News

PUMI_MESH (FMDB): FMDB Weekly #8 (Feb 21 - Feb 25)

Added by E. Seegyoung Seol over 6 years ago

[ Week 8 Activities ]

FMDB v1.2
  • (on-going) entity set/entity group integration
  • tagging improvement
    - (completed) byte type tagging for entity/entity set/part
    - (on-going) array/entity set tagging
  • (on-going) parallel FMDB.h development
    - constraint: single part per mesh
    - improving ghosting & related code (e.g. FMDB_Mesh_Verify, )
  • (completed) autotool packaging for Sandia release
  • (completed) User's Guide update: chap 7. compilation instructions
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - (completed) set/get adjacency table
    - See "Active Issues" for iMesh bug details
SCUtil/iBase
  • NONE
GRM/iRel
  • NONE

[ Planned Week 9 Activities ]

FMDB v1.2
  • entity set/entity group integration
  • array/entity set tagging
  • parallel FMDB.h development
  • packaging for ITAPS release (March 5)
  • User's Guide update: chap 6. programming instructions
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
SCUtil/iBase
  • NONE
GRM/iRel
  • NONE
[ Independent Study ]
  • NONE
[ Active Issues ]
  • vertex coordinate precision problem (reporter: Min&Kai)
  • mesh loading failure with process grouping (reporter: Fan)
  • iMesh compliance test: infinite loop with no-advancing entity iterator
  • FMDB/test/serial: mesh format portability (sms, vtk, ncdf conversion fails)
    . mesh file written from another mesh format cannot be loaded back
    . e.g. sms written from ncdf or vtk file cannot be loaded back, and vice versa
  • removed entity, entity set, part handle are not deleted from tagging storage
  • crash on partitioned mesh loading (+512 parts on B/G)
  • deleting multiple mesh instances fails (chronic bug)
  • improper memory management with tagging
    - simultaneous, multiple mesh instances give error due to single tag management scheme across multiple mesh objects
    - multiple mesh instances a in a row give error during tag data deletion (cause: Singleton tagging management class)

[ Pended Issues ]

  • FMDB
    - intelligent memory management & leak proof
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
    - (held) 2D NCDF mesh support
  • iMesh(P)
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCUtil/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Wednesday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

PUMI_MESH (FMDB): FMDB Weekly #7 (Feb 14 - Feb 18)

Added by E. Seegyoung Seol over 6 years ago

[ Week 7 Activities ]

FMDB v1.2
  • (on-going) entity set/entity group integration
  • tagging improvement
    - (completed) byte type tagging for entity
    - (on-going) byte type tagging for part & entity set
  • (on-going) parallel FMDB.h development
    - constraint: single part per mesh
    - improving 1-layer ghosting & related code (e.g. FMDB_Mesh_Verify, )
  • (completed) packaging for Sandia release (parallel)
  • (on-going) User's Guide update
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - (on-going) implementing with FMDB.h api calls
    - See "Active Issues" for iMesh bug details
SCUtil/iBase
  • NONE
GRM/iRel
  • NONE

[ Planned Week 8 Activities ]

FMDB v1.2
  • entity set/entity group integration
  • byte type tagging for part/entity set
  • parallel FMDB.h development
  • packaging for ITAPS release (Feb 25)
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
SCUtil/iBase
  • NONE
GRM/iRel
  • NONE
[ Independent Study ]
  • NONE
[ Active Issues ]
  • vertex coordinate precision problem (reporter: Min&Kai)
  • mesh loading failure with process grouping (reporter: Fan)
  • iMesh compliance test: infinite loop with no-advancing entity iterator
  • FMDB/test/iter: PartIter with type 4 (all types) returns double of # entities of all types
  • FMDB/test/serial: mesh format portability (sms, vtk, ncdf conversion fails)
    . mesh file written from another mesh format cannot be loaded back
    . e.g. sms written from ncdf or vtk file cannot be loaded back, and vice versa
  • removed entity, entity set, part handle are not deleted from tagging storage
  • crash on partitioned mesh loading (+512 parts on B/G)
  • deleting multiple mesh instances fails (chronic bug)
  • improper memory management with tagging
    - simultaneous, multiple mesh instances give error due to single tag management scheme across multiple mesh objects
    - multiple mesh instances a in a row give error during tag data deletion (cause: Singleton tagging management class)

[ Pended Issues ]

  • FMDB
    - intelligent memory management & leak proof
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
    - (held) 2D NCDF mesh support
  • iMesh(P)
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCUtil/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Wednesday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

PUMI_MESH (FMDB): FMDB Weekly #6 (Feb 07 - Feb 11)

