Poinot to try Version 3 fix for HDF-5 indexing problem, and
make sure it works for his applications.
Not done yet, but there was also some discussion on this
point. The index ordering for HDF-5 is now fixed in
Version 3.0, but this is for people who use the MLL. The
question arises whether one of Marc's HDF-5 files, if
given to someone else who uses MLL to read it, will be
successful or not. I.e., if a user like Marc does not
use MLL, it is up to them to order the indices correctly.
If they do not, then the file will not transfer correctly
to other users who use MLL. Hauser and Poinot will
work together to resolve the issue of the HDF-5 indexing
problem, both to make sure the fix in V3 works for Poinot
and to check consistency of files generated without
MLL.
Rumsey to add disclaimer about HDF-5 incorrect index ordering
in Version 2.x to website.
Poinot to send Wedan his coding and other information related
to HDF-5 compression.
Done.
Wedan to incorporate compression capability into Version 3
MLL.
Not done yet. Action item carries.
Wedan and Hauser to look into improving documentation and
build instructions for V3.0.
Not done yet. This action item will be combined with (h)
and (i) to make a new action item associated with the
upcoming release of V3.0. Hauser and Wedan to improve
documentation and build instructions for V3.0, remove
instructions regarding XML output from configure script,
make HDF-5 part of V3.0's tar file distribution, and
generally upgrade the build capability prior to release
of V3.0. Tentative plans to release V3.0 in January 2009.
This was also added to HDF-5 "to-do"
list.
Rumsey to email committee and ask for vote on two existing
proposals and collect responses.
Done. Allmaras' General SIDS Improvement extension is
ready to go, but Poinot's Rigid Motion Improvement
needs some specific examples added, and a renewed
consensus. Rumsey also brought up that we need to start
to tackle some of the accepted proposals "awaiting
implementation." Rumsey to move Allmaras' proposal to
"Awaiting implementation" column. Poinot to work with
Power and others to bring Rigid Motion Improvement
proposal to completion, including specific examples.
Rumsey to compile list of potential programmers who might
help to implement some of the outstanding proposals.
Hauser to remove instructions regarding XML output from
configure script in V3.0.
Not done yet. Combined to make a new action item.
Hauser to make HDF-5 part of V3.0's tar file distribution, and
set up makefile appropriately.
Not done yet. Combined to make a new action item.
Hauser will add HDF-5 parallel code to SourceForge and will
let users know when it is available.
Not done yet. Action item carries.
Towne to modify NGON+NFACE documentation for V3 to
include Feldman's suggestions, and to remove NGON +
parent data as an option.
Paul, Bussoletti, and Allmaras to continue to look into
finding manpower development (particularly for MLL
implementation) resources at Boeing.
Not done yet. Action item carries. There was also
discussion about NSF Industry Centers, which require
a collaboration of several groups each contributing
about 25K per year. Feldman also brought up the HPC
Consortium, which is trying to tie multi-physics codes
together... there is the question whether CGNS has
something to offer to this group and possibly could join
or contribute. Allmaras pointed out that if CGNS could
gain the consortium's support, that it might help CGNS to
gain more support from Boeing. Hauser to contact members
of the HPC Consortium to explore possibility of CGNS
contributing to their effort.
Duque and Hauser to continue to develop proposal for handling
sprays of unconnected points.
Not done yet. Action item carries.
Duque and Hauser to continue to pursue NSF funding and look
into idea of a fundraising collaborative.
Hauser has applied for NSF funding, including parallel
development for CGNS, but there is probably only a very
small chance that it will go through.
Poinot to forward a draft proposal on CGNS activities for
EUCASS to Hauser & Rumsey that will eventually go to the
EUCASS organizers.
Done. Poinot has forwarded the request to the EUCASS
organizers.
Hauser to attend an upcoming HDF-5 meeting in Colorado, and
find out more about what it means to become a non-profit like
HDF-5 and how it would affect things.
Done. Hauser gave a talk at the HDF-5 meeting and
spoke to members of that group. He found out that they
may be interested in attending the EUCASS meeting in
July 2009 and contributing or participating in the
CGNS activities there. Regarding becoming non-profit,
HDF-5 was helped considerably by the fact that it has
a long-term investment by NASA's earth observatory
mission. Another thing learned: In HDF-5, if there
are a very large number of zones, there are still
performance issues, but HDF-5 has a method to "fix" this,
using a different interface to HDF-5, which improves
performance. Hauser to send information on interface
"fix" to improve performance of HDF-5 when there are many
zones to Wedan, Poinot, and anyone else interested.
Wedan to investigate issue of MLL deleting the entire node
even though its new dimensions are the same.
