but as a matter of point - I'll trade faster developers for a couple CPU cycles any day of the week.

I sort of agree with you - if you need twitter type scale, you will need to rewrite.

Most frameworks are bloated (not setup to REALLY scale) out of the gate.

CakePHP is very flexible, and if used correctly can scale for 99% of workloads. Especially if you leverage some of the php5 only nuances (cake 1.3 has been developed to support php4 and php5).

CakePlate can steer you in the right direction - but is not an end all. It helps turn your idea into reality quicker. If your idea twitter scales, then you have cash to throw at the (good) problem.

If you give Cake an honest shot, I'm confident you will like it.

Could you expand on what you see as the bloat? I find the base distribution of CakePHP to be very lean, and using it on half dozen production projects, I have found no issues with speed, especially with proper use of memcached and media compression.

These benchmarks are from two years ago. The version of Cake used isn't even a stable release (1.2 RC1). Either way 1.3 has replaced 1.2 and is plenty fast for 99.99999% of sites.

No to mention that these kind of things are going to be hard to really measure. Each framework is doing such diverse things that each website will get different performance. Also I wouldn't trust a framework maker to do these tests.

Another benchmark source... http://www.yiiframework.com/performance/

Take with a grain of salt.

This benchmark seems to be either out dated or somewhat falsified. I ran the same benchmark on my system and CI out performed Yii ( https://github.com/JosephMoniz/BlueElephant/blob/master/docs... ) . The main point though is that with or without caching (APC) CakePHP is the slowest by an order of magnitude.

EDIT: yeah, pretty sure the original Yii benchmark is dated as the graph has older versions while the phpmark repo has slighlty newer ones ( http://code.google.com/p/phpmark/source/browse/#svn/trunk )

http://www.phpkode.com another PHP Framework source

Here's a more recent benchmark: http://symfony-reloaded.org/fast

Odd that they didn't include codeigniter in the test

Probably because if they had, wouldn't have been the fastest in all tests, or at least not by much.

fair enough, it was the newest one i could find.

this and similar reports from others are the reason i don't use cake, though, and it seems like it would be in their best interest to get some newer information out there. (i made my choices years ago though)

agreed. i like codeigniter, which is an order of magnitude or two faster than cake.

Consider Kohana, then, which forked off from CodeIgniter years ago and learned from the many mistakes CI made (most notably being a complete inability to adapt), and was designed with PHP 5 in mind from day one.

i considered it in the past, and i might make the switch with all the internal crap going on with CI. but my biggest current project is CI-based and making the conversion would be...tough. plus, their docs suck.

It looks like they've been working with the community to try and clear a lot of that stuff up, I really enjoy using CI myself so I've been interested. http://codeigniter.com/news/the_official_codeigniter_communi...

You may also want to look at CI's 2.0 development branch.

They've rebuilt much of the base code on top of PHP 5 natively now instead of being built on PHP 4 with hacks to make it also work with 5.

Although it is still in development, it has been said that it is stable.

i use it already. it isn't anything especially different.

Can you expand on "all the internal crap going on with CI"? I'm building with CI for a project so I'm curious.