charm.git
3 years agoUpdate changelog for release v6.5.1
Phil Miller [Tue, 2 Jul 2013 00:00:47 +0000 (19:00 -0500)]
Update changelog for release

3 years agompi-bluegeneq: look for indications of xlc build in base directory, no matter where...
Phil Miller [Wed, 26 Jun 2013 19:49:22 +0000 (19:49 +0000)]
mpi-bluegeneq: look for indications of xlc build in base directory, no matter where charmc is run

3 years agoBGQ: Move machine-specific topology code out of a public header, to reduce include...
Phil Miller [Tue, 25 Jun 2013 19:33:47 +0000 (19:33 +0000)]
BGQ: Move machine-specific topology code out of a public header, to reduce include pollution

3 years agoBGQ: Add implied header include explicitly
Phil Miller [Tue, 25 Jun 2013 19:31:12 +0000 (19:31 +0000)]
BGQ: Add implied header include explicitly

3 years agoconfig: add check for presence of mpi wrappers v6.5.1-rc2
Nikhil Jain [Tue, 25 Jun 2013 04:50:53 +0000 (04:50 +0000)]
config: add check for presence of mpi wrappers

(cherry picked from commit 9b1de6decea7dccf992f3690261892b6b2f32682)

3 years agomachine layer: configuration changes for BGQ
Nikhil Jain [Tue, 25 Jun 2013 04:22:52 +0000 (04:22 +0000)]
machine layer: configuration changes for BGQ

(cherry picked from commit 122042ff42359b56c4f50f8e1f3c6ae995f5d9d5)

3 years agostart up: pup read only data/msgs only for the zeroeth rank
Nikhil Jain [Wed, 5 Jun 2013 19:07:26 +0000 (19:07 +0000)]
start up: pup read only data/msgs only for the zeroeth rank

(cherry-picked from commit 4ee3b52e9d1)

4 years agochange to use sync() when available.
Gengbin Zheng [Wed, 29 May 2013 21:30:25 +0000 (16:30 -0500)]
change to use sync() when available.

4 years agoadd test for function sync()
Gengbin Zheng [Wed, 29 May 2013 21:23:10 +0000 (16:23 -0500)]
add test for function sync()

4 years agoa bug fix in disk-based restart under SMP. initDone can be called more than once...
Gengbin Zheng [Fri, 21 Jun 2013 20:37:37 +0000 (15:37 -0500)]
a bug fix in disk-based restart under SMP. initDone can be called more than once on many ranks, which caused CmiNodeBarrier() to hang the scheduler on some ranks.

4 years agoa bug in broadcast from a comm thread under SMP. PE number of CkNumPes() is also...
Gengbin Zheng [Fri, 21 Jun 2013 20:35:05 +0000 (15:35 -0500)]
a bug in broadcast from a comm thread under SMP. PE number of CkNumPes() is also a comm thread.

4 years agoRe-enable examples/bigsim/emulator, after fixing issue #233 v6.5.1-rc1
Phil Miller [Tue, 18 Jun 2013 22:44:58 +0000 (17:44 -0500)]
Re-enable examples/bigsim/emulator, after fixing issue #233

4 years agoIssue #161, Issue #233: Grab thread oversubscription behavior flags on all thread...
Phil Miller [Tue, 18 Jun 2013 22:42:21 +0000 (17:42 -0500)]
Issue #161, Issue #233: Grab thread oversubscription behavior flags on all thread ranks

4 years agoDisable recognized test failures to get a clearer picture of the remainder
Phil Miller [Tue, 18 Jun 2013 02:07:09 +0000 (21:07 -0500)]
Disable recognized test failures to get a clearer picture of the remainder

4 years agoIssue #158: Handle all types of non-checkpointable callbacks
Phil Miller [Sun, 16 Jun 2013 21:47:46 +0000 (16:47 -0500)]
Issue #158: Handle all types of non-checkpointable callbacks

4 years agonot set callback to plain chare when pointer is used
Xiang Ni [Thu, 6 Jun 2013 06:31:53 +0000 (01:31 -0500)]
not set callback to plain chare when pointer is used

4 years agoFT: add validation for callback
Xiang Ni [Wed, 5 Jun 2013 06:46:51 +0000 (01:46 -0500)]
FT: add validation for callback

