11 years agosince on craxy each node figure out its own cpu topology, comm thread does not have...
Gengbin Zheng [Sun, 11 Apr 2010 02:24:58 +0000 (22:24 -0400)]
since on craxy each node figure out its own cpu topology, comm thread does not have to call progress engine.

11 years agoCmiBarrier() in timer init should be called on all ranks.
Gengbin Zheng [Sat, 10 Apr 2010 02:50:58 +0000 (22:50 -0400)]
CmiBarrier() in timer init should be called on all ranks.

11 years agouse context thread
Gengbin Zheng [Sat, 10 Apr 2010 02:30:40 +0000 (22:30 -0400)]
use context thread

11 years agocharmc: add flag -gendeps to generate ci file dependencies
Ramprasad Venkataraman [Thu, 8 Apr 2010 19:35:07 +0000 (14:35 -0500)]
charmc: add flag -gendeps to generate ci file dependencies

First pass at making charmc take responsibility for providing dependency
information and compilation rules for charm's ci files. Flag -gendeps
used with a ci file should send dependency info and compilation rules to
stdout. It currently works only if preprocess flags are not specified
and is silently ignored for all other file types.

Several potential to-dos in order to make this cleaner, more complete and robust.
We could make charmc
- respect -I include paths in order to generate dependencies with complete path info
- respond to all file types by delegating c/c++ dep generation to the appropriate compilers
  One benefit of this would be automatically stripping charm headers from dependencies just like std headers
- usurp the -M -MM, -MG flags and provide portable behavior for all of these

11 years agobuild util: Script that will generate make dependencies from a ci file
Ramprasad Venkataraman [Thu, 8 Apr 2010 19:19:39 +0000 (14:19 -0500)]
build util: Script that will generate make dependencies from a ci file

One of the big issues with automating builds for charm++ programs is that decl and
def files are generated during the build. And since the ci file name does not correspond
to the names of the generated headers, pattern rules in make cannot help.

Facilitate the automation of builds by providing a script as part of charm that will generate
make-style dependency information from a ci file. This coupled with existing compiler mechanisms
for C/C++ code should go a long way in making large charm++ projects manageable.

11 years agodo the same thing to statsData, only allocated when needed instead of on every pe.
Gengbin Zheng [Thu, 8 Apr 2010 20:04:45 +0000 (15:04 -0500)]
do the same thing to statsData, only allocated when needed instead of on every pe.

11 years agoremoved PWD.
Gengbin Zheng [Thu, 8 Apr 2010 19:58:54 +0000 (14:58 -0500)]
removed PWD.

11 years agoinstead of allocate space of statsMsgsList in constructor (which is on all pes),...
Gengbin Zheng [Thu, 8 Apr 2010 19:14:40 +0000 (14:14 -0500)]
instead of allocate space of statsMsgsList in constructor (which is on all pes), only allocate it when needed (typically just on the central node). This removes the p^2 memory usage for bigsim.

11 years agogetObjTime() now gets the current timing from database, instead of the last known...
Gengbin Zheng [Tue, 6 Apr 2010 23:05:31 +0000 (18:05 -0500)]
getObjTime() now gets the current timing from database, instead of the last known (one saved from last lb)

11 years agockcomplex: Does not really need conv-mach.h
Ramprasad Venkataraman [Tue, 6 Apr 2010 22:25:24 +0000 (17:25 -0500)]
ckcomplex: Does not really need conv-mach.h

Vestigial include from ye olde days when ckcomplex had to worry
about which mallocs, memaligns etc. Refer 5b8df164.

11 years agockcomplex: delete old ifdef cruft
Ramprasad Venkataraman [Tue, 6 Apr 2010 21:41:51 +0000 (16:41 -0500)]
ckcomplex: delete old ifdef cruft

Ancient versions of openatom (and other programs?) used the type ckcomplex directly
Hence there was ifdef manipulation to keep them happy. However ckcomplex typedef-ed
to complex has been the default for ~4 years. Hence delete this old clutter.

note: There will (soon) be a need again to revert to a distinct charm complex offering
as complex becomes part of std libs: ck::complex?

11 years agofixed a typo for CmiMyPe()
Gengbin Zheng [Tue, 6 Apr 2010 13:21:10 +0000 (08:21 -0500)]
fixed a typo for CmiMyPe()

via git-CVS emulator

11 years agoadded -I. to compile controlPointsNoTrace.C. (windows needs unistd.h which is faked...
Gengbin Zheng [Tue, 6 Apr 2010 06:07:12 +0000 (01:07 -0500)]
added -I. to compile controlPointsNoTrace.C. (windows needs unistd.h  which is faked at charm/tmp)

11 years agofixed getObjTime to actually get obj walltime.
Gengbin Zheng [Tue, 6 Apr 2010 04:27:31 +0000 (23:27 -0500)]
fixed getObjTime to actually get obj walltime.

