It seems like you didn't understand what I was trying to explain. Have you received my E-mail containing my benchmarks? If so compare them to the ones the Yii author performed.
My "benchmarks" (quoted because I didn't intend to perform real benchmarks) were performed on a Windows XP everyday-computer (meaning a computer running software for daily use), his were performed on a Windows Vista everyday-computer (which would like NEVER be used for production but let's keep that aside for now). My benchmarks showed that Zend Framework performed better than Yii & Yiilite on a Windows XP + Apache + mod_php. This proves that there something went wrong with his ZFW benchmarks. And APC performing worse proves this even more. You have a good point meaning that since they were performed on the same machine while running the same software the results of Yii performing better than ZFW in general would hold some truth. Well, my benchmarks prove the opposite and I would agree with you if they were performed on a Linux + Apache production server, or in case of a Windows system a Windows (2003, etc) server + IIS. Why? Because benchmarks are supposed to be performed on a system running only software that would be used for production. And by performing them on a production server would take away the chance of other irrelevant software inflicting the results. "There is no Windows Vista-way and a Linux way", perhaps not but there is a real difference for software that is optimized for a certain OS. ZFW performs better than Yii on a Windows XP machine and would probably also perform better on a Linux OS (not talking about a production system here) so no one would go for the Vista benchmarks which show that ZFW performs worse than every other framework. And lastly, there is also a huge difference between the benchmarks of a framework in general (just a simple "hello world" script) and the ones of a real web application using all the features a framework has to offer in order to improve the performance (caching, etc) of that application. So in my opinion, if you plan to put some benchmarks on your website I would at least want to see the benchmarks of the framework in general and the benchmarks of a real web application using that framework. I suggest you read Matthew's E-mails on how they perform the benchmarks: http://www.nabble.com/Framework-speed-shotout----question-td19914787.html#a19914787 On Mon, Nov 3, 2008 at 2:17 PM, ekerazha <[EMAIL PROTECTED]> wrote: > > > Matthew Weier O'Phinney-3 wrote: > > > > You miss several important issues: it's not necessarily about testing on > > Windows, it's the entire methodology. If you want to have a good > > benchmark on windows, use a reasonable production environment: dedicated > > machine, using IIS + FastCGI. Apache on Windows is notoriously > > non-performant and unoptimized. Additionally, benching on a personal > > box, and not one tuned for production, will have skewed results due to > > having other applications and processes running. > > > > Really, what you call "important" is insignificant too, for the same reason > I've already explained: every framework was running on Windows Vista with > Apache, the environment was the same and it was a fair comparative. That > "you should use IIS + FastCGI" argument is not relevant, every framework > didn't use "IIS + FastCGI". > > "PHP for IIS + FastCGI" would be another nice label though ;-) > -- > View this message in context: > http://www.nabble.com/Framework-speed-shotout----question-tp19914787p20302597.html > Sent from the Zend Framework mailing list archive at Nabble.com. > > -- Isaak Malik Web Developer