4 years agobugfix: add a hack to get the correct thread count on BGQ
Nikhil Jain [Wed, 12 Jun 2013 06:31:21 +0000 (06:31 +0000)]
bugfix: add a hack to get the correct thread count on BGQ

4 years agostartup: provide more information
Nikhil Jain [Wed, 12 Jun 2013 04:56:57 +0000 (21:56 -0700)]
startup: provide more information

4 years agoRevert "bugfix: forward declaration should match the definition"
Nikhil Jain [Wed, 12 Jun 2013 04:42:39 +0000 (21:42 -0700)]
Revert "bugfix: forward declaration should match the definition"

This reverts commit d95f6446dd1d1021af9ca6138bb72160add34501.

4 years agoTopoManager: return absolute coordinates for cray machines
Nikhil Jain [Wed, 12 Jun 2013 04:35:35 +0000 (21:35 -0700)]
TopoManager: return absolute coordinates for cray machines

Conflicts:

src/util/XTTorus.h

4 years agobugfix: forward declaration should match the definition
Nikhil Jain [Wed, 12 Jun 2013 03:23:27 +0000 (20:23 -0700)]
bugfix: forward declaration should match the definition

4 years agoIssue #161 - Be explicit in handling of threads oversubscribing cores
Phil Miller [Tue, 4 Jun 2013 01:30:00 +0000 (20:30 -0500)]
Issue #161 - Be explicit in handling of threads oversubscribing cores

Add flags for explicitly spinning or sleeping when idle. If the RTS
autonomously chooses to sleep due to oversubscription, print a warning. Also
warn if users pass the older flag.

Note that this is another instance of the uniform (physical and logical) node
size assumption, pre-existing from 5c69d4462be87c41b06a4ab99db968bcba515fa40.

4 years agoIssue #161 - Rename _Cmi_noprocforcommthread to _Cmi_sleepOnIdle
Phil Miller [Mon, 3 Jun 2013 22:59:01 +0000 (17:59 -0500)]
Issue #161 - Rename _Cmi_noprocforcommthread to _Cmi_sleepOnIdle

4 years agofix +CmiNoProcForComThread warning by enable it when ppn is larger than physical...
Yanhua Sun [Sat, 4 May 2013 03:46:05 +0000 (22:46 -0500)]
fix +CmiNoProcForComThread warning by enable it when ppn is larger than physical cores

(cherry picked from commit 3723b6fd666544f810c1d2177773d558b547bd7c)

4 years agoIssue #104 - BGQ xlC workaround: Add -qsmp=noostls flag for MPI builds, too
Phil Miller [Fri, 7 Jun 2013 21:00:52 +0000 (16:00 -0500)]
Issue #104 - BGQ xlC workaround: Add -qsmp=noostls flag for MPI builds, too

4 years agobgq compiler option: add -qsmp=noostls to avoid compilation failure with -g
Nikhil Jain [Fri, 7 Jun 2013 06:15:15 +0000 (06:15 +0000)]
bgq compiler option: add -qsmp=noostls to avoid compilation failure with -g

4 years agomempool: bug fix to enable handle allocations till 1 GB
Nikhil Jain [Fri, 7 Jun 2013 01:36:03 +0000 (21:36 -0400)]
mempool: bug fix to enable handle allocations till 1 GB

4 years agoTopoManager: Enable nidtopid call in Constructor
Nikhil Jain [Mon, 25 Mar 2013 07:03:09 +0000 (02:03 -0500)]
TopoManager: Enable nidtopid call in Constructor

4 years agoTopoManager- drop pidtonid creation by default init for Cray
Nikhil Jain [Mon, 25 Mar 2013 07:01:08 +0000 (02:01 -0500)]
TopoManager- drop pidtonid creation by default init for Cray

4 years agoTopoManager: fix Cray API bug; table needs to be created for all nodes
Nikhil Jain [Thu, 6 Jun 2013 20:21:32 +0000 (16:21 -0400)]
TopoManager: fix Cray API bug; table needs to be created for all nodes