Added by E. Seegyoung Seol over 6 years ago

[ Week 6 Activities ]

FMDB v1.2
  • (on-going) entity set/entity group integration
  • (on-going) tagging improvement
    - byte type tagging & generic get/set functions
  • (on-going) parallel FMDB.h development
    - constraint: single part per mesh
    - migration/global&local partitioning/ghosting
  • (on-going) packaging for Sandia release
  • (on-going) User's Guide update
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - (on-going) implementing with FMDB.h api calls
    - See "Active Issues" for iMesh bug details
SCUtil/iBase
  • NONE
GRM/iRel
  • NONE

[ Planned Week 7 Activities ]

FMDB v1.2
  • entity set/entity group integration
  • tagging improvement
  • parallel FMDB.h development
  • memory leak check with Valgrind
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
SCUtil/iBase
  • NONE
GRM/iRel
  • prototype development
[ Independent Study ]
  • NONE
[ Active Issues ]
  • PartIter with type 4 (all types) returns double of # entities of all types
  • mEntity::get (upward adjacency) crashes when no upward adjacency exists
  • mesh format portability (sms, vtk, ncdf conversion fails)
    . mesh file written from another mesh format cannot be loaded back
    . e.g. sms written from ncdf or vtk file cannot be loaded back, and vice versa
  • generic data tag fails
  • tag array function fails
  • crash on partitioned mesh loading (+512 parts on B/G)
  • deleting multiple mesh instances fails (chronic bug)
  • improper memory management with tagging
    - simultaneous, multiple mesh instances give error due to single tag management scheme across multiple mesh objects
    - multiple mesh instances a in a row give error during tag data deletion

[ Pended Issues ]

  • FMDB
    - garbage collection (automatic or user-triggered)
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
    - (held) 2D NCDF mesh support
  • iMesh(P)
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCUtil/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Wednesday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

PUMI_MESH (FMDB): FMDB Weekly #5 (Jan 31 - Feb 4)

Added by E. Seegyoung Seol over 6 years ago

[ Week 5 Activities ]

FMDB v1.2
  • (on-going) entity set/entity group integration
  • (on-going) tagging improvement
    - traceable and non-traceable tag creation
    - byte type tagging & generic get/set functions
    - entity set type tagging
  • (on-going) parallel FMDB.h development
    - constraint: single part per mesh
    - defining api for partitioning related functions (Min)
  • (on-going) packaging for Sandia release
  • (on-going) User's Guide update
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - (on-going) implementing with FMDB.h api calls
    - See "Active Issues" for iMesh bug details
SCUtil/iBase
  • NONE
GRM/iRel
  • spec review

[ Planned Week 6 Activities ]

FMDB v1.2
  • entity set/entity group integration
  • tagging improvement
  • parallel FMDB.h development
  • User's Guide update
iMesh/iMeshP
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
SCUtil/iBase
  • NONE
GRM/iRel
  • prototype development
[ Independent Study ]
  • NONE
[ Active Issues ]
  • PartIter with type 4 (all types) returns double of # entities of all types
  • mEntity::get (upward adjacency) crashes when no upward adjacency exists
  • after mesh migration/ghosting, the mesh becomes invalid
  • mesh format portability (sms, vtk, ncdf conversion fails)
    . mesh file written from another mesh format cannot be loaded back
    . e.g. sms written from ncdf or vtk file cannot be loaded back, and vice versa
  • generic data tag fails
  • tag array function fails
  • crash on partitioned mesh loading (+512 parts on B/G)
  • deleting multiple mesh instances fails (chronic bug)
  • improper memory management with tagging
    - simultaneous, multiple mesh instances give error due to single tag management scheme across multiple mesh objects
    - multiple mesh instances a in a row give error during tag data deletion

[ Pended Issues ]

  • FMDB
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
    - (held) 2D NCDF mesh support
  • iMesh(P)
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCUtil/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Wednesday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

ITAPS: ITAPS PIs meeting at SIAM CSE11

Added by Mark Miller over 6 years ago

Our ITAPS PIs meeting at SIAM CSE11 will be held at 5:30pm on Wednesday,
3/2, in meeting room Ruby 1. We have the room until 8:30, so we can get
dinner delivered, or meet and then go out for dinner. If you have a
preference, let me know.

ITAPS: Listen to HPC Podcast with Carl, Mark S. and Tim T.

Added by Mark Miller over 6 years ago

Carl, Mark S. and Tim T. participated in the HPC podcast, at http://www.rce-cast.com/ to talk about ITAPS project.

