charm.git
16 years agofixed spelling mistake
Isaac Dooley [Tue, 2 Aug 2005 16:46:15 +0000 (16:46 +0000)]
fixed spelling mistake

16 years agoChecking in debug version temporarily.
Terry L. Wilmarth [Tue, 2 Aug 2005 15:59:04 +0000 (15:59 +0000)]
Checking in debug version temporarily.

16 years agotest program for smoothing, adjacency and modification
Dmitriy Ofman [Mon, 1 Aug 2005 23:55:01 +0000 (23:55 +0000)]
test program for smoothing, adjacency and modification

16 years agosome bugs fixed.. apeears to work now for most cases of edge_bisect, vertex_remove...
Nilesh Choudhury [Mon, 1 Aug 2005 21:52:43 +0000 (21:52 +0000)]
some bugs fixed.. apeears to work now for most cases of edge_bisect, vertex_remove & edge_contract

16 years agoChanges to do r2c tranform in plane with 2D r2c then 2D c2c. This
Yan Shi [Mon, 1 Aug 2005 20:27:08 +0000 (20:27 +0000)]
Changes to do r2c tranform in plane with 2D r2c then 2D c2c. This
way, no conversion of half-complex to complex is needed.

16 years agoput in conditions for graceful exit in uncertain cases
Nilesh Choudhury [Mon, 1 Aug 2005 16:49:02 +0000 (16:49 +0000)]
put in conditions for graceful exit in uncertain cases

16 years agoimits.h should be added, becaise INT_MIN is defined in this header, which is not...
Nilesh Choudhury [Mon, 1 Aug 2005 03:31:44 +0000 (03:31 +0000)]
imits.h should be added, becaise INT_MIN is defined in this header, which is not included
by default in many systems.
C
V

16 years agofixed some bugs
Nilesh Choudhury [Fri, 29 Jul 2005 18:26:44 +0000 (18:26 +0000)]
fixed some bugs

16 years agobug related to addnode in ghostsend
Nilesh Choudhury [Fri, 29 Jul 2005 17:10:41 +0000 (17:10 +0000)]
bug related to addnode in ghostsend

16 years agobugs for losing ghost nodes with no connectivity fixed... actually this should nto...
Nilesh Choudhury [Fri, 29 Jul 2005 16:08:27 +0000 (16:08 +0000)]
bugs for losing ghost nodes with no connectivity fixed... actually this should nto be a bug,
but the way it is used in the mesh modification operations,
bisect/contract, etc it is useful if it is not lost

16 years agoAdding PHOLD Benchmark for PDES.
Terry L. Wilmarth [Thu, 28 Jul 2005 22:33:16 +0000 (22:33 +0000)]
Adding PHOLD Benchmark for PDES.
Task: POSE

16 years agoTrying some adjustments.
Terry L. Wilmarth [Thu, 28 Jul 2005 22:31:12 +0000 (22:31 +0000)]
Trying some adjustments.

16 years ago*** empty log message ***
Terry L. Wilmarth [Thu, 28 Jul 2005 19:18:26 +0000 (19:18 +0000)]
*** empty log message ***

16 years agofixed a stupid bug in the ibm compiler, which was not able to build due to recursive...
Nilesh Choudhury [Thu, 28 Jul 2005 17:37:00 +0000 (17:37 +0000)]
fixed a stupid bug in the ibm compiler, which was not able to build due to recursive file includes.i

16 years agohacks for cases where all nodes of an element are shared
Nilesh Choudhury [Thu, 28 Jul 2005 07:44:01 +0000 (07:44 +0000)]
hacks for cases where all nodes of an element are shared

16 years agofixed a bug related to locking nodes/elements we need for any of the bisect, vertex...
Nilesh Choudhury [Thu, 28 Jul 2005 01:43:34 +0000 (01:43 +0000)]
fixed a bug related to locking nodes/elements we need for any of the bisect, vertex remove, flip, etc operations.

