[macstl-dev] Re: [altivec] Secret incantations of the vector masters...

[Glen Low]

On 03/06/2005, at 11:21 PM, Paul Russell wrote:

> On 3 Jun 2005, at 08:01, Glen Low wrote:
>
>
>> Hi All
>>
>> I've got vector code (macstl -- what else!) that needs to compile  
>> correctly on:
>>
>> 1.    Apple gcc 3.3 with -faltivec
>> 2.    Mainline gcc 3.4 with -maltivec
>> 3.    Apple gcc 4.0 with -faltivec -maltivec
>> 4.    Mainline gcc 4.0 with -maltivec
>>
>> On 1 and 3, the Altivec intrinsics are builtin and vector, bool  
>> and pixel are automatically contextualized keywords. On 2 and 4,  
>> the Altivec intrinsics need to be #include <altivec.h> and vector,  
>> bool and pixel are not keywords (but they are, ugh, defined  
>> through macros in <altivec.h>).
>>
>> What combination of #ifdef macros and secret handshakes will allow  
>> me to navigate this wasteland??
>>
>>
>>
>
> One possible empirical solution is to invoke:
>
> gcc -faltivec -dM -E - < /dev/null
>
> or
>
> gcc -maltivec -dM -E - < /dev/null
>
> on each of the compilers and then sort and diff the 4 out output  
> files against each other.
>
> Paul


That would be good except I don't possess mainline gcc 3.4. Can  
someone confirm that

__APPLE_CC__

is undefined or 1 for mainline gcc, and >1 for Apple gcc?




Cheers, Glen Low


---
pixelglow software | simply brilliant stuff
www.pixelglow.com
aim: pixglen

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.pixelglow.com/lists/archive/macstl-dev/attachments/20050604/dd94b98f/attachment.html