Listen to the full podcast here, http://www.rce-cast.com/components/com_podcast/media/47RCE-itaps.mp3

PUMI_MESH (FMDB): FMDB Weekly #4 (Jan 24 - Jan 28)

Added by E. Seegyoung Seol over 6 years ago

[ Week 4 Activities ]

FMDB v1.2
  • (on-going) entity set/entity group integration
  • (on-going) tagging improvement
    - traceable and non-traceable tag creation
    - byte type tagging & generic get/set functions
    - entity set type tagging
  • (on-going) parallel FMDB.h development
    - constraint: single part per mesh
    - defining api for partitioning related functions (Min)
  • (on-going) packaging for Sandia release
  • (on-going) User's Guide update
ITAPS
  • iMesh_FMDB improvement & debugging
    - (on-going) implementing with FMDB.h api calls
    - See "Active Issues" for iMesh bug details
SCU/iBase
  • NONE

[ Planned Week 5 Activities ]

FMDB v1.2
  • entity set/entity group integration
  • tagging improvement
  • parallel FMDB.h development
  • User's Guide update
  • FMDB release to SANDIA
ITAPS
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
SCUtil/iBase
  • folder restructuring
  • dependency clean-up (SCUtil shall be independent library)
[ Independent Study ]
  • NONE
[ Active Issues ]
  • mesh format portability (sms, vtk, ncdf conversion fails)
    . mesh file written from another mesh format cannot be loaded back
    . e.g. sms written from ncdf or vtk file cannot be loaded back, and vice versa
  • generic data tag fails
  • tag array function fails
  • crash on partitioned mesh loading (+512 parts on B/G)
  • deleting multiple mesh instances fails (chronic bug)
  • improper memory management with tagging
    - simultaneous, multiple mesh instances give error due to single tag management scheme across multiple mesh objects
    - multiple mesh instances a in a row give error during tag data deletion

[ Pended Issues ]

  • FMDB
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
    - (held) 2D NCDF mesh support
  • FMDB/iMesh(P)
    - iMesh entity handle validity check
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCU/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Tuesday 1pm-2pm
  • FMDB-itaps meeting: Thursday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

PUMI_MESH (FMDB): FMDB Weekly #3 (Jan 17 - Jan 21)

Added by E. Seegyoung Seol over 6 years ago

[ Week 3 Activities ]

FMDB v1.2
  • (on-going) entity set/entity group integration
  • (on-going) tagging improvement
    - single tag implementation for traceable tag (itaps) and non-traceable tag (scorec)
    - entity set/byte type tag data
  • (on-going) FMDB.h development and unit testing
    - constraint: single part per mesh
    - status: approx. 100 out of 150 api developed so far
    - verifying the current implementation (serial & parallel) for Sandia release
  • (on-going) autotool testing with "make check"
  • (on-going) User's Guide update
ITAPS
  • iMesh_FMDB improvement & debugging
    - (on-going) implementing with FMDB.h api calls
    - See "Active Issues" for iMesh bug details
SCU/iBase
  • NONE

[ Planned Week 4 Activities ]

FMDB v1.2
  • entity set/entity group integration
  • tagging improvement
  • FMDB.h development
  • autotool testing with "make check"
  • User's Guide update
  • FMDB release to SANDIA
ITAPS
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
SCUtil/iBase
  • folder restructuring
  • dependency clean-up (SCUtil shall be independent library)
[ Independent Study ]
  • NONE
[ Active Issues ]
  • sms mesh file written with a mesh loaded from vtk cannot be loaded back (classification info mismatch)
  • generic data tag fails
  • tag array function fails
  • crash on partitioned mesh loading (+512 parts on B/G)
  • deleting multiple mesh instances fails (chronic bug)
  • improper memory management with tagging
    - simultaneous, multiple mesh instances give error due to single tag management scheme across multiple mesh objects
    - multiple mesh instances a in a row give error during tag data deletion

[ Pended Issues ]

  • FMDB
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
    - (held) 2D NCDF mesh support
  • FMDB/iMesh(P)
    - iMesh entity handle validity check
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCU/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Tuesday 1pm-2pm
  • FMDB-itaps meeting: Thursday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

PUMI_MESH (FMDB): FMDB Weekly #2 (Jan 10 - Jan 14)

Added by E. Seegyoung Seol almost 7 years ago

[ Week 2 Activities ]

FMDB v1.2
  • FMDB.h development and unit testing (single part per mesh)
    - (on-going) iterator: part, entity set with various conditions (type, topo, reverse classification, entity status,etc.)
    - (on-going) single mesh instance per process
    - (on-going) file I/O (ncdf ascii/bin, sms, vtk) verification
ITAPS
  • iMesh_FMDB improvement & debugging
    - (on-going) revisiting itaps tag
    - (on-going) implementing with FMDB.h api calls
    - See "Active Issues" for iMesh bug details
SCU/iBase
  • (on-going) iterator debugging

