Re: [SMARTY] yet another benchmark

From: Date: Fri, 14 Jun 2002 14:25:56 +0000
Subject: Re: [SMARTY] yet another benchmark
References: 1  Groups: php.smarty.general 
Request: Send a blank email to php-smarty-general+get-4838@lists.php.net to get a copy of this message
Yes, you should demonstrate performance in realistic situations, which is what the current benchmarks do! If you try to benchmark one element of the engine againts one element of another engine, you may be comparing apples to oranges. Completely non-conclusive IMHO. Monte Ben Godfrey wrote:
I agree with Ivo. I think that the purpose of the benchmark should be to demonstrate to potential users how Smarty will perform against it's rivals in a set of realistic situations. The user wants to know why Smarty is great for them, which is that when you use, it will deliver 1 million pages a minute off the back of a Celeron 500, not that it uses PHP style XYZ. The user shouldn't need to know whether Smarty parses then prints, prints then parses or makes toffee. She needs to know if it's fast. Also, I think the more technical you get, the more you will find yourself taking advantage of Smarty good bits and (inadvertently of course) slanting the tests in Smarty's favour. I raised the point a while back that the other template system authors might want to see the benchmark code, they would probably ask that you change the test to favour them. I think that you need to consider how Smarty will be used and test that more. Perhaps test producing 500 versions of one page but with different themes or creating a portal page from 25 sub-templates. IMHO! Ben On Friday, June 14, 2002, at 03:00 , Monte Ohrt wrote:
I still don't see how the benchmark results would be anything conclusive. For instance, Smarty parses/prints in one function. If I run display() 50 times in Smarty, it will parse/compile the first time and execute the compiled template 50 times. Fine. So in FastTemplate, I execute FastPrint() 50 times, which just echoes the template (parsing done). In ModeliXe, the page is cached so all subsequent calls come from the cache. So maybe Smarty takes 3 seconds, FastTemplate takes 2.7 seconds and Modelixe takes 3.1 seconds. What does this mean? One is faster at printing than the other is a parsing/printing? One engine is better than the other because of these numbers? I think the results would be non-conclusive, or even misleading. Ivo Jansch wrote:
Hi, Monte Ohrt wrote:
Well this is a tough benchmark to make fair. Some template engines combine the parse/print functionality, some do not. So a fetch of a template 50 times might be parsing and printing, or just printing. ModeliXe uses a cached copy so subsequent fetches are just a file fetch.
What is 'fair' in this case anyway? Benchmarks are there for users to find out which solution is the best for their needs. The purpose of any template engine is 'to turn a template into an html file'. How the engine does that is irrelevant. That Smarty parses the template only once, making it ultrafast, does not make it an 'unfair' template engine. Benchmarking template engines without measuring their 'core business' would be like trying to benchmark formula 1 cars by just measuring how fast you can put fuel in it. :) Greetings, Ivo
-- Monte Ohrt <monte@ispi.net> Director of Technology, ispi Inc. -- Smarty General Mailing List (http://smarty.php.net/) To unsubscribe, visit: http://www.php.net/unsub.php
-- Monte Ohrt <monte@ispi.net> Director of Technology, ispi Inc.

« previous php.smarty.general (#4838) next »