11 years agoalways define ConverDeliver() function in case it is needed (for example, some part...
Gengbin Zheng [Tue, 6 Apr 2010 04:12:44 +0000 (23:12 -0500)]
always define ConverDeliver() function in case it is needed (for example, some part of code is compiled without CMK_OPTIMIZE).

11 years agoRevert "always define ConverDeliver() function in case it is needed (for example...
Gengbin Zheng [Tue, 6 Apr 2010 04:10:27 +0000 (23:10 -0500)]
Revert "always define ConverDeliver() function in case it is needed (for example, some part of code is compiled without CMK_OPTIMIZE)."

This reverts commit 736919f969853865e782513d01ec4d196245c94c.

11 years agoalways define ConverDeliver() function in case it is needed (for example, some part...
Gengbin Zheng [Tue, 6 Apr 2010 04:08:52 +0000 (23:08 -0500)]
always define ConverDeliver() function in case it is needed (for example, some part of code is compiled without CMK_OPTIMIZE).

11 years agoFixed because of capitalization problem with controlPointsNoTrace in Makefile.
Jonathan Lifflander [Tue, 6 Apr 2010 00:43:45 +0000 (19:43 -0500)]
Fixed because of capitalization problem with controlPointsNoTrace in Makefile.

11 years agoAdding a ControlPointsNoTrace module that provides the control point interface but...
Isaac Dooley [Mon, 5 Apr 2010 20:20:43 +0000 (15:20 -0500)]
Adding a ControlPointsNoTrace module that provides the control point interface but doesn't need to use the controlPoints tracing module.

11 years agofixed _validProcessors declaraction.
Gengbin Zheng [Mon, 5 Apr 2010 13:34:49 +0000 (08:34 -0500)]
fixed _validProcessors declaraction.

via git-CVS emulator

11 years agominor change to print out the solutions in satisfiability program
Yanhua YanhuaSunLaptop [Sun, 4 Apr 2010 06:03:43 +0000 (01:03 -0500)]
minor change to print out the solutions in satisfiability program

11 years agoMerge branch 'charm' of charmgit:charm into charm
Yanhua YanhuaSunLaptop [Sun, 4 Apr 2010 05:40:10 +0000 (00:40 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agoreplace ckvec with map in satisfiability program
Yanhua YanhuaSunLaptop [Sun, 4 Apr 2010 05:40:01 +0000 (00:40 -0500)]
replace ckvec with map in satisfiability program

11 years agoAdding command line argument for disabling control point filtering of output file
Isaac Dooley [Sat, 3 Apr 2010 19:54:47 +0000 (14:54 -0500)]
Adding command line argument for disabling control point filtering of output file

11 years agoMerge branch 'charm' of charmgit:charm into charm
Isaac Dooley [Sat, 3 Apr 2010 16:47:04 +0000 (11:47 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agocontrol point writing data file is more verbose.
Isaac Dooley [Sat, 3 Apr 2010 16:46:54 +0000 (11:46 -0500)]
control point writing data file is more verbose.

11 years agoFixed POSE PUP bug: random seed in rep.h needs to be PUPed
Ryan Mokos [Fri, 2 Apr 2010 19:44:07 +0000 (14:44 -0500)]
Fixed POSE PUP bug: random seed in rep.h needs to be PUPed

11 years agoFixed bug with deterministic events in POSE (+deterministic_pose)
Ryan Mokos [Fri, 2 Apr 2010 19:04:56 +0000 (14:04 -0500)]
Fixed bug with deterministic events in POSE (+deterministic_pose)

11 years agosatisfiability can be run as sequential in order to get speed up
Yanhua YanhuaSunLaptop [Fri, 2 Apr 2010 17:53:15 +0000 (12:53 -0500)]
satisfiability can be run as sequential in order to get speed up

11 years agomove the declaration of _validProcessors from init.C to convcore.c, because otherwise...
Gengbin Zheng [Fri, 2 Apr 2010 17:20:40 +0000 (12:20 -0500)]
move the declaration of _validProcessors from init.C to convcore.c, because otherwise on some system a pure converse program can not be compiled due to undef'ed variable.

