charm.git
15 years agoonly start timer after warmup. and only reset memory count once
Yan Shi [Wed, 12 Apr 2006 16:06:51 +0000 (16:06 +0000)]
only start timer after warmup. and only reset memory count once
every proc

15 years agoImplement alltoall with iget. Currently shadowed as MPI_Alltoall2
Yan Shi [Wed, 12 Apr 2006 04:20:20 +0000 (04:20 +0000)]
Implement alltoall with iget. Currently shadowed as MPI_Alltoall2

15 years agoFixed a bug. Also to work with iget for alltoall.
Yan Shi [Wed, 12 Apr 2006 04:13:09 +0000 (04:13 +0000)]
Fixed a bug. Also to work with iget for alltoall.

15 years agochange for iget
Yan Shi [Wed, 12 Apr 2006 04:08:32 +0000 (04:08 +0000)]
change for iget

15 years agomore conservative locking while coarsening
Nilesh Choudhury [Wed, 12 Apr 2006 02:31:29 +0000 (02:31 +0000)]
more conservative locking while coarsening

15 years agosome memory optimizations
Nilesh Choudhury [Tue, 11 Apr 2006 22:40:01 +0000 (22:40 +0000)]
some memory optimizations

15 years agofixed a bug when it can not be negative when using unsigned
Gengbin Zheng [Tue, 11 Apr 2006 21:35:43 +0000 (21:35 +0000)]
fixed a bug when it can not be negative when using unsigned

15 years agofixed the remaining parallel memory leaks... hopefully parfum adaptivity is
Nilesh Choudhury [Tue, 11 Apr 2006 20:42:15 +0000 (20:42 +0000)]
fixed the remaining parallel memory leaks... hopefully parfum adaptivity is
now memory-leak free :)

15 years agofixed a bunch of leaks in the adaptive higher level algorithms
Nilesh Choudhury [Tue, 11 Apr 2006 17:55:46 +0000 (17:55 +0000)]
fixed a bunch of leaks in the adaptive higher level algorithms

15 years agosome more memory leak fixes and made fem_repair to not iterate more than 4 times
Nilesh Choudhury [Tue, 11 Apr 2006 15:13:42 +0000 (15:13 +0000)]
some more memory leak fixes and made fem_repair to not iterate more than 4 times

15 years agoa new implementation of Irecv and alltoall.
Gengbin Zheng [Tue, 11 Apr 2006 07:22:54 +0000 (07:22 +0000)]
a new implementation of Irecv and alltoall.
Irecv now posted receives in a cmmtable and messages will copied to the buffer in inorder().
Using new irecv to implement alltoall.

15 years agofixed some memory leaks for serial version of adaptivity
Nilesh Choudhury [Tue, 11 Apr 2006 00:31:13 +0000 (00:31 +0000)]
fixed some memory leaks for serial version of adaptivity

15 years agocompilation error.. i don't know how long it was not compiling..
Nilesh Choudhury [Mon, 10 Apr 2006 22:38:14 +0000 (22:38 +0000)]
compilation error.. i don't know how long it was not compiling..
and definitely not many people are using this

15 years agoAdded FEM_Repair.
Terry L. Wilmarth [Mon, 10 Apr 2006 22:38:09 +0000 (22:38 +0000)]
Added FEM_Repair.

15 years agoname changes for functions to calculate offset of a pointer to a stack and retrive...
Gengbin Zheng [Mon, 10 Apr 2006 20:56:07 +0000 (20:56 +0000)]
name changes for functions to calculate offset of a pointer to a stack and retrive a pointer with an offset

15 years agofixed a bug when mem is negative
Gengbin Zheng [Mon, 10 Apr 2006 20:10:49 +0000 (20:10 +0000)]
fixed a bug when mem is negative

15 years agoupdated
Gengbin Zheng [Mon, 10 Apr 2006 19:49:27 +0000 (19:49 +0000)]
updated

15 years agoMakefile change.
Yan Shi [Mon, 10 Apr 2006 19:13:40 +0000 (19:13 +0000)]
Makefile change.

15 years agoUpdates: more correct refine loop. Still need some fixing.
Yan Shi [Mon, 10 Apr 2006 18:39:37 +0000 (18:39 +0000)]
Updates: more correct refine loop. Still need some fixing.

