(j3.2006) (SC22WG5.4003) [ukfortran] New summary of coarrays

N.M. Maclaren nmm1
Thu Jun 11 04:41:20 EDT 2009


>Once again I reiterate my "watch what you are replying to" : J3 list 
>readers please use "Reply All" or manually reinsert the WG5 list when 
>replying to a WG5 message.

Yes, please!  I don't get J3 messages, for a start.

>No, as I understand it existing hardware and physical constraints 
>encourages some vendors not to have global consistency; *and* we allow 
>them not to have global consistency.

Yes.  Bill is right that the chances of decent documentation are not good,
unfortunately.  I was hoping to encourage the vendors ....

My reasoning for wanting to leave this open is that we really, but REALLY,
can't predict where this area is going.  There are two aspects:

    1) Will Fortran coarrays be implement much on commodity clusters, what
semantics will they deliver, how many programs will be written for them,
and what semantics will they rely on?  We haven't even selected the jury
to answer that one yet!

    2) What memory model will the multi-core CPUs of 2019 (128 is estimated)
use?  Few (none?) of the current ones deliver sequential consistency, and
some experts believe that even causal consistency will have to be abandoned
to get decent efficiency on those systems.

And that is why I believe that tying down the memory model should be a job
for Fortran 2018.  The current wording is clean (and clear to an expert),
and does not prejudge a future decision.

>As for "confusing to an ordinary user", not nearly as confusing as 
>getting wildly different answers or program hangs/crashes from the same 
>data on the same machine on a Thursday.

Precisely - been there, seen that, spent hours trying to explain it to the
users :-(  That is why I feel that there should be a warning in all relevant
documents "Here Be Dragons", but I don't care how it is phrased.


Regards,
Nick Maclaren,
University of Cambridge Computing Service,
New Museums Site, Pembroke Street, Cambridge CB2 3QH, England.
Email:  nmm1 at cam.ac.uk
Tel.:  +44 1223 334761    Fax:  +44 1223 334679




More information about the J3 mailing list