11 years agoMerge branch 'charm' of charmgit:charm into charm
Gengbin Zheng [Fri, 2 Apr 2010 17:19:21 +0000 (12:19 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agomonir fixes for smp on bluebiou
Gengbin Zheng [Fri, 2 Apr 2010 17:18:41 +0000 (12:18 -0500)]
monir fixes for smp on bluebiou

11 years agoAdding externally callable write data to disk function. For use in NAMD where shutdow...
Isaac Dooley [Fri, 2 Apr 2010 16:23:18 +0000 (11:23 -0500)]
Adding externally callable write data to disk function. For use in NAMD where shutdown is not done properly.

11 years agomake sure charmxi is first built. this seems to fix the broken headerlinks rule,...
Gengbin Zheng [Fri, 2 Apr 2010 02:46:52 +0000 (21:46 -0500)]
make sure charmxi is first built. this seems to fix the broken headerlinks rule, where ci files are not all built at once.

11 years agoMerge branch 'charm' of charmgit:charm into charm
Eric Bohm [Thu, 1 Apr 2010 22:26:54 +0000 (17:26 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agoauto translate chpt<stuff> to rep for pups as else clause for ifndef SEQUENTIAL
Eric Bohm [Thu, 1 Apr 2010 22:25:38 +0000 (17:25 -0500)]
auto translate chpt<stuff> to rep for pups as else clause for ifndef SEQUENTIAL

11 years agoin isomalloc, use getpagesize to determin page size instead of using CMK_MEMORY_PAGES...
Gengbin Zheng [Thu, 1 Apr 2010 21:59:52 +0000 (16:59 -0500)]
in isomalloc, use getpagesize to determin page size instead of using CMK_MEMORY_PAGESIZE defined in conv-mach.
getpagesize is tested in configure.

11 years agoBuild: Include optimization as part of --with-production
Phil Miller [Thu, 1 Apr 2010 20:33:58 +0000 (15:33 -0500)]
Build: Include optimization as part of --with-production

11 years agoMerge branch 'charm' of charmgit:charm into charm
Gengbin Zheng [Thu, 1 Apr 2010 20:47:54 +0000 (15:47 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agoallow ibverbs to run in standalone mode.
Gengbin Zheng [Thu, 1 Apr 2010 20:47:24 +0000 (15:47 -0500)]
allow ibverbs to run in standalone mode.

11 years agoBGP: Use make variables instead of repeating file names
Phil Miller [Thu, 1 Apr 2010 16:30:39 +0000 (11:30 -0500)]
BGP: Use make variables instead of repeating file names

11 years agoBGP: Compile trace-controlPoints-BGP.a with optimization off
Phil Miller [Thu, 1 Apr 2010 16:19:06 +0000 (11:19 -0500)]
BGP: Compile trace-controlPoints-BGP.a with optimization off

Passing xlc optimization flags to g++ makes a mess. Specify -no-optimize on
the one file that forces g++ usage so that this doesn't happen.

11 years agoBGP: Remove extraneous -g for control points object
Phil Miller [Thu, 1 Apr 2010 16:17:59 +0000 (11:17 -0500)]
BGP: Remove extraneous -g for control points object

11 years agoMake charmc -no-optimize actually work
Phil Miller [Thu, 1 Apr 2010 16:14:23 +0000 (11:14 -0500)]
Make charmc -no-optimize actually work

Charmc would act as if -optimize were passed if either of that or -no-optimize
actually appeared on the command line, since both set $OPTIMIZE_MODE to a
non-null value. Make charmc actually check that it's set affirmatively.

11 years agoDon't pass the '-optimize' flag to a hard-coded g++
Phil Miller [Thu, 1 Apr 2010 15:43:24 +0000 (10:43 -0500)]
Don't pass the '-optimize' flag to a hard-coded g++

The charmc compiler wrapper switched from expecting -O to -optimize for the
meaning "pass the best known compiler optimization flags to the underlying
compiler." This was written on the build command line and passed through make
in $OPTS. That works fine as long as charmc gets to rewrite it, but causes
problems when passed directly to a real compiler, like g++.

The file trace-controlPoints-BPG.C needs to be compiled with g++ and not xlC,
and this need was encoded as an explicit invocation of g++ in the BGP-specific
Makefile. Thus, we passed -optimize to it, and got an object file named
'ptimize'. Get this right by using charmc's '-c++' compiler override option

11 years agoadded a new feature to charmrun so that it works with mpiexec now.
Gengbin Zheng [Thu, 1 Apr 2010 15:11:25 +0000 (10:11 -0500)]
added a new feature to charmrun so that it works with mpiexec now.
++mpiexec new command line option is added for tis purpose.

