|
|
Posted by Jochem Maas on 06/06/05 16:13
Mark Cain wrote:
> I see your point about including the timing code in the calculation itself
> and while it does add time to the process, he is not just timing this
> code -- he is timing this code against another. Meaning we don't want to
> know "How fast is this code?" per se -- we want to know "Which is faster?"
> So, arguably from the faster point of view, as long as the time mech is in
> both tests, the question will be answered with a degree of certainty. From
> the days of my boyhood, even if the race route were "two times around the
> big oak tree," the faster runner would still win the race.
>
> Question for you, please:
>
> In your post you say:
>
>
>>time. For example, using something like ab will let you test your code
>>in its raw form - without the timing and looping.
>
>
> What is ab? I am somewhat limited in my depth and scope of php and so I
> have never seen this before. Care to point me in a direction where I could
> learn more?
apache benchmark, try:
> man ab
on your local linux cmd line (assuming you have apache installed)
>
> Mark Cain
>
>
>
>
> ----- Original Message -----
> From: "Chris Shiflett" <shiflett@php.net>
> To: "Mark Cain" <mark@markcain.com>
> Cc: "Andy Pieters" <mailings@vlaamse-kern.com>; <php-general@lists.php.net>
> Sent: Sunday, June 05, 2005 9:56 PM
> Subject: Re: [PHP] What is faster?
>
>
>
>>Mark Cain wrote:
>>
>>>I checked the first expression with 1,000 iterations and it
>>>took 0.00745 seconds.
>>>
>>>Here is the code I used.
>>
>>[snip]
>>
>>
>>>$start1 = vsprintf('%d.%06d', gettimeofday());
>>
>>Although many would argue that it's pointless to worry over such small
>>details (and they have valid arguments), it's better to construct your
>>benchmark so that the timing mechanism impacts your code as little as
>>possible.
>>
>>Although consistency is difficult to achieve without some effort, you
>>can at least get your timing mechanism out of the code that you want to
>>time. For example, using something like ab will let you test your code
>>in its raw form - without the timing and looping.
>>
>>I also dislike unit tests that are part of the code they're meant to
>>test. :-)
>>
>>Hope that helps.
>>
>>Chris
>>
>>--
>>Chris Shiflett
>>Brain Bulb, The PHP Consultancy
>>http://brainbulb.com/
>>
>>
>
>
Navigation:
[Reply to this message]
|