Re: How does compare gcc to VS C++ ?

From: John-Paul Stewart (jpstewart_at_binaryfoundry.ca)
Date: 08/10/04 

Date: Tue, 10 Aug 2004 17:45:08 -0400

Jason Bowen wrote:
> John-Paul Stewart wrote:
>
>> Jason Bowen wrote:
>>
>>> John-Paul Stewart wrote:
>>>
>>>> Jason Bowen wrote:
>>>>
>>>>> Dave Uhring wrote:
>>>>>
>>>>>>
>>>>>> There is no possible way to compare the Microsfot pumpkin to the
>>>>>> Linux
>>>>>> apple.
>>>>>>
>>>>>
>>>>> There sure is. Consider an application written in C that is
>>>>> computationaly intensive with a command line interface. If the
>>>>> Microsoft compiler created an executable that finished the problem
>>>>> 10% faster everytime you'd be a fool to not chose it, especially if
>>>>> it was something that ran 24/7.
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> But is that hypothetical 10% difference due to the compiler alone,
>>>> the O/S alone, or both together? Comparing the MS compiler on
>>>> Windows to GCC on Linux doesn't tell you anything about the
>>>> compiler. You're comparing both the compiler and the O/S
>>>> simultaneously.
>>>
>>>
>>>
>>>
>>> Did you read what I wrote? I'm not taking into account the operating
>>> system.
>>
>>
>>
>> Sure you are. (Well, *you* aren't, but your experiment is.) If
>> you're using VC++ on Windows and comparing it to GCC on Linux, whether
>> or not you realize it, you're comparing the effeciencies of the two
>> O/Ses. Period. GCC on Windows would be a valid comparison of compilers
>> only, but way off-topic in this newsgroup.
>>
>
> Christ, you don't get it. I'm comparing nothing other than code
> generation.

Then explain to me how you can be certain the observed performance
difference can be attributed solely to code generation.