[Mpi-21] AUTHORS: new consistent version and write-token to you

Adam Moody moody20 at [hidden]
Wed Apr 16 20:32:59 CDT 2008 


Hi Rajeev,
Thanks for all of your feedback.  I've made every change but two, which 
I think we should delay to a decision for the Forum (or at least wait to 
get more votes from reviewers).

Rolf,
Can you add the following items to some ballot to be considered by the 
Forum?

* pg 134, ln 32-33: What does matching arguments mean? Does it mean same
arguments? That's not true for the root argument of rooted collectives on
the group containing the root. Maybe just delete the sentence.

* pg 163, Sec 5.9.3: Delete the sentence "This is an extension to MPI-1.2,
..." since we are all one now.

I've incorporated the rest of the edits Rajeev suggested and committed 
the new chapter.
-Adam

Rajeev Thakur wrote:

>Adam,
>     You have done a wonderful job of fixing the merged collectives chapter,
>which was a nontrivial task (and hence I didn't volunteer for it :-)). I
>only have some minor comments listed below.
>
>Rajeev
>
>
>* pg 133, ln 27-28: change to the plural -- "patterns of" and "do not fit"
>  
>
Done.

>* pg 134, ln 32-33: What does matching arguments mean? Does it mean same
>arguments? That's not true for the root argument of rooted collectives on
>the group containing the root. Maybe just delete the sentence.
>  
>
Yes, I agree, but it's not clear to me how to clarify this point 
either.  There's probably a clean way to explain the intent here, but I 
don't want to make that decision.  I'll leave this in to get more input 
from the Forum.

>* pg 136, ln 22-23: move "for intracommunicators" to the beginning of the
>sentence
>  
>
Done.

>* pg 140, ln 28: change "are using" to "use"
>  
>
Done.

>* pg 149, ln 45: change "are using" to "use"
>  
>
Done.

>* pg 154, ln 33: change "are using" to "use"
>  
>
Done.

>* pg 162, ln 10: change "are using" to "use"
>  
>
Done.

>* pg 163, Sec 5.9.3: Delete the sentence "This is an extension to MPI-1.2,
>..." since we are all one now. Merge the next paragraph on heterogeneous
>into this one, and fix spelling of "charater" to character.
>  
>
I fixed the "charater" mis-spelling and merged the paragraphs.  I left 
the statement about MPI-1.2 for now, although I moved it so that, after 
merging the other paragraph, the resulting paragraph reads a little more 
smoothly.  This statement seems to highlight an important, but subtle 
change.  I want to get feedback from the Forum on this item.

>* pg 165, ln 10: change "are using" to "use"
>  
>
Done.

>* pg 167, ln 38-39: Question - Do we want to also list MPI_Accumulate with a
>forward reference to it?
>
>* pg 170, ln 3: change "is using" to "uses"
>  
>
Done.

>* pg 171, ln 38: should be "In such a case" or better "In this case"
>  
>
Done.

>* pg 171, ln 43: change "is using" to "uses" and "an intracommunicators" to
>the singular
>  
>
Done.

>* pg 174, ln 11: change "illegal" to "invalid"
>  
>
Done.

>* pg 174, ln 12-20: Delete the entire rationale. It was included in MPI-1 to
>justify not having an exclusive scan. Doesn't apply any more. You deleted
>only the last sentence, which was explicit, but the whole rationale should
>go.
>  
>
Done.

>* pg 175, ln 13-20: By the same token, delete the entire rationale here. It
>was used to justify adding an exclusive scan to MPI-2. Even the "no in-place
>version" is covered above in lin 5. Maybe you can change that line 5 to
>include the justification as follows: "No in-place version is supported
>because it is not clear what this means for the process with rank zero".
>Note that it should be "process with rank zero" rather than "process for
>rank zero" as in the current text.
>  
>
I kept the rationale, because I think it provides some useful 
information here, but I changed it to the following:

The exclusive scan is more general than the inclusive scan.  Any 
inclusive scan operation can be achieved by using the exclusive scan and 
then locally combining the local contribution.  Note that for 
non-invertable operations such as MPI_MAX, the exclusive scan cannot be 
computed with the inclusive scan.  No in-place version is specified for 
MPI_EXSCAN because it is not clear what this means for the process with 
rank zero.

>* pg 175, ln 25: change "is using" to "uses"
>  
>
Done.

