[X4U] OT - Unscientific Speed Test - PC-Mac-FileMaker

[Edward Hillenbrand]

On Jul 17, 2005, at 2:05 PM, Mark Phillips wrote:

> On Jul 17, 2005, at 12:44 PM, Edward Hillenbrand wrote:
>
>> On Jul 15, 2005, at 7:51 AM, revDAVE wrote:
>>
>>> Test 2: loop 1 to 20000 in filemaker 6...
>>>
>>>  activity Monitor = 96.5% CPU
>>
>> As a software developer this is unbelievable. If I'm understanding  
>> you correctly, your essentially telling FileMaker to count and do  
>> nothing during each iteration.  Right?
>>
>> I wrote a program similar to yours in C to speed test my system.   
>> On any modern system it should complete almost instantly.  On my  
>> single processor 1 Ghz G4 counting to 20000 and doing nothing  
>> takes 0.000544 seconds.  This is so fast that the program doesn't  
>> even show up in Activity Monitor.  If I do something like output  
>> some text during the loop then the program uses about 1% of the  
>> CPU. So, either FileMaker is unbelievably slow (someone said that  
>> had already been established) or I've misunderstood you.
>
> I am feeling feisty today so... Comparing a c/c++ program to  
> something like FileMaker isn't a fair comparison. C vs Pascal  
> works; FileMaker vs 4th Dimension works too. While your remark is  
> enlightening, it is a bit like comparing the speed test of a quark  
> and that of a minivan.
>
> Given the very high level of FileMaker "zero programming" scripting,

You're quite right.  FileMaker by its very nature should be slower.   
It's a compiled vs interpreted kind of thing.  The comparison,  
however, was to illustrate the difference between raw performance  
(what the machine is capable of) vs the apparent performance of the  
machine when using FileMaker.

> I suggest a better test would involve activities typical of the  
> business application of the software. Say, create 20,000 invoice  
> records, modify each of them, then delete them all one by one. Run  
> this on the target systems and see how you do.

No why would I want to do that, the point was only to show that you  
shouldn't judge your machine by the measure of a single application.   
FileMaker is a fine program I've even considered using it my self.   
I'm not trying to come of as an arrogant C programer.  I don't think  
that the solution to everything is to write a custom C application.   
On the contrary, you shouldn't build something yourself if it's  
already available to you.  Even Apple has embraced this concept with  
its use of open source software.

Now that being said, FileMaker took 19 seconds to "count" to 20,000  
and my C program took 0.000544 seconds, that means FileMaker is about  
34,926 times slower at "counting" than the machine is capable of.   
All I'm saying is that's interesting and by my book quite slow.  
Please don't read any further into it than that.