11 years agofixed a string to integer comparison.
Gengbin Zheng [Thu, 1 Apr 2010 15:01:39 +0000 (10:01 -0500)]
fixed a string to integer comparison.

11 years agoMerge branch 'charm' of charmgit:charm into charm
Gengbin Zheng [Thu, 1 Apr 2010 04:40:13 +0000 (23:40 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agochange mem checkpoint so that when checkpointing, the buddy that is likely to host...
Gengbin Zheng [Thu, 1 Apr 2010 04:38:11 +0000 (23:38 -0500)]
change mem checkpoint so that when checkpointing, the buddy that is likely to host the array element will notify checkpointing. Make DoItNow() an expedited entry method.

11 years agoChangelog: Copy over from 6.2
Phil Miller [Wed, 31 Mar 2010 21:31:15 +0000 (16:31 -0500)]
Changelog: Copy over from 6.2

11 years agocall resumeAfterLB instead resumeAfterCheckpoint after load balancing.
Gengbin Zheng [Wed, 31 Mar 2010 21:21:51 +0000 (16:21 -0500)]
call resumeAfterLB instead resumeAfterCheckpoint after load balancing.

11 years agoremoving prepending `pwd` to the executable, which does not work if already using...
Gengbin Zheng [Wed, 31 Mar 2010 07:00:58 +0000 (02:00 -0500)]
removing prepending `pwd` to the executable, which does not work if already using absolute path.

11 years agoredo the previous fix.
Gengbin Zheng [Wed, 31 Mar 2010 06:16:18 +0000 (01:16 -0500)]
redo the previous fix.

11 years agoMerge branch 'charm' of charmgit:charm into charm
Gengbin Zheng [Wed, 31 Mar 2010 06:13:11 +0000 (01:13 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agofixed an error in testing MPI. BUILD_MPI may be an empty string.
Gengbin Zheng [Wed, 31 Mar 2010 06:12:30 +0000 (01:12 -0500)]
fixed an error in testing MPI. BUILD_MPI may be an empty string.

11 years agofixed compiler error in VC++.
U-HOME2\Gengbin [Wed, 31 Mar 2010 05:02:42 +0000 (00:02 -0500)]
fixed compiler error in VC++.

11 years agoAdded documentation for the new POSE checkpoint command line parameters to the POSE...
Ryan Mokos [Wed, 31 Mar 2010 03:15:25 +0000 (22:15 -0500)]
Added documentation for the new POSE checkpoint command line parameters to the POSE manual.

11 years agomodify the pose array creation to work with load balancer.
Gengbin Zheng [Wed, 31 Mar 2010 03:09:57 +0000 (22:09 -0500)]
modify the pose array creation to work with load balancer.

11 years agoadded a command line option +lb_gvt_pose to set the gvt intervals between load balancing.
Gengbin Zheng [Wed, 31 Mar 2010 03:07:46 +0000 (22:07 -0500)]
added a command line option +lb_gvt_pose to set the gvt intervals between load balancing.
modified to set usesAtSync in sim array.

11 years agoadded basic support for load balancing in pose (using the general LB framework).
Gengbin Zheng [Wed, 31 Mar 2010 02:41:26 +0000 (21:41 -0500)]
added basic support for load balancing in pose (using the general LB framework).

11 years agoChanged POSE checkpointing so it's activated with command-line parameters. Also...
Ryan Mokos [Tue, 30 Mar 2010 21:30:18 +0000 (16:30 -0500)]
Changed POSE checkpointing so it's activated with command-line parameters.  Also added POSE checkpointing based on time.

11 years agocontrolPoints: Explicitly include climits for INT_MAX
Ramprasad Venkataraman [Tue, 30 Mar 2010 04:38:41 +0000 (23:38 -0500)]
controlPoints: Explicitly include climits for INT_MAX

g++ 4.4 seems to have cleaned up their internal includes which results
in unrecognized INT_MAX if climits is not included explicitly by user code

Tested in net-linux with gcc 4.4

11 years agoAdding warning to Envelope mtype field
Filippo Gioachin [Tue, 30 Mar 2010 03:04:14 +0000 (22:04 -0500)]
Adding warning to Envelope mtype field

11 years agofix the problem that the binary of parsat_tnm is too large in satisifiablity
Yanhua YanhuaSunLaptop [Tue, 30 Mar 2010 02:52:38 +0000 (21:52 -0500)]
fix the problem that the binary of parsat_tnm is too large in satisifiablity

11 years agoUpdating to an API change.
Isaac Dooley [Mon, 29 Mar 2010 20:01:01 +0000 (15:01 -0500)]
Updating to an API change.

