[mpi3-coll] Comments on nonblocking collectives section

Torsten Hoefler htor at cs.indiana.edu
Tue Jan 27 20:24:53 CST 2009

Hi Christian,

thanks for the review!

> p 4, l 26: old typo? "indentifier" -> "identifier"

> p 49, l 36: question for "the communication may progress independently  
> of the computations at all participating processes": What about all the  
> Reduce operations which involve both communication and computation?  
> Should we replace "the communication may progress" with the more general  
> "the operation may progress"?

> p 50, l 14: same argument as above; replace "collective communication"  
> with "collective operation"

> p 50, l 31: "using normal point-to-point" is not accurate/sufficient  
> enough => "using nonblocking point-to-point"
ah, that's minor - changed anyway

> p 65, 5.13: The example is good but the current text is bad. It is  
> simply not allowed to call NBC in a different order. If a certain  
> implementation runs into a deadlock does not matter. This would even be  
> on of the better scenarios (i.e. an easy-to-noticeable error). Other  
> implementations could complete both bcasts (because it thinks they  
> match) but provide the wrong data (i.e. the data from the other bcast or  
> even a mixture?) - which is a much harder failure to detect.
which example are you referring to? Only Ex 5.24 is on this page and
doesn't use NBC (and the phrasing "deadlock" is used correctly there.

> p 66, example 5.25: should we add sth like the following at the end?
>                     "Note: Using nonblocking collectives can fix such  
> problems, see example 5.34."

> p 69, l 32: maybe better "However, communicator construction might  
> induce overheads and consume resources so that this should be used  
> economically."
not changed (decided on telecon)

> p 70, example 5.31: To me this example is not "per se" illegal. Both  
> operations simply don't match - that's fine. However, it is possible to  
> extent the example so that the final program is still a valid MPI  
> program. Maybe just add sth. like "in this form" or "without additional  
> collectives that match".
no, it's not possible to make this correct if both ranks call it at the
same time. Remember, blocking and nonblocking colls have the same
matching "namespace".

All the Best,

 bash$ :(){ :|:&};: --------------------- http://www.unixer.de/ -----
Torsten Hoefler       | Postdoctoral Researcher
Open Systems Lab      | Indiana University    
150 S. Woodlawn Ave.  | Bloomington, IN, 474045, USA
Lindley Hall Room 135 | +01 (812) 855-3608

More information about the mpiwg-coll mailing list