15 years agoAdding files for GridMetisLB.
Greg Koenig [Mon, 10 Apr 2006 17:43:51 +0000 (17:43 +0000)]
Adding files for GridMetisLB.
Not yet added to Makefile(s), and not debugged.

15 years agochanged the scheme to calculate max mem
Gengbin Zheng [Sun, 9 Apr 2006 19:21:26 +0000 (19:21 +0000)]
changed the scheme to calculate max mem

15 years ago*** empty log message ***
Nilesh Choudhury [Sun, 9 Apr 2006 05:13:38 +0000 (05:13 +0000)]
*** empty log message ***

15 years agominor cosmetic changes
Gengbin Zheng [Sun, 9 Apr 2006 05:04:52 +0000 (05:04 +0000)]
minor cosmetic changes

15 years agosome odd special cases because in fem_entity last_invalid and first_invalid
Nilesh Choudhury [Sun, 9 Apr 2006 04:15:44 +0000 (04:15 +0000)]
some odd special cases because in fem_entity last_invalid and first_invalid
are equal to 0, when there are no rows in the entity.

15 years agopass all user attributes across chunks when nodes or elems are lost/added
Nilesh Choudhury [Sat, 8 Apr 2006 21:49:48 +0000 (21:49 +0000)]
pass all user attributes across chunks when nodes or elems are lost/added

15 years agoc syntax.
Yan Shi [Sat, 8 Apr 2006 05:18:16 +0000 (05:18 +0000)]
c syntax.

15 years agosize should be in terms of byte.
Yan Shi [Fri, 7 Apr 2006 19:39:54 +0000 (19:39 +0000)]
size should be in terms of byte.

15 years agoFixed to get ghost with user defined type. Still need to
Yan Shi [Fri, 7 Apr 2006 19:39:08 +0000 (19:39 +0000)]
Fixed to get ghost with user defined type. Still need to
update the refine loop to use the ghost for compute.

15 years agomentioning the old inclusion syntax for charm objects
Filippo Gioachin [Fri, 7 Apr 2006 18:30:20 +0000 (18:30 +0000)]
mentioning the old inclusion syntax for charm objects

15 years agoAdded implementation for MPI_IGet, MPI_IGet_Wait, MPI_IGet_Free.
Yan Shi [Fri, 7 Apr 2006 18:16:26 +0000 (18:16 +0000)]
Added implementation for MPI_IGet, MPI_IGet_Wait, MPI_IGet_Free.

15 years agofirst versions
Laxmikant Kale [Fri, 7 Apr 2006 16:05:50 +0000 (16:05 +0000)]
first versions

15 years agoTest code for Put/Get, and IGet
Yan Shi [Fri, 7 Apr 2006 05:33:27 +0000 (05:33 +0000)]
Test code for Put/Get, and IGet

15 years agoProgram to do jacobi with MPI_IGet. Need correct packing now.
Yan Shi [Fri, 7 Apr 2006 05:32:40 +0000 (05:32 +0000)]
Program to do jacobi with MPI_IGet. Need correct packing now.

15 years agoMinor changes for more memory-efficient, less comm intensive solution transfer.
Terry L. Wilmarth [Fri, 7 Apr 2006 03:51:48 +0000 (03:51 +0000)]
Minor changes for more memory-efficient, less comm intensive solution transfer.

15 years agoExperimenting with quality improvement.
Terry L. Wilmarth [Fri, 7 Apr 2006 03:03:20 +0000 (03:03 +0000)]
Experimenting with quality improvement.

15 years ago*** empty log message ***
Terry L. Wilmarth [Fri, 7 Apr 2006 01:01:50 +0000 (01:01 +0000)]
*** empty log message ***

15 years agoCommented out print statements and increased the quality requirements.
Terry L. Wilmarth [Thu, 6 Apr 2006 23:34:48 +0000 (23:34 +0000)]
Commented out print statements and increased the quality requirements.

15 years ago*** empty log message ***
Nilesh Choudhury [Wed, 5 Apr 2006 22:07:10 +0000 (22:07 +0000)]
*** empty log message ***

15 years ago*** empty log message ***
Nilesh Choudhury [Wed, 5 Apr 2006 21:57:50 +0000 (21:57 +0000)]
*** empty log message ***

15 years agosome minor changes
Gengbin Zheng [Wed, 5 Apr 2006 20:51:04 +0000 (20:51 +0000)]
some minor changes

15 years ago*** empty log message ***
Nilesh Choudhury [Wed, 5 Apr 2006 20:50:22 +0000 (20:50 +0000)]
*** empty log message ***

15 years agodump OPTS options to file OPTS
Gengbin Zheng [Wed, 5 Apr 2006 18:34:17 +0000 (18:34 +0000)]
dump OPTS options to file OPTS

15 years agoAdded code to reset notify and wait counters.
Chao Huang [Wed, 5 Apr 2006 16:54:28 +0000 (16:54 +0000)]
Added code to reset notify and wait counters.

