[mpiwg-rma] [EXTERNAL] Re: same_op_no_op and SHMEM
Barrett, Brian W
bwbarre at sandia.gov
Thu Oct 24 14:45:01 CDT 2013
On 10/24/13 1:24 PM, "Jeff Hammond" <jeff.science at gmail.com> wrote:
>> I would have no objection to adding yet another info key. I think if we
>> keep at this for another year, we can make sure we have the longest
>> pre-defined info key in the spec.
>
>Or we can focus on the values. Instead of "same_op_no_op_replace =
>true" we can have "op_list = no_op,replace,sum" and let the
>implementation instantiate same_op_no_op_replace on its own.
The advantage here would be that if the bad thing is FP divide, I can list
no_op,sum,replace,max,min and still have goodness. The disadvantage would
be all the test to list the options. So maybe long true/false keys have
some advantage.
>> I admit to having very little medium term memory; which is the
>> type-homogeneity suggestion?
>
>Medium-term meaning in the last 2 hours? :-)
Fine, no memory at all.
>It addresses the lack of type-specification in MPI window creating.
>If I can say at window allocation time that I only want to use
>MPI_INT, then maybe the implementation can avoid a software
>implementation that might be required if I was to use MPI_FLOAT, for
>example. In my experience, NICs are more likely to support
>fixed-point arithmetic in hardware than floating-point.
As an info key? I think that would be helpful for exactly the reason you
said. If I know the user isn't going to use long double complex data
types, that makes life a bit easier :).
Brian
--
Brian W. Barrett
Scalable System Software Group
Sandia National Laboratories
More information about the mpiwg-rma
mailing list