11 years agoa new function CmiPrintCPUAffinty() that can be called on any processor to print...
Gengbin Zheng [Mon, 29 Mar 2010 15:42:05 +0000 (10:42 -0500)]
a new function CmiPrintCPUAffinty() that can be called on any processor to print current cpuaffinity.
added a new option:  +showcpuaffinity to print current cpuaffinity after setting cpuaffinity.
support affinity notation of e.g. -1, which means the CmiNumCores()-1, or the last core on a node, similarly for -2, -3, and so on.

11 years ago fix some mistakes in TNM
Yanhua YanhuaSunLaptop [Mon, 29 Mar 2010 05:57:03 +0000 (00:57 -0500)]
 fix some mistakes in TNM

11 years agoadd minisat and TNM sequential solvers in sat problem
Yanhua YanhuaSunLaptop [Mon, 29 Mar 2010 03:05:44 +0000 (22:05 -0500)]
add minisat and TNM sequential solvers in sat problem

11 years agoone charm parallel two sequential SAT solver
Yanhua YanhuaSunLaptop [Mon, 29 Mar 2010 03:00:36 +0000 (22:00 -0500)]
one charm parallel two sequential SAT solver

11 years agorename charm related files into par_* to distiguish the sequential code in satisfiabi...
Yanhua YanhuaSunLaptop [Sun, 28 Mar 2010 19:17:50 +0000 (14:17 -0500)]
rename charm related files into par_* to distiguish the sequential code in satisfiability problem

11 years agoMerge branch 'charm' of charmgit:charm into charm
Yanhua YanhuaSunLaptop [Sun, 28 Mar 2010 19:08:34 +0000 (14:08 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agoport the minisat sequential part into charm parallel satisfiablity problem
Yanhua YanhuaSunLaptop [Sun, 28 Mar 2010 19:08:18 +0000 (14:08 -0500)]
port the minisat sequential part into charm parallel satisfiablity problem

11 years agofix for -pedantic
Gengbin Zheng [Sat, 27 Mar 2010 01:41:33 +0000 (20:41 -0500)]
fix for -pedantic

11 years agorenamed Set.h to ckset.h
Gengbin Zheng [Sat, 27 Mar 2010 01:37:18 +0000 (20:37 -0500)]
renamed Set.h to ckset.h

11 years agoMerge branch 'charm' of charmgit:charm into charm
Gengbin Zheng [Sat, 27 Mar 2010 00:24:48 +0000 (19:24 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agorenamed heap.h to ckheap.h to avoid using too general file name.
Gengbin Zheng [Sat, 27 Mar 2010 00:23:42 +0000 (19:23 -0500)]
renamed heap.h to ckheap.h to avoid using too general file name.

11 years agoMerge branch 'charm' of charmgit:charm into charm
Ryan Mokos [Fri, 26 Mar 2010 23:24:02 +0000 (18:24 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agoAdded checkpointing to POSE. Also added a couple tests to the Ring Makefile in charm...
Ryan Mokos [Fri, 26 Mar 2010 23:21:52 +0000 (18:21 -0500)]
Added checkpointing to POSE.  Also added a couple tests to the Ring Makefile in charm/examples/pose/Ring for easy checking of this ability.

11 years agoAdding an always use defaults option to tuning framework.
Isaac Dooley [Fri, 26 Mar 2010 19:02:41 +0000 (14:02 -0500)]
Adding an always use defaults option to tuning framework.

11 years agoVC++ has _snprintf, instead of snprintf
Gengbin Zheng [Fri, 26 Mar 2010 13:25:36 +0000 (08:25 -0500)]
VC++ has _snprintf, instead of snprintf

via git-CVS emulator

11 years agoibverbs for net-linux-ppc
Gengbin Zheng [Fri, 26 Mar 2010 05:16:56 +0000 (00:16 -0500)]
ibverbs for net-linux-ppc

11 years agoMerge branch 'charm' of charmgit:charm into charm
Gengbin Zheng [Fri, 26 Mar 2010 05:13:03 +0000 (00:13 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agoadded support for avail_vector, so it can be used with FT.
Gengbin Zheng [Fri, 26 Mar 2010 05:11:36 +0000 (00:11 -0500)]
added support for avail_vector, so it can be used with FT.
Won't assign work to the failed processors.