15 years agocommlib to comlib
Gengbin Zheng [Wed, 5 Apr 2006 16:23:43 +0000 (16:23 +0000)]
commlib to comlib

15 years agocompiling obsolete libmodulecommlib.a like usual, this fixed bug in myrinet.
Gengbin Zheng [Wed, 5 Apr 2006 16:22:41 +0000 (16:22 +0000)]
compiling obsolete libmodulecommlib.a like usual, this fixed bug in myrinet.

15 years agoFixed wait notify implementation
Chao Huang [Wed, 5 Apr 2006 03:33:47 +0000 (03:33 +0000)]
Fixed wait notify implementation

15 years agoinclude <complex>
Gengbin Zheng [Tue, 4 Apr 2006 20:53:57 +0000 (20:53 +0000)]
include <complex>

15 years agoall remote methods can now call tcharm specific methods, and also fixed gradateMesh...
Nilesh Choudhury [Tue, 4 Apr 2006 20:24:27 +0000 (20:24 +0000)]
all remote methods can now call tcharm specific methods, and also fixed gradateMesh so that
it works inside a barrier, thus not interfering with other chunks who might be
modifying the same nodes. Locking would be very costly and since we go over all nodes
together, it is better to sync everyone and do this once

15 years agoAdded gfortran option
Chao Huang [Tue, 4 Apr 2006 18:57:55 +0000 (18:57 +0000)]
Added gfortran option

15 years ago*** empty log message ***
Nilesh Choudhury [Tue, 4 Apr 2006 07:43:06 +0000 (07:43 +0000)]
*** empty log message ***

15 years agoAdded pup'er for std::complex<T>
Chao Huang [Tue, 4 Apr 2006 00:23:48 +0000 (00:23 +0000)]
Added pup'er for std::complex<T>

15 years agosome more set of tests to verify the preserveness of orderedness of nodes in elements
Nilesh Choudhury [Mon, 3 Apr 2006 23:50:06 +0000 (23:50 +0000)]
some more set of tests to verify the preserveness of orderedness of nodes in elements
A compiler flag FEM_ELEMSORDERED should be used when the input mesh has the
connectivity of all elements as ordered.

15 years agoadded a function getSignedArea(int,int,int)
Nilesh Choudhury [Mon, 3 Apr 2006 23:41:31 +0000 (23:41 +0000)]
added a function getSignedArea(int,int,int)

15 years agosmoothing across chunks will not crash now. MeshSizing needs to be
Nilesh Choudhury [Mon, 3 Apr 2006 21:43:12 +0000 (21:43 +0000)]
smoothing across chunks will not crash now. MeshSizing needs to be
handled properly across chunks by the gradation code

15 years agoadded priority to locking, so if two chunks are fighting for a set of resources,
Nilesh Choudhury [Sun, 2 Apr 2006 23:35:27 +0000 (23:35 +0000)]
added priority to locking, so if two chunks are fighting for a set of resources,
the chunk with lowest id, will have priority. Also, added some randomization
of how many times edge_bisect and edge_flip will try before they give up
on the operation. This means that longest edge bisection will have to
ensure that the operation is complete. So, more work needs to be done there.
This now avoids all livelocks for node locks for virtual processors,
which plagued us for a while.

15 years agoAdded a file for flow control for iget.
Yan Shi [Sun, 2 Apr 2006 19:25:10 +0000 (19:25 +0000)]
Added a file for flow control for iget.

15 years agoChange of futureattach function for iget.
Yan Shi [Sun, 2 Apr 2006 19:23:52 +0000 (19:23 +0000)]
Change of futureattach function for iget.

15 years agoChanged interface for a attachfuture call. only used in
Yan Shi [Sun, 2 Apr 2006 19:18:58 +0000 (19:18 +0000)]
Changed interface for a attachfuture call. only used in
iget code.