4 years agoIssue #118 - Examples: Drop FEM testing, since we're deprecating it
Phil Miller [Tue, 4 Jun 2013 18:57:54 +0000 (13:57 -0500)]
Issue #118 - Examples: Drop FEM testing, since we're deprecating it

4 years agoIssue #191: Note deprecation of CmiBool typedefs
Phil Miller [Mon, 3 Jun 2013 22:38:22 +0000 (17:38 -0500)]
Issue #191: Note deprecation of CmiBool typedefs

4 years agoIssue #118 - Tests: Drop FEM testing, since we're deprecating it
Phil Miller [Mon, 3 Jun 2013 22:29:06 +0000 (17:29 -0500)]
Issue #118 - Tests: Drop FEM testing, since we're deprecating it

4 years agoClear up craycc issues:
Eric Bohm [Fri, 17 May 2013 22:49:45 +0000 (17:49 -0500)]
Clear up craycc issues:
1. split tests for sync_synchronize and fetch_add
2. split implementation selection for CmiMemory*Fence from CmiMemoryAtomic* to use whichever underlying intrinsic support is available based
on the newly split configure test
3. Add test and support for use of __builtin_ia32_lfence() so that craycc builds can leverage native intrinsics for CmiMemory*Fence
4. fixed typo in configure.in

With these changes in place, AMPI builds and passes make test when using craycc.

(cherry picked from commit 92c60d3c5536df2aacedd92f04b3deb338d3c3cc, omitted extraneous envelope.h change that doesn't apply to the older branch)

4 years agoadd test to reject CCE 8.1.4
Eric Bohm [Thu, 9 May 2013 20:27:33 +0000 (15:27 -0500)]
add test to reject CCE 8.1.4
(cherry picked from commit fffdb85027d221259cf129702773b37a052d6134)

4 years agoPUP-STL: add forward decls for pup for fix clang build
Jonathan Lifflander [Wed, 29 May 2013 19:44:33 +0000 (14:44 -0500)]
PUP-STL: add forward decls for pup for fix clang build
(cherry picked from commit 95a6e29dba09827ce153f4d52a8a05a120e59e20)

4 years agoCentralLB: don't leak stats count message (fixes #187)
Phil Miller [Tue, 23 Apr 2013 21:27:18 +0000 (16:27 -0500)]
CentralLB: don't leak stats count message (fixes #187)

4 years agozlib on mpi-bgq
Yanhua Sun [Thu, 18 Apr 2013 20:34:13 +0000 (20:34 +0000)]
zlib on mpi-bgq
(cherry picked from commit 7d27753bfeea2e8a955138822deb27966bf33f49)

4 years agofix the ZLIB path problem on BGQ
Yanhua Sun [Thu, 18 Apr 2013 20:07:40 +0000 (20:07 +0000)]
fix the ZLIB path problem on BGQ
(cherry picked from commit df206d1897961b913c312ced89da23aee7841c32)

4 years agoexamples/charm++/leanmd: Fix mismatched new[]/delete
Phil Miller [Thu, 4 Apr 2013 14:40:22 +0000 (09:40 -0500)]
examples/charm++/leanmd: Fix mismatched new[]/delete

(cherry picked from commit 7c9cf21f2f70746c9732e7b9f87d4481b4f25db6)

4 years agoReference binary relative to ./ to appease Edison's aprun
Phil Miller [Sat, 30 Mar 2013 18:18:14 +0000 (11:18 -0700)]
Reference binary relative to ./ to appease Edison's aprun

4 years agotests: make more ci file processing safe for make -j
Phil Miller [Sat, 30 Mar 2013 17:10:57 +0000 (12:10 -0500)]
tests: make more ci file processing safe for make -j

4 years agoCrayXC: Add target gni-crayxc and mpi-crayxc for the new cray system based on
Nikhil Jain [Sun, 10 Mar 2013 01:04:39 +0000 (17:04 -0800)]
CrayXC: Add target gni-crayxc and mpi-crayxc for the new cray system based on
Aries interconnect. All the code is borrowed from CrayXE. In the process also
remove the GEMINI tag from CRAYXE build and call it GNI instead.