[ Planned Week 3 Activities ]

FMDB v1.2
  • parallel mesh file I/O (single part per mesh)
  • file I/O verification
  • part/entity set iterator api
  • User's Guide update (SERIAL FMDB release to SANDIA in Week 4)
ITAPS
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
    - implementing missing itaps tag features (generic data)
    - merging itaps tag and attachable data's data structure for better memory cost (temporary fix)
SCUtil/iBase
  • iterator debugging
  • folder restructuring
  • dependency clean-up (SCUtil shall be independent library)
[ Independent Study ]
  • NONE
[ Active Issues ]
  • sms mesh file written with a mesh loaded from vtk cannot be loaded back (classification info mismatch)
  • iMesh compliance test fails during iterator check-up
  • generic data tag fails
  • tag array function fails
  • crash on partitioned mesh loading (+512 parts on B/G)

[ Pended Issues ]

  • FMDB
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
  • FMDB/iMesh(P)
    - iMesh entity handle validity check
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCU/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Tuesday 1pm-2pm
  • FMDB-itaps meeting: Thursday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

PUMI_MESH (FMDB): FMDB Weekly #1 (Jan 3 - Jan 7)

Added by E. Seegyoung Seol almost 7 years ago

[ Week 1 Activities ]

FMDB v1.2
  • FMDB.h development and unit testing (single part per mesh)
    - (on-going) iterator: part, entity set with various conditions (type, topo, classification, entity status,etc.)
    - (completed) 1-layer ghosting code integration
ITAPS
  • iMesh_FMDB improvement & debugging
    - (on-going) revisiting itaps tag
    - See "Active Issues" for iMesh bug details
  • Released v1.2 tarball to itaps community
SCU/iBase
  • (on-going) iterator debugging

[ Planned Week 2 Activities ]

FMDB v1.2
  • multi-part mesh file I/O
  • ncdf file I/O extension and verification (Kai)
  • vtk file I/O verification
  • part/entity set iterator api
ITAPS
  • iMesh_FMDB improvement & debugging
    - revising with FMDB.h
    - revisiting itaps tag implementation
    - merging itaps tag and attachable data's data structure for better memory cost (temporary fix)
SCUtil/iBase
  • iterator debugging
  • folder restructuring
  • dependency clean-up (SCUtil shall be independent library)
[ Independent Study ]
  • NONE
[ Active Issues ]
  • sms mesh file written with a mesh loaded from vtk cannot be loaded back (classification info mismatch)
  • iMesh compliance test fails during iterator check-up
  • generic data tag fails
  • tag array function fails
  • crash on partitioned mesh loading (+512 parts on B/G)

[ Pended Issues ]

  • FMDB
    - regression testing suite incorporated with Automake and Nightly Build
    - local partitioning (requestor: Min)
    - FMDB.h consolidation
    - automatic attached data migration
    - merging entity group with entity set
    - improved algorithm for migration procedure
    - (held) Exodus II mesh format (requestor: Sandia)
  • FMDB/iMesh(P)
    - iMesh entity handle validity check
    - entity set/tag file I/O with vtk and ncdf files
    - iMeshP implmentation: linking with iZoltan
    - iMeshP compliance test: standardized autotool options
    - distributed vtk file I/O with partition info
  • GMI/iGeom
    - incorporating with entity set and iterator
    - implementing GMI.h
  • SCU/iBase
    - folder reorganization
    - implementing api
    - removing FMDB dependency
    - revisiting entity set for improvement (removing tight coupling with FMDB)
    - combinging entity group with entity set
    - templated tag
  • GRM/iRel
    - NONE
  • MeshAdapt
    - incorporating decision procedure to compute partition objects to migrate before calling FMDB::M_migrate
  • Miscellaneous
    - SCOREC S/W web page update
    - Doxygen
[ Remarks ]
  • FMDB team meeting: Tuesday 1pm-2pm
  • FMDB-itaps meeting: Thursday 11am-12pm
  • ITAPS telecon: Thursday 12pm-1pm
  • iMeshP compliance test verification with FMDB pended due to incomplete implementation such as ghosting & multiple part per process
  • entity group migration with "splitted" entity data stream won't be considered to put into the repository
    Instead, better algorithm for migration procedure will be investigated.

« Previous 1 2 3 4 5 6 Next »

Also available in: Atom