15 years agoChanges for Iget flow control: better modular coding.
Yan Shi [Sun, 2 Apr 2006 19:18:02 +0000 (19:18 +0000)]
Changes for Iget flow control: better modular coding.

15 years agoprint error mesg when CMK_CF90 is not defined
Gengbin Zheng [Fri, 31 Mar 2006 23:26:59 +0000 (23:26 +0000)]
print error mesg when CMK_CF90 is not defined

15 years agoupdated
Gengbin Zheng [Fri, 31 Mar 2006 22:39:19 +0000 (22:39 +0000)]
updated

15 years agormeoved CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:35:47 +0000 (22:35 +0000)]
rmeoved CMK_MOD_EXT

15 years agoget rid of CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:35:26 +0000 (22:35 +0000)]
get rid of CMK_MOD_EXT

15 years agoremoved
Gengbin Zheng [Fri, 31 Mar 2006 22:24:27 +0000 (22:24 +0000)]
removed

15 years agoremoved, not supported anymore
Gengbin Zheng [Fri, 31 Mar 2006 22:23:59 +0000 (22:23 +0000)]
removed, not supported anymore

15 years agoremoved CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:22:15 +0000 (22:22 +0000)]
removed CMK_MOD_EXT

15 years agoremoved CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:21:28 +0000 (22:21 +0000)]
removed CMK_MOD_EXT

15 years agoremoved CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:13:11 +0000 (22:13 +0000)]
removed CMK_MOD_EXT

15 years agormeoved CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:10:03 +0000 (22:10 +0000)]
rmeoved CMK_MOD_EXT

15 years agoremoved defintion of CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 22:09:15 +0000 (22:09 +0000)]
removed defintion of CMK_MOD_EXT

15 years agoremoved CMK_MOD_EXT definition
Gengbin Zheng [Fri, 31 Mar 2006 22:07:39 +0000 (22:07 +0000)]
removed CMK_MOD_EXT definition

15 years agoremove manual definition of CMK_MOD_EXT.
Gengbin Zheng [Fri, 31 Mar 2006 22:05:14 +0000 (22:05 +0000)]
remove manual definition of CMK_MOD_EXT.

15 years agoadded a lot of comments to help understanding
Gengbin Zheng [Fri, 31 Mar 2006 20:57:38 +0000 (20:57 +0000)]
added a lot of comments to help understanding

15 years agoadded support for carrying userFlag field.
Eric Bohm [Fri, 31 Mar 2006 20:45:35 +0000 (20:45 +0000)]
added support for carrying userFlag field.
userFlag is optional and defaults to -1 as in the ck-core array reduction.

15 years agolet autoconfig test fortran mod ext
Gengbin Zheng [Fri, 31 Mar 2006 18:56:25 +0000 (18:56 +0000)]
let autoconfig test fortran mod ext

15 years agolet autoconfig decide fortran mod name ext
Gengbin Zheng [Fri, 31 Mar 2006 18:54:10 +0000 (18:54 +0000)]
let autoconfig decide fortran mod name ext

15 years agolet autoconfig decide fortran mod name
Gengbin Zheng [Fri, 31 Mar 2006 18:50:31 +0000 (18:50 +0000)]
let autoconfig decide fortran mod name

15 years agolet autoconfig figure out CMK_MOD_NAME_ALLCAPS amd CMK_MOD_EXT
Gengbin Zheng [Fri, 31 Mar 2006 18:44:44 +0000 (18:44 +0000)]
let autoconfig figure out CMK_MOD_NAME_ALLCAPS amd CMK_MOD_EXT

15 years agoauto test if fortran mod name is captialized and what's the extension of fortran...
Gengbin Zheng [Fri, 31 Mar 2006 18:43:54 +0000 (18:43 +0000)]
auto test if fortran mod name is captialized and what's the extension of fortran mod.

15 years agoadded a stronger test for flip test
Nilesh Choudhury [Fri, 31 Mar 2006 06:55:27 +0000 (06:55 +0000)]
added a stronger test for flip test

15 years agoupdating syntax of chare object inheritance: all should inherit from CBase_<classname>
Filippo Gioachin [Fri, 31 Mar 2006 02:14:22 +0000 (02:14 +0000)]
updating syntax of chare object inheritance: all should inherit from CBase_<classname>

15 years ago*** empty log message ***
Nilesh Choudhury [Fri, 31 Mar 2006 01:01:56 +0000 (01:01 +0000)]
*** empty log message ***

15 years agoBugfix to correct message queue corruption.
David Kunzman [Thu, 30 Mar 2006 23:53:55 +0000 (23:53 +0000)]
Bugfix to correct message queue corruption.