(cherry picked from commit 831051eed8beca089943238a70c0b36265739eae)

4 years agoPersistent communication: link in zlib automatically
Phil Miller [Fri, 29 Mar 2013 21:36:39 +0000 (16:36 -0500)]
Persistent communication: link in zlib automatically

(cherry picked from commit fbaa9813af3d7d6ec16d1c6955ad9d60de8be0bc)

4 years agoMPI w/ PGCC: Don't pass C++-specific compiler options to the C compiler
Phil Miller [Fri, 29 Mar 2013 20:12:49 +0000 (13:12 -0700)]
MPI w/ PGCC: Don't pass C++-specific compiler options to the C compiler

4 years agoMPI PGCC: Don't pass PGI-specific flags to the native GNU compilers
Phil Miller [Fri, 29 Mar 2013 19:42:05 +0000 (14:42 -0500)]
MPI PGCC: Don't pass PGI-specific flags to the native GNU compilers
(cherry picked from commit f0bc4a173563db34eae22c53fc688f8bf3b86b42)

4 years agoTopoManager: fix for Titan inconsistent value for dims and NID
Phil Miller [Thu, 28 Mar 2013 02:07:01 +0000 (21:07 -0500)]
TopoManager: fix for Titan inconsistent value for dims and NID

(cherry-picked from commit 9bda4c8a272d9badc7f0c7dec718e8095b53776a)

4 years agoAlign conditional function & variable definition with their usage
Phil Miller [Wed, 27 Mar 2013 21:45:25 +0000 (16:45 -0500)]
Align conditional function & variable definition with their usage

(cherry picked from commit 96a08a1ba8f5eea783639ee89e52f881a0ea3ebb)

4 years agoBug #89: Disable test that disregards cross-compilation to allow rest of testing v6.5.0
Phil Miller [Tue, 12 Mar 2013 18:20:24 +0000 (13:20 -0500)]
Bug #89: Disable test that disregards cross-compilation to allow rest of testing

4 years agoBug #87: Restore NodeGroups by node ID, not PE id
Phil Miller [Mon, 11 Mar 2013 23:19:22 +0000 (18:19 -0500)]
Bug #87: Restore NodeGroups by node ID, not PE id

4 years agoSlightly modifying the QD helper functions that allow us to avoid including some...
Lukasz Wesolowski [Mon, 11 Mar 2013 03:56:19 +0000 (22:56 -0500)]
Slightly modifying the QD helper functions that allow us to avoid including some of the Charm++ header files in GPU Manager.

4 years agodoc: adding redirection text for partition
Nikhil Jain [Sun, 10 Mar 2013 03:54:50 +0000 (21:54 -0600)]
doc: adding redirection text for partition

4 years agoAdding cuda option for building GPU Manager on gemini systems
Lukasz Wesolowski [Sun, 10 Mar 2013 03:22:57 +0000 (21:22 -0600)]
Adding cuda option for building GPU Manager on gemini systems
and fixing build errors.

4 years agoCrayXE: move charmrun from the source folder to machine target
Nikhil Jain [Sat, 9 Mar 2013 23:12:41 +0000 (23:12 +0000)]
CrayXE: move charmrun from the source folder to machine target

4 years agoBGP: force compilation with dec2011 compilers, else exit
Nikhil Jain [Fri, 8 Mar 2013 22:30:54 +0000 (22:30 +0000)]
BGP: force compilation with dec2011 compilers, else exit

4 years agoBGQ: remove the env from mpi charmrun
Nikhil Jain [Fri, 8 Mar 2013 18:52:16 +0000 (18:52 +0000)]
BGQ: remove the env from mpi charmrun

4 years agoRevert "BGQ: modify charmrun for BGQ to accept exit status 1 as a successful exit...
Nikhil Jain [Fri, 8 Mar 2013 11:37:44 +0000 (11:37 +0000)]
Revert "BGQ: modify charmrun for BGQ to accept exit status 1 as a successful exit as per"

This reverts commit cce8c68cd8df0c858c9023c7ac0c89db7ded09ae.