16 years agofixed a comment.
Gengbin Zheng [Wed, 27 Jul 2005 19:32:11 +0000 (19:32 +0000)]
fixed a comment.

16 years agodeleted by error one include file...
Filippo Gioachin [Wed, 27 Jul 2005 17:06:34 +0000 (17:06 +0000)]
deleted by error one include file...

16 years agochanging random to srand48 for solaris problems
Filippo Gioachin [Wed, 27 Jul 2005 17:05:05 +0000 (17:05 +0000)]
changing random to srand48 for solaris problems

16 years agoghost node indices now translated
Nilesh Choudhury [Tue, 26 Jul 2005 20:06:30 +0000 (20:06 +0000)]
ghost node indices now translated

16 years agosome bug fixes
Nilesh Choudhury [Tue, 26 Jul 2005 16:02:45 +0000 (16:02 +0000)]
some bug fixes

16 years agointerpolation is done on new node creation now. Also a bug fix for
Nilesh Choudhury [Mon, 25 Jul 2005 22:17:38 +0000 (22:17 +0000)]
interpolation is done on new node creation now. Also a bug fix for
ghost node deletion has been checked in.

16 years agonew include file
Filippo Gioachin [Mon, 25 Jul 2005 16:41:52 +0000 (16:41 +0000)]
new include file

16 years agoinclude file
Filippo Gioachin [Mon, 25 Jul 2005 16:40:59 +0000 (16:40 +0000)]
include file

16 years agoforgot comlib.h file...
Filippo Gioachin [Mon, 25 Jul 2005 16:37:47 +0000 (16:37 +0000)]
forgot comlib.h file...

16 years agochanging comlib.h into convcomlib.h, and adding a new comlib.h file under
Filippo Gioachin [Mon, 25 Jul 2005 16:37:07 +0000 (16:37 +0000)]
changing comlib.h into convcomlib.h, and adding a new comlib.h file under
ck-com. This will be used as include file by user programs, in the place of all
the different include files comlib has.

16 years ago some bug fixes done and locking code in place for vertex split & edge contraction.
Nilesh Choudhury [Mon, 25 Jul 2005 01:22:07 +0000 (01:22 +0000)]
 some bug fixes done and locking code in place for vertex split & edge contraction.
Both these operations seem to work in serial, but there seems to be some issues in parallel.

16 years ago added a couple of functions in fem_mutil needed by refine_leb
Nilesh Choudhury [Sun, 24 Jul 2005 21:41:35 +0000 (21:41 +0000)]
 added a couple of functions in fem_mutil needed by refine_leb

16 years agovertex_remove should be working now for shared edges as well.
Nilesh Choudhury [Sun, 24 Jul 2005 21:02:56 +0000 (21:02 +0000)]
vertex_remove should be working now for shared edges as well.
However, if any of the chunks involved has one of the involved nodes as
a ghost node from more than one chunks.. I am not sure if the explanation is clear.
Anyway this happens because the initial ghost node layer needs to have
some more information than it currently has. This is also the reason
for edge_bisect not working at places where three chunks meet. Because
such a situation arises only if there is a node shared by more than 2 chunks.

16 years agoedge bisect should work now for all cases except when three chunks meet at a point.
Nilesh Choudhury [Sun, 24 Jul 2005 19:17:08 +0000 (19:17 +0000)]
edge bisect should work now for all cases except when three chunks meet at a point.
In such a situation there is some ambiguity because one chunk does not know
what the ghost numbers for all nodes on another chunk. Will fix this soon.