15 years agoremoved a c++ comment in c
Gengbin Zheng [Thu, 30 Mar 2006 15:28:43 +0000 (15:28 +0000)]
removed a c++ comment in c

15 years agofixed a bunch of small bugs including GetElementOnEdge, modifying the lock even when a
Nilesh Choudhury [Thu, 30 Mar 2006 01:15:11 +0000 (01:15 +0000)]
fixed a bunch of small bugs including GetElementOnEdge, modifying the lock even when a
chunk is losing a node and it won't be a ghost later

15 years agoadded FEM_TESTMESH and FEM_SIMPLEREFINE and FEM_SIMPLECOARSEN to the interface
Nilesh Choudhury [Thu, 30 Mar 2006 01:13:25 +0000 (01:13 +0000)]
added FEM_TESTMESH and FEM_SIMPLEREFINE and FEM_SIMPLECOARSEN to the interface

15 years agoRunscript for the benchmark
Isaac Dooley [Wed, 29 Mar 2006 23:08:21 +0000 (23:08 +0000)]
Runscript for the benchmark

15 years agoAdded another all to all test.
Isaac Dooley [Wed, 29 Mar 2006 22:49:16 +0000 (22:49 +0000)]
Added another all to all test.

15 years agoAdded a High-Watermark Memory Usage function: CmiMaxMemoryUsage();
Isaac Dooley [Wed, 29 Mar 2006 22:11:17 +0000 (22:11 +0000)]
Added a High-Watermark Memory Usage function: CmiMaxMemoryUsage();

15 years agoRemoved windows line feeds.
Isaac Dooley [Wed, 29 Mar 2006 19:04:22 +0000 (19:04 +0000)]
Removed windows line feeds.

15 years agoAdded some additional printouts for debugging.
David Kunzman [Wed, 29 Mar 2006 16:52:55 +0000 (16:52 +0000)]
Added some additional printouts for debugging.

15 years agoUpdated SPE Runtime to be able to handle buffers of size greater than 16k. Made
David Kunzman [Wed, 29 Mar 2006 16:51:07 +0000 (16:51 +0000)]
Updated SPE Runtime to be able to handle buffers of size greater than 16k.  Made
some additional small modifications (updated comments, zero writeOnly memory, etc.)

15 years agoremove_node_local might not be totally disconnected
Nilesh Choudhury [Wed, 29 Mar 2006 00:09:36 +0000 (00:09 +0000)]
remove_node_local might not be totally disconnected

15 years agoensure that is_valid is only called on the correct entity, it cannot be
Nilesh Choudhury [Tue, 28 Mar 2006 23:39:27 +0000 (23:39 +0000)]
ensure that is_valid is only called on the correct entity, it cannot be
called with a negative index

15 years agofixed a bunch of bugs including some race conditions, some cases of using stale
Nilesh Choudhury [Tue, 28 Mar 2006 22:20:02 +0000 (22:20 +0000)]
fixed a bunch of bugs including some race conditions, some cases of using stale
data, i.e. having a remote call between when some data has been acquired and used.
This remote call could effectively change this data.

15 years agoFixed small typo.
Esteban Pauli [Tue, 28 Mar 2006 21:24:25 +0000 (21:24 +0000)]
Fixed small typo.

15 years agoSome changes by Aaron. Should be close to working!
Isaac Dooley [Tue, 28 Mar 2006 21:20:28 +0000 (21:20 +0000)]
Some changes by Aaron. Should be close to working!

15 years agoBug in a modified variable name.
Greg Koenig [Tue, 28 Mar 2006 21:06:10 +0000 (21:06 +0000)]
Bug in a modified variable name.

15 years agoAdded GridCommRefineLB to the default build process.
Greg Koenig [Sat, 25 Mar 2006 03:57:41 +0000 (03:57 +0000)]
Added GridCommRefineLB to the default build process.

15 years agofix meta_free for memaligned data (allocated with meta_memalign)
Filippo Gioachin [Fri, 24 Mar 2006 23:33:10 +0000 (23:33 +0000)]
fix meta_free for memaligned data (allocated with meta_memalign)