phpmark - Project Hosting on Google Code

Earlier this year Apr 25, 2009 issue 2 (Add Kohana Framework) reported by jeff@tjwallace.ca - Please add the Kohana framework for more comparisons. http://kohanaphp.com/ Please add the Kohana framework for more comparisons. http://kohanaphp.com/ Mar 01, 2009 r22 (Added Yii 1.0.3.) committed by qiang.xue - Added Yii 1.0.3. Added Yii 1.0.3. Feb 07, 2009 issue 1 (Flawed benchmarks) commented on by drechin - http://blog.orite.com.au/tag/yii-framework/ shows what some may consider a more proper benchmark. The thing with benchmarks is that they are not precise, and any application can be tuned to a certain benchmark but that does not necessarily make it better. An overall comparison like how CPUs and GPUs are compared would be a better benchmark. http://blog.orite.com.au/tag/yii-framework/ shows what some may consider a more proper benchmark. The thing with benchmarks is that they are not precise, and any application can be tuned to a certain benchmark but that does not necessarily make it better. An overall comparison like how CPUs and GPUs are compared would be a better benchmark. Jan 31, 2009 r21 ([No log message]) committed by qiang.xue - [No log message] [No log message] Jan 31, 2009 r20 (Added yii 1.0.2.) committed by qiang.xue - Added yii 1.0.2. Added yii 1.0.2. Jan 22, 2009 r19 ([No log message]) committed by qiang.xue - [No log message] [No log message] Jan 22, 2009 r18 ([No log message]) committed by qiang.xue - [No log message] [No log message] Jan 22, 2009 r17 (Added CakePHP 1.2.1. Added Symfony 1.2.2. Changed all tests ...) committed by qiang.xue - Added CakePHP 1.2.1. Added Symfony 1.2.2. Changed all tests from a die() statement to an echo statement. Added CakePHP 1.2.1. Added Symfony 1.2.2. Changed all tests from a die() statement to an echo statement. Jan 22, 2009 r16 (Added zend 1.7.3) committed by qiang.xue - Added zend 1.7.3 Added zend 1.7.3 Jan 22, 2009 r15 (fixed external dependency.) committed by qiang.xue - fixed external dependency. fixed external dependency. Jan 11, 2009 issue 1 (Flawed benchmarks) commented on by qiang.xue - Please read my earlier comments. Also please take a look at Rasmus' "simple is hard" presentation: http://talks.php.net/show/drupal08 Even for this simple "hello world", a framework needs to undergo many steps: initialization, resolving the request and dispatching it to the appropriate handler. And this is exactly what this project is trying to identify. Any overhead beyond this is the responsibility of application developers. If you insist in comparing frameworks with more realistic applications, please find it in a different place or do it yourself. I highly doubt you could do it correctly or in a fair way. Please read my earlier comments. Also please take a look at Rasmus' "simple is hard" presentation: http://talks.php.net/show/drupal08 Even for this simple "hello world", a framework needs to undergo many steps: initialization, resolving the request and dispatching it to the appropriate handler. And this is exactly what this project is trying to identify. Any overhead beyond this is the responsibility of application developers. If you insist in comparing frameworks with more realistic applications, please find it in a different place or do it yourself. I highly doubt you could do it correctly or in a fair way. Jan 10, 2009 issue 1 (Flawed benchmarks) commented on by kureikain - I think that pcdinh's idea is correct! You can say that 'hello world' don't need any extra working! And you used die()? But please worry about real web application? With die(), you almost don't make anything with your frameworks! Please change your approach! I think that pcdinh's idea is correct! You can say that 'hello world' don't need any extra working! And you used die()? But please worry about real web application? With die(), you almost don't make anything with your frameworks! Please change your approach! Jan 02, 2009 issue 1 (Flawed benchmarks) commented on by qiang.xue - I know that frameworks usually do extra work after an action and a die() would 'early' terminate an application. The question is that these extra work are NOT essential to the 'hello world' display. So using die() is similar to what we do to disable session, logging and other non-essential features of a framework. They all serve for the same goal: what is the fastest way to produce some output by going through the typical life cycles of a framework? I know that frameworks usually do extra work after an action and a die() would 'early' terminate an application. The question is that these extra work are NOT essential to the 'hello world' display. So using die() is similar to what we do to disable session, logging and other non-essential features of a framework. They all serve for the same goal: what is the fastest way to produce some output by going through the typical life cycles of a framework? Jan 02, 2009 issue 1 (Flawed benchmarks) commented on by 1stvamp - I think you completely misunderstood his point. The point is the output stage of some frameworks may hit the die() statement before other stages, and so by killing the application at this point you are treating some frameworks more evenly than others. Let the script complete as normal with a echo statement, otherwise the benchmark is utter bumph. I think you completely misunderstood his point. The point is the output stage of some frameworks may hit the die() statement before other stages, and so by killing the application at this point you are treating some frameworks more evenly than others. Let the script complete as normal with a echo statement, otherwise the benchmark is utter bumph. Older Dec 31, 2008 issue 1 (Flawed benchmarks) Status changed by qiang.xue - Please understand the goal of this benchmark : to find out the minimal overhead of each framework. No matter how powerful a framework is, it needs to handle the simplest case like 'hello world'. For example, an AJAX request would want to show the current server time. In this more realistic request, neither database nor full view rendering would be involved. And this scenario is quite similar to the 'hello world' case. As a matter of fact, in today's Web 2.0 applications, frameworks need to be more careful with performance because it needs to handle much more but smaller requests like the example above. Having explained as above, we should also conclude that it is not an easy task to make a fair comparison even for this simple 'hello world' test. If you include database and full view rendering into the test, you will find there are more unfairness involved. For example, in one framework, caching is available to make DB query or view rendering faster, while in another framework, there is simply no caching feature at all. As always, if you find the test application for a framework in this benchmark is not fully optimized, please let us know. We are more than happy to update it. Thanks. Status: Invalid Please understand the goal of this benchmark : to find out the minimal overhead of each framework. No matter how powerful a framework is, it needs to handle the simplest case like 'hello world'. For example, an AJAX request would want to show the current server time. In this more realistic request, neither database nor full view rendering would be involved. And this scenario is quite similar to the 'hello world' case. As a matter of fact, in today's Web 2.0 applications, frameworks need to be more careful with performance because it needs to handle much more but smaller requests like the example above. Having explained as above, we should also conclude that it is not an easy task to make a fair comparison even for this simple 'hello world' test. If you include database and full view rendering into the test, you will find there are more unfairness involved. For example, in one framework, caching is available to make DB query or view rendering faster, while in another framework, there is simply no caching feature at all. As always, if you find the test application for a framework in this benchmark is not fully optimized, please let us know. We are more than happy to update it. Thanks. Status: Invalid Dec 30, 2008 issue 1 (Flawed benchmarks) reported by pcdinh - Dear all, Please fix your benchmark approach. A framework does not work as if it is a single page script. So please don't use die('hello world'); in controllers. Just use echo 'hello world'; to allow frameworks do the rest of its processing. By calling die(), you makes frameworks run immaturely. It is not fair. Dear all, Please fix your benchmark approach. A framework does not work as if it is a single page script. So please don't use die('hello world'); in controllers. Just use echo 'hello world'; to allow frameworks do the rest of its processing. By calling die(), you makes frameworks run immaturely. It is not fair. Dec 28, 2008 r14 (This marks the benchmark performed on Dec. 26, 2008) committed by qiang.xue - This marks the benchmark performed on Dec. 26, 2008 This marks the benchmark performed on Dec. 26, 2008 Dec 26, 2008 r13 ([No log message]) committed by qiang.xue - [No log message] [No log message] Dec 26, 2008 r12 ([No log message]) committed by qiang.xue - [No log message] [No log message] Dec 26, 2008 r11 ([No log message]) committed by qiang.xue - [No log message] [No log message] Dec 26, 2008 r10 ([No log message]) committed by qiang.xue - [No log message] [No log message] Dec 26, 2008 r9 (added cakephp 1.1.20 and 1.2.0.) committed by qiang.xue - added cakephp 1.1.20 and 1.2.0. added cakephp 1.1.20 and 1.2.0. Dec 26, 2008 r8 (added zend 1.6.2 and 1.7.2) committed by qiang.xue - added zend 1.6.2 and 1.7.2 added zend 1.6.2 and 1.7.2 Dec 26, 2008 r7 (added yii 1.0.0 and 1.0.1) committed by qiang.xue - added yii 1.0.0 and 1.0.1 added yii 1.0.0 and 1.0.1 Dec 26, 2008 r6 (added symfony 1.1.6 and 1.2.1) committed by qiang.xue - added symfony 1.1.6 and 1.2.1 added symfony 1.1.6 and 1.2.1 Dec 26, 2008 r5 ([No log message]) committed by qiang.xue - [No log message] [No log message] Dec 26, 2008 r4 (added prado 3.1.3) committed by qiang.xue - added prado 3.1.3 added prado 3.1.3 Dec 26, 2008 r3 (added CI 1.7.0) committed by qiang.xue - added CI 1.7.0 added CI 1.7.0 Dec 26, 2008 r2 (Added baseline and CI 1.6.3) committed by qiang.xue - Added baseline and CI 1.6.3 Added baseline and CI 1.6.3 Dec 26, 2008 Project phpmark created by qiang.xue - This project compares different PHP frameworks This project compares different PHP frameworks

