Framework Benchmarking

So Paul M. Jones did some benchmarking but the numbers looked a little funny so I ran just the Zend benchmark on my own system, Will get back with further results later but here is what I found.

Pauls Benchmarks

Quick Run Base zend, no incubator just straight zend 0.2.0
Document Path: /
Document Length: 12 bytes
Requests per second: 405.95 [#/sec] (mean)
Time per request: 24.634 [ms] (mean)
Time per request: 2.463 [ms] (mean, across all concurrent requests)
Transfer rate: 80.88 [Kbytes/sec] received

Test.html
Document Path: /test.html
Document Length: 12 bytes
Requests per second: 2234.28 [#/sec] (mean)
Time per request: 4.476 [ms] (mean)
Time per request: 0.448 [ms] (mean, across all concurrent requests)
Transfer rate: 595.66 [Kbytes/sec] received

HTML Vs Zend, 405 / 2234, you come out with PHP running at only 18% compared to plain HTML, however Pauls example, 12 / 808 runs more like 1.5%???

Of course I didn't use incubator, and I made a couple code changes as I couldn't get his to run, for example why does he call dirname(dirname(__FILE__)) . '/views/index.php', plus I had problems getting certain files to load, I suspect he ended up with a mix of 0.2.0 normal and incubator code.

Basically it looks like he threw everything in a single directory, I normally put library files outside the main path.. Got his code working at get the following results.

Zend 0.2.0 with incubator hopefully being used?? Files match Pauls exactly now.
Document Path: /
Document Length: 12 bytes
Requests per second: 409.39 [#/sec] (mean)
Time per request: 24.426 [ms] (mean)
Time per request: 2.443 [ms] (mean, across all concurrent requests)
Transfer rate: 81.55 [Kbytes/sec] received

No real diffrence, matter fact no matter what I do I can not replicate the extreme poor performance he managed to got. Matter fact, all his stats are so bad, with Solar only getting 5% vs PHP that I would suspect something is up on his machine.

Will try to run the same tests he ran on solar and symfony later.