Not done yet. Action item carries.
Hauser will take control of cgns.org domain name and
will consider possibility of moving website to Utah State
site.
Not successful in obtaining an invoice from Ansys
yet. Action item carries.
Hauser to contact Hunten about ISO/STEP and ask the best way
for us to comment.
Not done. Decided to delete this action item and instead
keep this as an open item.
Wedan to look into issue brought up by Miller (10/19/07) to
write info one element at a time to save memory.
Not done yet. Action item carries.
Wedan to implement patch sent to him by Bussoletti to fix
excessive memory usage by MLL.
Not done yet. Action item carries.
Rumsey to email Hauser information on how to reserve a room
for AIAA conference.
Done.
Open items (purpose of these is to not lose track of long-term
issues that come up)
Need for official certification process (test for compliance).
Develop method to write info one element at a time to save
memory via something like cg_section_partialwrite
(Miller 10/19/07 CGNSTalk)
Address issue of possible need for "long" instead of "int" to
avoid signed 2 Gig limit (Miller 10/19/07 CGNSTalk).
Problem with 64-bit cg_goto_f (Gibeling 08/29/08
CGNSTalk)
Hauser plans to have face-to-face meeting at the AIAA meeting in
Orlando in January 2009. Tentatively set for Wednesday evening of
the conference, January 7 at 7:00 PM. Hauser will ask for telecon
capability in case anyone is interested in calling in.
Meeting was adjourned at 12:30 PM Eastern. Next Telecon date will
be set at the AIAA meeting in Orlando.
Summary of action items:
Hauser and Poinot will work together to resolve the issue of
the HDF-5 indexing problem, both to make sure the fix in V3
works for Poinot and to check consistency of files generated
without MLL.
Wedan to incorporate compression capability into Version 3
MLL.
Hauser and Wedan to improve documentation and build
instructions for V3.0, remove instructions regarding XML
output from configure script, make HDF-5 part of V3.0's tar
file distribution, and generally upgrade the build capability
prior to release of V3.0. Tentative plans to release V3.0 in
January 2009.
Rumsey to move Allmaras' proposal to "Awaiting implementation"
column.
Poinot to work with Power and others to bring Rigid Motion
Improvement proposal to completion, including specific
examples.
Rumsey to compile list of potential programmers who might help
to implement some of the outstanding proposals.
Hauser will add HDF-5 parallel code to SourceForge and will
let users know when it is available.
Paul, Bussoletti, and Allmaras to continue to look into
finding manpower development (particularly for MLL
implementation) resources at Boeing.
Hauser to contact members of the HPC Consortium to explore
possibility of CGNS contributing to their effort.
Duque and Hauser to continue to develop proposal for handling
sprays of unconnected points.
Hauser to send information on interface "fix" to improve
performance of HDF-5 when there are many zones to Wedan,
Poinot, and anyone else interested.
Wedan to investigate issue of MLL deleting the entire node
even though its new dimensions are the same.
Hauser will take control of cgns.org domain name and
will consider possibility of moving website to Utah State
site.
Wedan to look into issue brought up by Miller (10/19/07) to
write info one element at a time to save memory.
Wedan to implement patch sent to him by Bussoletti to fix
excessive memory usage by MLL.
Attendees
Steve Allmaras
Boeing Commercial
Mark Anderson
Concepts NREC
Steve Feldman
CD-ADAPCO
Thomas Hauser
Utah State
Chris Rumsey
NASA Langley
Tentative to-do list in association with HDF-5 switchover
Complete and test parallel implementation
under development
Complete and test API capability to automatically detect and switch
between ADF and HDF-5 (transparent to user?)
done in V3.0
Assess/minimize impact on software vendors using CGNS
under development (success with automatic detection transparent
to user will minimize impact)
Make changes recommended by HDF-5 to improve usability with HDF-5
(e.g., character strings as opposed to character arrays)
done
Assess compression capability of HDF-5
done by Poinot - capability will be added to V3 MLL
CGNS configure scripts will need to be modified to check for
availability of appropriate HDF-5 libraries.
done in V3.0
Possibly add flag-based options when opening CGNS files. For
example: "follow links" vs. "don't follow links"; "use ADF" vs. "use
HDF-5"; "translate file automatically" vs. "leave the file as-is";
"compress" vs. "don't compress".
Look into eliminating need for ID's from MLL
Resolve "link-of-links" problem: will require HDF-5 fix by NCSA
under development (1.8.0 beta)
done in V3.0 in combination with HDF-5 V1.8
Are any changes needed in SIDS-to-HDF5 document? Or is it
up-to-date?