Solution to MacOSX on Intel?

[Ben Abbott]

Jack, I did some bench marking using LA intensive code last year and  
found g95 and gfortran to very similar in speed ... Gfortran was a bit  
faster.

I had max'd out the optimization for each.

However both f2c and g77 were about 3x slower.

How did you do your comparison?

Ben

P.s. I only ask in case I am missing out on an opportunity to  
accelerate my programs.


On Dec 7, 2007, at 11:24 AM, Jack Howarth <howarth at bromo.msbb.uc.edu>  
wrote:

> John,
>   Using Apple's gcc/g++ compilers and gfortran from FSF gcc
> 4.2.2 on powerpc-apple-darwin9, I get 8 failures in the
> make check of octave 2.9.18. If I use only the FSF gcc
> 4.2.2 compilers on powerpc-apple-darwin9, I get a bus
> error in test_sparse.m. This bus error doesn't happen with
> the gcc 4.3 snapshot on powerpc-apple-darwin9 so it must
> be a compiler bug in 4.2.2. On i386-apple-darwin9, the
> Apple gcc/g++ compilers cause a bus error in the
> ov_fnct_handle.cc test case during make check. Using
> only the FSF gcc 4.2.2 compilers on i386-apple-darwin9
> eliminates that and yields 7 failures in the make check.
>   How many failures do you get on the linux with gcc 4.2.2
> for octave? I would rather stick with the gfortran compiler
> because it generates code that is significantly faster than
> g95.
>                 Jack
>
> On Fri, Dec 07, 2007 at 10:51:49AM -0500, John W. Eaton wrote:
>> On  7-Dec-2007, Ben Abbott wrote:
>>
>> | > In any event, I've successfully built octave using g95 on MacOSX
>> | > 10.5 on Intel. The details are below.
>>
>> Weren't people complaining that failures were happening with  
>> gfortran,
>> not g95?
>>
>> jwe
> _______________________________________________
> Bug-octave mailing list
> Bug-octave at octave.org
> https://www.cae.wisc.edu/mailman/listinfo/bug-octave