>Rest looks great!
>
>
>
> 
>
>  
>
>>-----Original Message-----
>>From: mpi-21-bounces_at_[hidden] 
>>[mailto:mpi-21-bounces_at_[hidden]] On Behalf Of Rolf 
>>Rabenseifner
>>Sent: Saturday, April 12, 2008 10:15 AM
>>To: MPI 2.1 Mailing List
>>Subject: Re: [Mpi-21] AUTHORS: new consistent version and 
>>write-token to you
>>
>>Hi Adam,
>>
>>this is really a good merge that you have done.
>>Good ideas and very carefully implemented.
>>
>>The reviewers can see (nearly) all of your modifications in red.
>>I expect that your reviewersand the Forum will enjoy your 
>>modifications.
>>
>>I have restored two minor mods you have done:
>>(Line numbers refer to the new file you'll get with svn update)
>>
>>coll.tex, line 412 - I restored the "intracommunicator"
>>  because the categories describe the behavior only on
>>  intracommunicators, not on all, e.g., ALLGATHER has
>>  one result on intra and two on inter and the wording is
>>  "the result".
>>
>>coll.tex, line 491 - restored the word "also".
>>  These routines are valid for intra and inter.
>>  Writing "apply to inter" may be interpreted that they are
>>  only valid for inter. 
>>  Writing "also apply to inter" should solve this problem. 
>>  I also colored this sentence red.
>>
>>I hop, this isokay with you.
>>
>>Best regards
>>Rolf
>>
>>
>>On Fri, 11 Apr 2008 15:22:10 -0700
>> Adam Moody <moody20_at_[hidden]> wrote:
>>    
>>
>>>Hi Rolf,
>>>I have checked in my changes for the Collectives chapter.  
>>>      
>>>
>>It's ready to 
>>    
>>
>>>be reviewed by the collective chapter reviewers.
>>>-Adam
>>>
>>>Rolf Rabenseifner wrote:
>>>
>>>      
>>>
>>>>>>write-token again to the chapter authors    <<
>>>>>>- but only until Saturday 8:00 European GMT <<
>>>>>>     
>>>>>>
>>>>>>            
>>>>>>
>>>>Hi all authors and reviewers, 
>>>>
>>>>Please read this mail carfully, if you want to make any further 
>>>>modification of the text.
>>>>
>>>>You must now call 
>>>> svn update 
>>>>to get the newest release!!   
>>>>We have a again a consistent version. Nearly all of 
>>>>
>>>> 
>>>>        
>>>>
>>http://www.hlrs.de/mpi/mpi21/doc/MPI-2.1draft-2008-02-23-review.txt
>>    
>>
>>>>is now implemented in the tex-source.
>>>>Only a few items are still open, see statistics at the beginning.
>>>>
>>>>You can view the current version (it is svn revision 92) at
>>>>
>>>> http://www.hlrs.de/mpi/mpi21/doc/mpi-report.pdf
>>>>
>>>>It is only today and Friday remaining for finishing MPI-2.1.
>>>>I give back the write-token to the authors.
>>>>
>>>>On Saturday morning European time, I take the write-token
>>>>automatically on Saturday morning to build final pdf for the 
>>>>reviews.
>>>>
>>>>As I can see, many chapter authors made fine cleaning to the text.
>>>>All larger changes should be defered to MPI-2.2!
>>>>Especially the merge of MPI Environment (inquiry.tex), 
>>>>        
>>>>
>>Misc (misc-2.tex)
>>    
>>
>>>>and External Interfaces (ei-2.tex) 
>>>>to one chapter "MPI Environment and Miscellany" (with 45 pages) 
>>>>may be defered also to MPI-2.2.
>>>>
>>>>Up to now, I was able to check most changes and they look fine.
>>>>
>>>>Best regards
>>>>Rolf
>>>>
>>>>
>>>>
>>>>Dr. Rolf Rabenseifner . . . . . . . . . .. email 
>>>>        
>>>>
>>rabenseifner_at_[hidden]
>>    
>>
>>>>High Performance Computing Center (HLRS) . phone 
>>>>        
>>>>
>>++49(0)711/685-65530
>>    
>>
>>>>University of Stuttgart . . . . . . . . .. fax ++49(0)711 
>>>>        
>>>>
>>/ 685-65832
>>    
>>
>>>>Head of Dpmt Parallel Computing . . . 
>>>>        
>>>>
>>www.hlrs.de/people/rabenseifner
>>    
>>
>>>>Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)
>>>>_______________________________________________
>>>>mpi-21 mailing list
>>>>mpi-21_at_[hidden]
>>>>http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-21
>>>>
>>>> 
>>>>
>>>>        
>>>>
>>>_______________________________________________
>>>mpi-21 mailing list
>>>mpi-21_at_[hidden]
>>>http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-21
>>>      
>>>
>>
>>Dr. Rolf Rabenseifner . . . . . . . . . .. email rabenseifner_at_[hidden]
>>High Performance Computing Center (HLRS) . phone ++49(0)711/685-65530
>>University of Stuttgart . . . . . . . . .. fax ++49(0)711 / 685-65832
>>Head of Dpmt Parallel Computing . . . www.hlrs.de/people/rabenseifner
>>Nobelstr. 19, D-70550 Stuttgart, Germany . (Office: Allmandring 30)
>>_______________________________________________
>>mpi-21 mailing list
>>mpi-21_at_[hidden]
>>http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-21
>>
>>
>>    
>>
>
>_______________________________________________
>mpi-21 mailing list
>mpi-21_at_[hidden]
>http://lists.mpi-forum.org/mailman/listinfo.cgi/mpi-21
>
>  
>

More information about the Mpi-21 mailing list