(j3.2006) (SC22WG5.5240) Interesting article

Bill Long longb
Fri May 9 17:25:44 EDT 2014


On May 8, 2014, at 6:57 PM, Van Snyder <van.snyder at jpl.nasa.gov> wrote:

> On Thu, 2014-05-08 at 16:42 -0700, Damian Rouson wrote:
>> 
>> On May 8, 2014, at 11:28 AM, Van Snyder <Van.Snyder at jpl.nasa.gov>
>> wrote:
>> 
>>> On Thu, 2014-05-08 at 10:35 +0100, John Reid wrote:
>>> 
>>> If Julia really wants to get past half as fast as Fortran, maybe the
>>> target ought to be Fortran instead of LLVM.
>> 
>> These aren?t mutually exclusive.  There are Fortran front ends that
>> use LLVM as the back end, e.g., DragonEgg: http://dragonegg.llvm.org.
> 
> Do such compilers produce code with performance comparable to
> purpose-built compilers, such as Intel's, or is the code half as fast?

It depends (of course).  Some commercial compilers use part of LLVM and result in code with very good performance.  But if the question is about ?hobby? compiler projects that use all of LLVM, the realistic question is how they compare to gfortran (i.e the default free compiler).  If they can?t keep up with gfortran, then the hobby seems of questionable value.  Unless the purpose is to train students in compiler development, for example. 


> 
> If the latter, maybe Julia ought to target Fortran, and compile the
> resulting code with a purpose-built Fortran compiler.
> 

It would be better if the *programmer* targeted Fortran in the first place.  Eliminate the middle man.

Cheers,
Bill




>> Damian
> 
> 
> _______________________________________________
> J3 mailing list
> J3 at mailman.j3-fortran.org
> http://mailman.j3-fortran.org/mailman/listinfo/j3

Bill Long                                                                       longb at cray.com
Fortran Technical Suport  &                                  voice:  651-605-9024
Bioinformatics Software Development                     fax:  651-605-9142
Cray Inc./ Cray Plaza, Suite 210/ 380 Jackson St./ St. Paul, MN 55101





More information about the J3 mailing list