4 years agoBGP: bug fix for broken Manytomany operation - function arguments were being
Nikhil Jain [Fri, 8 Mar 2013 11:21:50 +0000 (11:21 +0000)]
BGP: bug fix for broken Manytomany operation - function arguments were being
passed in wrong order

4 years agoUse the new dynamic CPU set allocation API to support very large SMP machines
Jim Phillips [Fri, 8 Mar 2013 05:48:29 +0000 (23:48 -0600)]
Use the new dynamic CPU set allocation API to support very large SMP machines

In particular, allow a multicore binary built on Stampede to run on
Blacklight.  (It's not needed if you build on Blacklight, because the
default CPU set size in the headers has been increased on that
machine.) This should be safe, as it doesn't use the new interface
unless the default size is too small.

4 years agoIn cuda builds of charm, make sure int128 and int128_t are not used.
Lukasz Wesolowski [Fri, 8 Mar 2013 05:28:52 +0000 (23:28 -0600)]
In cuda builds of charm, make sure int128 and int128_t are not used.

4 years agoCHANGES: Update BGQ info to reflect multiple layers and their status
Phil Miller [Fri, 8 Mar 2013 02:29:04 +0000 (20:29 -0600)]
CHANGES: Update BGQ info to reflect multiple layers and their status

4 years agoCHANGES: typo
Phil Miller [Fri, 8 Mar 2013 02:28:18 +0000 (20:28 -0600)]
CHANGES: typo

4 years agoBGQ: compiler path changes to other BGQ builds
Nikhil Jain [Fri, 8 Mar 2013 00:00:52 +0000 (00:00 +0000)]
BGQ: compiler path changes to other BGQ builds

4 years agoBGQ: updating xlc paths - making them non-relative
Nikhil Jain [Thu, 7 Mar 2013 23:51:57 +0000 (23:51 +0000)]
BGQ: updating xlc paths - making them non-relative

4 years agoBGQ: optimizing the memalloc routine - done with it
Nikhil Jain [Thu, 7 Mar 2013 20:53:37 +0000 (20:53 +0000)]
BGQ: optimizing the memalloc routine - done with it

4 years agorednmgr: remove derived class overload of virtual flushStates() that hid base class...
Ramprasad Venkataraman [Wed, 6 Mar 2013 23:48:49 +0000 (17:48 -0600)]
rednmgr: remove derived class overload of virtual flushStates() that hid base class method

global and local element counters in CkReductionMgr need to be reset only when
the redn mgr is serving a chare array (and not when its serving a group). Implementing
this via an input flag to flushStates() caused flushStates() to become overloaded in
the child class. This overload hid the base class virtual method causing some compilers
(icc?) to complain (warnings). This somewhat cleaner solution should avoid the
warnings, but it still stinks of a hack.

4 years agoTopoManager: make topomanager node aware with partitioning
Nikhil Jain [Thu, 7 Mar 2013 00:03:21 +0000 (00:03 +0000)]
TopoManager: make topomanager node aware with partitioning

4 years agoAMPI: bug fix for mempool + isomalloc interaction - cannot free the mempool till
Nikhil Jain [Wed, 6 Mar 2013 18:30:56 +0000 (12:30 -0600)]
AMPI: bug fix for mempool + isomalloc interaction - cannot free the mempool till
the very end; infact there is not a need for freeing it, but keeping it

4 years agonet: keep byte and message counts in unsigned ints to avoid undefined overflow
Phil Miller [Wed, 6 Mar 2013 08:18:49 +0000 (02:18 -0600)]
net: keep byte and message counts in unsigned ints to avoid undefined overflow

Change the counts of sent and received messages and bytes from int to
unsigned int, to avoid the undefined behavior that occurs when the
eventually overflow. Incidentally, this doubles their useful range as
well.

Detected with clang -fsanitize=integer

4 years agoCkHashCode: return unmodified when rotating by 0 to avoid undefined behavior
Phil Miller [Wed, 6 Mar 2013 07:46:26 +0000 (01:46 -0600)]
CkHashCode: return unmodified when rotating by 0 to avoid undefined behavior

Shifting by more the full width of a variable is undefined behavior in
C and C++. When trying to rotate by 0, a shift of the full width
resulted. Avoid that by returning the unmodified argument when no
rotation is to be applied. The test should be free, either because the
value was just computed and so the condition code is set, or because
the compiler emits a rotate opcode anyway.