11 years agorenamed old directories as *.old.nobackup so backup system ignores them
Filippo Gioachin [Fri, 26 Mar 2010 00:51:06 +0000 (19:51 -0500)]
renamed old directories as *.old.nobackup so backup system ignores them
ignoring a few errors while making

11 years agothe temp buf does not need to be that big now for printing only ints and floats,...
Gengbin Zheng [Thu, 25 Mar 2010 22:59:24 +0000 (17:59 -0500)]
the temp buf does not need to be that big now for printing only ints and floats, and string has been treated separately in the previous commit. Still, a buffer overflow checking is added.

11 years agoAdding expedited to performance gathering entry methods in tuning framework.
Isaac Dooley [Thu, 25 Mar 2010 22:45:31 +0000 (17:45 -0500)]
Adding expedited to performance gathering entry methods in tuning framework.

11 years agoOrion's change to fix ckstream when a temp print buffer overflow for printing strings.
Gengbin Zheng [Thu, 25 Mar 2010 21:40:40 +0000 (16:40 -0500)]
Orion's change to fix ckstream when a temp print buffer overflow for printing strings.

11 years agoMake pup definitions visible to ckcomplex
Ramprasad Venkataraman [Thu, 25 Mar 2010 19:38:21 +0000 (14:38 -0500)]
Make pup definitions visible to ckcomplex

Including ckcomplex shouldnt really force user code to supply charm++/pup.h

11 years agoEmit scoped names for messages inside a namespace
Aaron Becker [Thu, 25 Mar 2010 16:31:50 +0000 (11:31 -0500)]
Emit scoped names for messages inside a namespace

11 years agoAdding a timer to determine time spent generating control point configuration plans.
Isaac Dooley [Thu, 25 Mar 2010 16:30:57 +0000 (11:30 -0500)]
Adding a timer to determine time spent generating control point configuration plans.

11 years agoonly test MPI build when CMK_BUILD_MPI is set in
Gengbin Zheng [Thu, 25 Mar 2010 13:40:45 +0000 (08:40 -0500)]
only test MPI build when CMK_BUILD_MPI is set in

11 years agoadded some dummy stuff so that it is not completely empty, something xlc compiler...
Gengbin Zheng [Thu, 25 Mar 2010 06:14:57 +0000 (01:14 -0500)]
added some dummy stuff so that it is not completely empty, something xlc compiler would complain about.

11 years agodon't explicitly specify at link time to use -lmpi or -lmpich,
Gengbin Zheng [Thu, 25 Mar 2010 06:09:39 +0000 (01:09 -0500)]
don't explicitly specify at link time to use -lmpi or -lmpich,
let charmconfig test it.

11 years agofixed CMK_LDXX to be CMK_CXX
Gengbin Zheng [Thu, 25 Mar 2010 05:14:10 +0000 (00:14 -0500)]
fixed CMK_LDXX to be CMK_CXX

11 years agowhen compiler test shows it is 64 bit compiler, and CMK_64BIT is not defined, define it.
Gengbin Zheng [Thu, 25 Mar 2010 05:12:02 +0000 (00:12 -0500)]
when compiler test shows it is 64 bit compiler, and CMK_64BIT is not defined, define it.

11 years agoFixing a bug in the divide & conquer steering scheme.
Isaac Dooley [Wed, 24 Mar 2010 20:13:18 +0000 (15:13 -0500)]
Fixing a bug in the divide & conquer steering scheme.

11 years agoMerge branch 'charm' of charmgit:charm into charm
Yanhua YanhuaSunLaptop [Wed, 24 Mar 2010 05:01:16 +0000 (00:01 -0500)]
Merge branch 'charm' of charmgit:charm into charm

11 years agofix a bug that occur in sequential part in satisfiability problem
Yanhua YanhuaSunLaptop [Wed, 24 Mar 2010 05:00:48 +0000 (00:00 -0500)]
fix a bug that occur in sequential part in satisfiability problem

11 years agoreplaced strdup (which is a problem on aix) by malloc and strcpy
Gengbin Zheng [Wed, 24 Mar 2010 04:00:02 +0000 (23:00 -0500)]
replaced strdup (which is a problem on aix) by malloc and strcpy

11 years agoMerge branch 'charm' of charmgit:charm into charm
Yanhua Yanhua [Wed, 24 Mar 2010 01:36:36 +0000 (20:36 -0500)]
Merge branch 'charm' of charmgit:charm into charm