16 years agoRemoved inline from the definition of some functions
Sayantan Chakravorty [Sat, 23 Jul 2005 20:13:20 +0000 (20:13 +0000)]
Removed inline from the definition of some functions

16 years agoAdded a special case to always compile bglmachine.C w/ GNU g++
Chao Huang [Sat, 23 Jul 2005 19:54:36 +0000 (19:54 +0000)]
Added a special case to always compile bglmachine.C w/ GNU g++

16 years agoaddVarIndexAttribute added to FEM_Entity to add an attribute of
Sayantan Chakravorty [Sat, 23 Jul 2005 16:39:56 +0000 (16:39 +0000)]
addVarIndexAttribute added to FEM_Entity to add an attribute of
type FEM_VarIndexAttribute

16 years agoAdded a propagating refinement method for longest edge bisection. Moved FEM_Adapt...
Terry L. Wilmarth [Fri, 22 Jul 2005 22:13:51 +0000 (22:13 +0000)]
Added a propagating refinement method for longest edge bisection. Moved FEM_Adapt object inside of femMeshModify.  Lots of integration stuff.
Task: FEM

16 years agosome more bug fixes
Nilesh Choudhury [Fri, 22 Jul 2005 21:45:44 +0000 (21:45 +0000)]
some more bug fixes

16 years agochanged bzero to memset for solaris
Filippo Gioachin [Fri, 22 Jul 2005 17:52:21 +0000 (17:52 +0000)]
changed bzero to memset for solaris

16 years agoAdding interpolate header back to Makefile. Was mysteriously removed.
Terry L. Wilmarth [Fri, 22 Jul 2005 16:21:06 +0000 (16:21 +0000)]
Adding interpolate header back to Makefile.  Was mysteriously removed.

16 years agoAdding back headers for refine which mysteriously were removed in CVS.
Terry L. Wilmarth [Fri, 22 Jul 2005 16:18:18 +0000 (16:18 +0000)]
Adding back headers for refine which mysteriously were removed in CVS.

16 years agoincluding comlib multicast example in the nightly autobuild
Filippo Gioachin [Fri, 22 Jul 2005 02:58:22 +0000 (02:58 +0000)]
including comlib multicast example in the nightly autobuild

16 years agodeleted duplicated definitions (from comlib.h)
Filippo Gioachin [Fri, 22 Jul 2005 02:47:45 +0000 (02:47 +0000)]
deleted duplicated definitions (from comlib.h)

16 years agostyle change for commlib.tex
Filippo Gioachin [Fri, 22 Jul 2005 01:11:33 +0000 (01:11 +0000)]
style change for commlib.tex

16 years agoupdated commlib manual, so to ease the user understanding. now only one single
Filippo Gioachin [Fri, 22 Jul 2005 00:41:53 +0000 (00:41 +0000)]
updated commlib manual, so to ease the user understanding. now only one single
interface is present, which include everything

16 years agoFixing bugs and handling of the var_id types.
Isaac Dooley [Thu, 21 Jul 2005 23:03:41 +0000 (23:03 +0000)]
Fixing bugs and handling of the var_id types.

16 years agoFixed broken makefile
Isaac Dooley [Thu, 21 Jul 2005 20:47:59 +0000 (20:47 +0000)]
Fixed broken makefile

16 years agosome more bug fixes
Nilesh Choudhury [Thu, 21 Jul 2005 20:29:26 +0000 (20:29 +0000)]
some more bug fixes

16 years agoNow element copy-based default interpolation method is implemented. Again, it compil...
Terry L. Wilmarth [Thu, 21 Jul 2005 19:15:24 +0000 (19:15 +0000)]
Now element copy-based default interpolation method is implemented.  Again, it compiles, but has not been tested.
Task: FEM

16 years agoFixed irritating tabbing to make code readable.
Terry L. Wilmarth [Thu, 21 Jul 2005 18:42:11 +0000 (18:42 +0000)]
Fixed irritating tabbing to make code readable.

16 years agoInterpolation code can now be overridden via inheritance as well as by setting partic...
Terry L. Wilmarth [Thu, 21 Jul 2005 18:18:26 +0000 (18:18 +0000)]
Interpolation code can now be overridden via inheritance as well as by setting particular function pointers.  All node interpolation functions have a default behavior that averages real data and minimizes integer data, for all attributes.  These should work for all 2D and 3D cases, for inserting nodes on edges with a weight, for inserting on a face of 3 or 4 nodes by strict averaging, and for inserting in an arbitrary volume of 8 nodes or less, again by strict averaging.  This code compiles but has not been tested.  The Makefile has been modified to build it.
Task: FEM