issue 1 (Flawed benchmarks) Status changed by qiang.xue - Please understand the goal of this benchmark : to find out the minimal overhead of each framework. No matter how powerful a framework is, it needs to handle the simplest case like 'hello world'. For example, an AJAX request would want to show the current server time. In this more realistic request, neither database nor full view rendering would be involved. And this scenario is quite similar to the 'hello world' case. As a matter of fact, in today's Web 2.0 applications, frameworks need to be more careful with performance because it needs to handle much more but smaller requests like the example above. Having explained as above, we should also conclude that it is not an easy task to make a fair comparison even for this simple 'hello world' test. If you include database and full view rendering into the test, you will find there are more unfairness involved. For example, in one framework, caching is available to make DB query or view rendering faster, while in another framework, there is simply no caching feature at all. As always, if you find the test application for a framework in this benchmark is not fully optimized, please let us know. We are more than happy to update it. Thanks.

Status: Invalid

Please understand the goal of this benchmark : to find out the minimal overhead of each framework. No matter how powerful a framework is, it needs to handle the simplest case like 'hello world'. For example, an AJAX request would want to show the current server time. In this more realistic request, neither database nor full view rendering would be involved. And this scenario is quite similar to the 'hello world' case. As a matter of fact, in today's Web 2.0 applications, frameworks need to be more careful with performance because it needs to handle much more but smaller requests like the example above. Having explained as above, we should also conclude that it is not an easy task to make a fair comparison even for this simple 'hello world' test. If you include database and full view rendering into the test, you will find there are more unfairness involved. For example, in one framework, caching is available to make DB query or view rendering faster, while in another framework, there is simply no caching feature at all. As always, if you find the test application for a framework in this benchmark is not fully optimized, please let us know. We are more than happy to update it. Thanks.

Status: Invalid

Earlier this year

Older