(j3.2006) Heterogeneous parallelism

Dan Nagle dannagle
Wed Aug 22 16:36:02 EDT 2007


Aleksandar Donev wrote:
> Hi,
> Craig and Lawrie have brought up homogeneous versus heterogeneous parallelism 
> and how it affects co-arrays. I am trying to understand exactly what we are 
> talking about and what they want that is not there now.

This whole argument is entirely spurious, as was explained at 181.

The Fortran standard simply doesn't explain what the processor
actually *is*, so the entire debate about what the standard says
or not about homogeneous environments versus heterogeneous ones
is off-topic.

You can't debate green oobleck versus blue oobleck
if you can't discuss oobleck.

The vendor has the option, under the existing "size and complexity"
rule, to limit or not the environments wherein programs compiled
by the vendor's compiler will execute.

As a further off-topic expedition, the efficiency or not
of homogeneous environments versus heterogeneous environments
should not be compared against each other.  The question becomes,
will co-arrays be more or less efficient than any other message passing
scheme *on the same* hardware.

Sorry, Larwie and David, but a pile of red herrings is still
red herrings.

<snip the rest>

And I have yet to hear a single reason withstanding scrutiny
why co-arrays should be outside the main standard.  Endit.


Dan Nagle
Purple Sage Computing Solutions, Inc.

More information about the J3 mailing list