16 years agoAdded additional interpolate operations for handling default case of node interpolati...
Terry L. Wilmarth [Thu, 21 Jul 2005 18:02:57 +0000 (18:02 +0000)]
Added additional interpolate operations for handling default case of node interpolation for a variety of node insertion methods.
Task: FEM

16 years agoUse FEM_Is_ghost_index instead of hardcoded comparison.
Isaac Dooley [Thu, 21 Jul 2005 17:30:23 +0000 (17:30 +0000)]
Use FEM_Is_ghost_index instead of hardcoded comparison.

16 years agoFixed n2n and n2e creation bugs. Now I think the work in parallel with ghosts!
Isaac Dooley [Thu, 21 Jul 2005 17:25:07 +0000 (17:25 +0000)]
Fixed n2n and n2e creation bugs. Now I think the work in parallel with ghosts!

16 years agoBug in ListSend is fixed.
Chao Huang [Thu, 21 Jul 2005 15:37:12 +0000 (15:37 +0000)]
Bug in ListSend is fixed.

16 years agoAdded xlc option and a readme file stating this is experimental and
Chao Huang [Thu, 21 Jul 2005 15:02:26 +0000 (15:02 +0000)]
Added xlc option and a readme file stating this is experimental and
temporary config

16 years agoUpdated conv-mach.sh per Gengbin's suggestion.
Chao Huang [Thu, 21 Jul 2005 14:38:05 +0000 (14:38 +0000)]
Updated conv-mach.sh per Gengbin's suggestion.

16 years agosome bug fixes
Nilesh Choudhury [Thu, 21 Jul 2005 05:18:06 +0000 (05:18 +0000)]
some bug fixes

16 years agoupdated with an example in bound array.
Gengbin Zheng [Thu, 21 Jul 2005 04:33:08 +0000 (04:33 +0000)]
updated with an example in bound array.

16 years agoexample program for commlib multicast strategies
Filippo Gioachin [Thu, 21 Jul 2005 01:52:33 +0000 (01:52 +0000)]
example program for commlib multicast strategies

16 years agofixed a bug in MultiRing where the sorting of the data was causing problems.
Filippo Gioachin [Thu, 21 Jul 2005 01:43:00 +0000 (01:43 +0000)]
fixed a bug in MultiRing where the sorting of the data was causing problems.
Now getNewMessage has a parameter for requesting the sorting of the processors.

16 years agoadded a few functionalities to commlib
Filippo Gioachin [Thu, 21 Jul 2005 01:42:04 +0000 (01:42 +0000)]
added a few functionalities to commlib

16 years agobug fixes... print functions, etc edge_flip & edge_bisect work in serial!!!
Isaac Dooley [Thu, 21 Jul 2005 00:06:59 +0000 (00:06 +0000)]
bug fixes... print functions, etc edge_flip & edge_bisect work in serial!!!

16 years agotest fortran version
Gengbin Zheng [Wed, 20 Jul 2005 22:40:29 +0000 (22:40 +0000)]
test fortran version
currently ignores return status.

16 years agofixed a bug in previous checkin.
Gengbin Zheng [Wed, 20 Jul 2005 22:39:44 +0000 (22:39 +0000)]
fixed a bug in previous checkin.

16 years agoRemoved spurious print.
Terry L. Wilmarth [Wed, 20 Jul 2005 21:40:02 +0000 (21:40 +0000)]
Removed spurious print.

16 years agobetter testing for both f77 and f90 compiler.
Gengbin Zheng [Wed, 20 Jul 2005 21:34:19 +0000 (21:34 +0000)]
better testing for both f77 and f90 compiler.

16 years agoModified to use new interface
Isaac Dooley [Wed, 20 Jul 2005 21:26:15 +0000 (21:26 +0000)]
Modified to use new interface

16 years agofollow link to determine fortran lib correctly.
Gengbin Zheng [Wed, 20 Jul 2005 21:19:18 +0000 (21:19 +0000)]
follow link to determine fortran lib correctly.

