(j3.2006) Order of application of elemental procedures

Aleksandar Donev donev1
Tue Mar 17 16:54:47 EDT 2009


Hi,

> I don't think it matters, 
Exactly.

> but "in array element order"  
> might be construed to prohibit parallelizing references to pure ones,
> even though the restrictions on pure ones imply that the order can't
> matter.
No it doesn't. If you can't tell the code was optimized, all is well. 
The purity guarantees that the ordering does not matter so the compiler 
can parallelize just as it could in F03.

Aleks



More information about the J3 mailing list