[Pythonmac-SIG] unixy python and applescript

[Bob Ippolito]

On Tuesday, Sep 2, 2003, at 18:47 America/New_York, Rob Managan wrote:

>> On Thursday, Aug 28, 2003, at 14:24 America/New_York, Russell Finn  
>> wrote:
>>
>>> On Wednesday, August 27, 2003, at 06:41  PM, Bob Ippolito wrote:
>>>> I'll probably get around to compiling Numpy+ATLAS, Scipy, etc. for  
>>>> MacPython sometime.  I'd have done it by now, but the damn thing  
>>>> needs a Fortran compiler and I haven't come across one that's  
>>>> reasonably easy to install/use with the gcc 3.3 toolchain.  If you  
>>>> find one, let me know, and I'll see about doing it.
>>>
>>> Just the thing you need, courtesy of IBM:   
>>> <http://www14.software.ibm.com/webapp/download/ 
>>> search.jsp?go=y&rs=vacpp3>
>>>
>>> Disclaimer: It's technically beta, and I haven't actually used it.   
>>> But it's supposed to work with (nay, requires) gcc 3.3.
>>
>> Just FYI, the IBM compiler seems to install and compile trivial  
>> software (PRINT *, "Hello World!") correctly on my pbg4 1ghz even  
>> though it lists a PowerMac G5 as a requirement.  I'll let you all  
>> know how Numpy+ATLAS and SciPy go when I get the chance to try those  
>> out.
>>
>
> i did try compiling ATLAS and the settings I gave it resultedin lower  
> MFLOPS numbers in the SUMMARY.LOG file than the default gcc  
> installation!!

Well it seems that the OS X High Performance Computing page (  
http://gravity.psu.edu/~khanna/hpc.html ) has been updated to include  
an experimental version of g77 for gcc 3.3, so I'll try both g77 and  
xlf90 with a couple different flags (time willing) and see which  
benchmarks better on my g4.  I'd be willing to bet that IBM's compilers  
work MUCH better with a ppc970 target than GCC does.. I doubt IBM was  
shooting for much more than compatibility with Motorola processors,  
especially for this initial beta.

-bob