16 years agofixed to support symbolic link of ifc binary so that fortran lib dir is correctly...
Gengbin Zheng [Wed, 20 Jul 2005 21:15:53 +0000 (21:15 +0000)]
fixed to support symbolic link of ifc binary so that fortran lib dir is correctly set.

16 years agoadded a new removenode in an idxl list
Nilesh Choudhury [Wed, 20 Jul 2005 20:36:24 +0000 (20:36 +0000)]
added a new removenode in an idxl list

16 years agoadded functionality for fem_remove_element for both shared and ghost elements.
Nilesh Choudhury [Wed, 20 Jul 2005 20:05:47 +0000 (20:05 +0000)]
added functionality for fem_remove_element for both shared and ghost elements.

16 years agoAdding the outline for a solution transfer module.
Terry L. Wilmarth [Wed, 20 Jul 2005 20:03:12 +0000 (20:03 +0000)]
Adding the outline for a solution transfer module.
Task: FEM

16 years agoreverted previous "optimization" that trying to poll network in CmiGetNonLocal()...
Gengbin Zheng [Wed, 20 Jul 2005 19:29:13 +0000 (19:29 +0000)]
reverted previous "optimization" that trying to poll network in CmiGetNonLocal() when msg is NULL (this slow namd down).

16 years agoFixed an assertion which incorrectly caused a crash.
Isaac Dooley [Wed, 20 Jul 2005 16:35:13 +0000 (16:35 +0000)]
Fixed an assertion which incorrectly caused a crash.

16 years agoAdding refinement header and empty function.
Terry L. Wilmarth [Wed, 20 Jul 2005 16:03:04 +0000 (16:03 +0000)]
Adding refinement header and empty function.

16 years agoA bug fix.
Isaac Dooley [Wed, 20 Jul 2005 14:47:24 +0000 (14:47 +0000)]
A bug fix.

16 years agodded some bug fixes for add_eleemnt
Nilesh Choudhury [Wed, 20 Jul 2005 06:01:06 +0000 (06:01 +0000)]
dded some bug fixes for add_eleemnt

16 years agoInitialized a pointer to 0 so it can be checked to see if it should be deleted.
Isaac Dooley [Wed, 20 Jul 2005 02:08:01 +0000 (02:08 +0000)]
Initialized a pointer to 0 so it can be checked to see if it should be deleted.

16 years agoAdded some more default parameters.
Isaac Dooley [Wed, 20 Jul 2005 02:07:15 +0000 (02:07 +0000)]
Added some more default parameters.

16 years agoAdded assertions to check the range of indices passed into the *valid* functions
Isaac Dooley [Wed, 20 Jul 2005 02:02:18 +0000 (02:02 +0000)]
Added assertions to check the range of indices passed into the *valid* functions

16 years agomost cases for add_element other than the flip should work now
Nilesh Choudhury [Tue, 19 Jul 2005 23:28:19 +0000 (23:28 +0000)]
most cases for add_element other than the flip should work now

16 years agoTightened up the constraints on adaptivity by size.
Terry L. Wilmarth [Tue, 19 Jul 2005 22:38:29 +0000 (22:38 +0000)]
Tightened up the constraints on adaptivity by size.

16 years agoFixed mpi-blugenel script to work on "typical" config on BGL machines.
Chao Huang [Tue, 19 Jul 2005 21:04:23 +0000 (21:04 +0000)]
Fixed mpi-blugenel script to work on "typical" config on BGL machines.

16 years agoChecking in Rocrem-related fixes to collision detection. Most of the files just...
Terry L. Wilmarth [Tue, 19 Jul 2005 18:26:49 +0000 (18:26 +0000)]
Checking in Rocrem-related fixes to collision detection.  Most of the files just got retabified during debugging stages...
Task: Rocrem

16 years agoadd_node_local() and add_element_local() now have flag for whether to add the new...
Isaac Dooley [Tue, 19 Jul 2005 17:03:54 +0000 (17:03 +0000)]
add_node_local() and add_element_local()  now have flag for whether to add the new item as a ghost.

