Search found 73 matches

by fewl9012
Wed Oct 09, 2019 3:55 pm
Forum: Using OpenMP
Topic: compunity: dead or alive
Replies: 3
Views: 1850

Re: compunity: dead or alive

If the purpose of cOMPunity has been superceded by the contributor membership, and there is admittedly no activity in the cOMPunity organization, perhaps it's time to formally close it down. It's still listed as a member, which according to MarkB's reply, is misleading at best. If a business shuts d...
by fewl9012
Sat Aug 31, 2019 5:46 pm
Forum: Using OpenMP
Topic: compunity: dead or alive
Replies: 3
Views: 1850

Re: compunity: dead or alive

Today I see the compunity.org website is at least resolving to something, but it's some old information. It says "The cOMPunity website provide up-to-date information" but the newest information on the site is from 2006. I still wonder if cOMPunity is really alive. Is Dr. Chapman still the president...
by fewl9012
Wed Jul 24, 2019 10:49 am
Forum: Using OpenMP
Topic: compunity: dead or alive
Replies: 3
Views: 1850

compunity: dead or alive

The compunity.org address has expired [at least as of July 24 2019]. Is cOMPunity still active? I see it is still listed as an OpenMP member. Does it have any presence, and does it add any value? There is a compunity twitter account, and it's had precisely one post, in 2008. For some reason, zoominf...
by fewl9012
Wed Sep 26, 2018 8:55 am
Forum: TR7 - OpenMP 5.0 Draft Discussion
Topic: OpenMP Context Definitions
Replies: 0
Views: 7906

OpenMP Context Definitions

Page 64, lines 12 and 21, refer to the OpenMP Context Definitions document which is available on http://www.openmp.org. Is this document available?
by fewl9012
Wed Aug 01, 2018 9:38 am
Forum: TR7 - OpenMP 5.0 Draft Discussion
Topic: Program order
Replies: 0
Views: 2369

Program order

The comment for the glossary item for program order says "For C11 and C++11, program order corresponds to the sequenced before relation..." In this sentence, "program order" is italicized. "sequenced before" should be similarly italicized or quoted.
by fewl9012
Wed Aug 01, 2018 9:35 am
Forum: TR7 - OpenMP 5.0 Draft Discussion
Topic: Can a device have zero processors
Replies: 0
Views: 2301

Can a device have zero processors

In the Glossary for device, "A device could have one or more processors." What is a device with zero processors? Can an OpenMP thread execute on anything other than a processor?
by fewl9012
Thu Sep 10, 2015 6:48 pm
Forum: Conferences and Events
Topic: Please proof read your IWOMP program
Replies: 0
Views: 6352

Please proof read your IWOMP program

You misspelled the name of one of your keynote speakers both on the program web page and on the program PDF. The keynote talk title for John Mellor-Crummey's talk (who is not Crummy) is also missing from the program PDF. A couple other misspelled words: Application-level Enery Awareness, I don't kno...
by fewl9012
Mon Aug 31, 2015 2:38 pm
Forum: OpenMP 4.1 DRAFT - Public Comments
Topic: omp_in_target?
Replies: 2
Views: 5127

Re: omp_in_target?

What if the 'target' region is executing on the initial device? Is that behavior well-defined? Is there a concept of an inactive target region, as there is with an inactive parallel region? omp_in_parallel is well defined to distinguish between in an active parallel region vs. only in inactive paral...
by fewl9012
Mon Aug 31, 2015 2:35 pm
Forum: OpenMP 4.1 DRAFT - Public Comments
Topic: elemental functions
Replies: 4
Views: 23524

Re: elemental functions

Clearly you are not a Fortran programmer. In Fortran, the !$omp declare target would appear within the subprogram to which it applies. Fortran, unlike C/C++, doesn't have the concept of a file or file scope, so there is no way to say 'declare target' from here in the file to some 'end declare target...
by fewl9012
Fri Aug 14, 2015 6:03 pm
Forum: OpenMP 4.1 DRAFT - Public Comments
Topic: target construct: If an array section...
Replies: 0
Views: 5174

target construct: If an array section...

page 98 line 7, C/C++ "If an array section is a list item in a map clause, and the array section is derived from a variable for which the type is pointer, reference to array or reference to pointer then the data-sharing attribute for that variable in the construct is private." I think would be bette...