(j3.2006) (SC22WG5.3998) [ukfortran] New summary of coarrays

[Bill Long]

Malcolm Cohen wrote:
> Hi Bill and Craig,
> 
> Craig Dedo, and then Bill Long, wrote to J3 replying to a WG5 message...
> 
> 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.
> 
> I am copying this to WG5 just so I can stick my oar in on the substance.
> 
> Craig Dedo wrote replying to Nick (I think):
>> Everyone:
>> 	I believe that a better wording for the last sentence would be:
>>
>> 	The exact details are left processor dependent.  Therefore,
>> programmers should read their processor documentation before using atomic
>> subroutines.
>>
>> 	I believe that "processor documentation" is the recommended term.
>> Also, breaking it up into two sentences makes it read better.
>>   
> 
> And Bill Long replied:
>> Both versions presume something that seems very unlikely - that vendors
>> discuss such implementation and performance details in their
>> documentation. I think the whole addition should be dropped. At best it
>> is confusing to an ordinary user; at worst is encourages vendors to not
>> bother doing a good job.
>>   
> 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.

I didn't see any relation to global consistency here. The context was 
user written synchronization schemes. The issue is whether the atomic 
operations take near-infinite time to execute.



> 
> 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.

The same machine should have pretty consistent performance for a single 
atomic memory operation.


Cheers,
Bill



> 
> Cheers,

-- 
Bill Long                                   longb at cray.com
Fortran Technical Support    &              voice: 651-605-9024
Bioinformatics Software Development         fax:   651-605-9142
Cray Inc., 1340 Mendota Heights Rd., Mendota Heights, MN, 55120