16 years agoadding alloca include for non-linux machines
Filippo Gioachin [Tue, 19 Jul 2005 16:38:53 +0000 (16:38 +0000)]
adding alloca include for non-linux machines

16 years agoFixed real-to-complex , now produces the correct result. Approach taken to put the
Yan Shi [Tue, 19 Jul 2005 14:30:59 +0000 (14:30 +0000)]
Fixed real-to-complex , now produces the correct result. Approach taken to put the
real-to-complex call within a local plane, thus plane creation will be the same.

16 years agoTweaked to handle section multicast and migration smoothly.
Eric Bohm [Tue, 19 Jul 2005 04:41:19 +0000 (04:41 +0000)]
Tweaked to handle section multicast and migration smoothly.
Stripped out redundant code.
Stripped out martialling wrappers which just slow things down.

16 years agoAdded case to properly update edge boundaries along element collapses.
Terry L. Wilmarth [Tue, 19 Jul 2005 04:10:28 +0000 (04:10 +0000)]
Added case to properly update edge boundaries along element collapses.
Task: TMRC2D

16 years agointermediate stage of add_remote_element
Nilesh Choudhury [Tue, 19 Jul 2005 03:02:56 +0000 (03:02 +0000)]
intermediate stage of add_remote_element

16 years agocommented out a few useless routines
Filippo Gioachin [Tue, 19 Jul 2005 02:06:02 +0000 (02:06 +0000)]
commented out a few useless routines

16 years agodeleting CkPrintf...
Filippo Gioachin [Tue, 19 Jul 2005 01:09:08 +0000 (01:09 +0000)]
deleting CkPrintf...

16 years agofix of the MultiRingMulticast strategy
Filippo Gioachin [Tue, 19 Jul 2005 01:02:24 +0000 (01:02 +0000)]
fix of the MultiRingMulticast strategy

16 years agoFixed a bug handling ghosts in FEM_Node::fillNodeAdjacency
Isaac Dooley [Mon, 18 Jul 2005 21:57:09 +0000 (21:57 +0000)]
Fixed a bug handling ghosts in FEM_Node::fillNodeAdjacency

16 years agotouch up section reduction callback during initialize
Eric Bohm [Mon, 18 Jul 2005 21:53:49 +0000 (21:53 +0000)]
touch up section reduction callback during initialize

16 years agowrappers for interacting with section reduction paircalc
Eric Bohm [Mon, 18 Jul 2005 21:17:03 +0000 (21:17 +0000)]
wrappers for interacting with section reduction paircalc

16 years agoAdded ghost handling to n2n and n2e functions.
Isaac Dooley [Mon, 18 Jul 2005 21:16:03 +0000 (21:16 +0000)]
Added ghost handling to n2n and n2e functions.

16 years agoSmaller fixes missed in last checkin
Chao Huang [Mon, 18 Jul 2005 19:58:26 +0000 (19:58 +0000)]
Smaller fixes missed in last checkin

16 years agoModified the communication advance scheme to be more aggressive.
Chao Huang [Mon, 18 Jul 2005 19:57:23 +0000 (19:57 +0000)]
Modified the communication advance scheme to be more aggressive.

16 years agoModified some comments & whitespace.
Isaac Dooley [Mon, 18 Jul 2005 19:37:04 +0000 (19:37 +0000)]
Modified some comments & whitespace.

16 years agoRemoved duplicate entries in n2e creation. fillElemAdjacencyTable() now checks if...
Isaac Dooley [Mon, 18 Jul 2005 19:21:43 +0000 (19:21 +0000)]
Removed duplicate entries in n2e creation. fillElemAdjacencyTable() now checks if an entry is in the table before adding it.

16 years agoFixing some problems with initial n2e and n2n generation.
Isaac Dooley [Mon, 18 Jul 2005 18:59:24 +0000 (18:59 +0000)]
Fixing some problems with initial n2e and n2n generation.