Error detected by
clang -fsanitize=integer -fno-sanitize=unsigned-integer-overflow

This leaves the entire execution of tests/charm++ clean of undefined
signed integer overflow.

4 years agoAvoid undefined behavior of left-shifting through the sign bit
Phil Miller [Wed, 6 Mar 2013 07:08:17 +0000 (01:08 -0600)]
Avoid undefined behavior of left-shifting through the sign bit

In C and C++, shifting left into or past the sign bit of a signed
integer results in undefined behavior. In cases where we're just
setting up a bit mask, use the bitwise inverse of an unsigned 0
instead, on which shifting is perfectly well defined.

Detected with
clang -fsanitize=integer -fno-sanitize=unsigned-integer-overflow

4 years agopup: Use unsigned int in hash computation to avoid undefined behavior
Phil Miller [Wed, 6 Mar 2013 06:52:48 +0000 (00:52 -0600)]
pup: Use unsigned int in hash computation to avoid undefined behavior

Overflowing a signed int is undefined behavior in C and C++, but the
hash code seems to assume 2's-complement wrap-around. That's exactly
the specified behavior of unsigned arithmetic, to use that explicitly.

4 years ago#62: A few more missing CkArgMsg parameters to mainchare constructors
Phil Miller [Tue, 5 Mar 2013 23:08:13 +0000 (17:08 -0600)]
#62: A few more missing CkArgMsg parameters to mainchare constructors

4 years agoBGQ: bug fix in async mode
Nikhil Jain [Tue, 5 Mar 2013 20:57:12 +0000 (20:57 +0000)]
BGQ: bug fix in async mode

4 years agolocmgr: nuke unused proxy var
Ramprasad Venkataraman [Tue, 5 Mar 2013 20:31:14 +0000 (14:31 -0600)]
locmgr: nuke unused proxy var

4 years agoMakefile: specify compilation of lz4.o independent of other objects and libraries
Phil Miller [Tue, 5 Mar 2013 20:08:00 +0000 (12:08 -0800)]
Makefile: specify compilation of lz4.o independent of other objects and libraries

4 years agobarnes-charm: fix compiler path; code is still broken/dependent on outside stuff
Phil Miller [Mon, 4 Mar 2013 23:55:36 +0000 (17:55 -0600)]
barnes-charm: fix compiler path; code is still broken/dependent on outside stuff

4 years ago#62: Missing CkArgMsg* for tests and examples that don't build
Phil Miller [Mon, 4 Mar 2013 23:54:53 +0000 (17:54 -0600)]
#62: Missing CkArgMsg* for tests and examples that don't build

4 years agoIssue #62: More missing explicit CkArgMsg* parameters
Phil Miller [Mon, 4 Mar 2013 23:53:21 +0000 (17:53 -0600)]
Issue #62: More missing explicit CkArgMsg* parameters

4 years agosimplest hello world example: fix syntax error
Phil Miller [Mon, 4 Mar 2013 23:41:40 +0000 (17:41 -0600)]
simplest hello world example: fix syntax error

4 years agoMPI has its own throttling built in and has its own API for this where
Eric Bohm [Mon, 4 Mar 2013 21:40:44 +0000 (15:40 -0600)]
MPI has its own throttling built in and has its own API for this where
available, so we disable this test on the MPI layer because it doesn't
visibly provide the gni_pub.h header and consequently the BI API.

4 years agoWe were never reading from the logs being written anyway, so there was no
Phil Miller [Mon, 4 Mar 2013 18:13:32 +0000 (12:13 -0600)]
We were never reading from the logs being written anyway, so there was no
reason for it. Since the flag causes crashes on Mac OS X (and potentially Blue
Gene) when generating compressed logs, drop it.

4 years agoBGQ: bug fix in optimized list send
Nikhil Jain [Mon, 4 Mar 2013 10:58:50 +0000 (10:58 +0000)]
BGQ: bug fix in optimized list send

4 years agoMakefile whitespace: line up target directories of gatherflat/tree
Phil Miller [Mon, 4 Mar 2013 05:55:00 +0000 (23:55 -0600)]
Makefile whitespace: line up target directories of gatherflat/tree

4 years agoBGQ: fixing a typo pointed by Phil in previous commit
Nikhil Jain [Mon, 4 Mar 2013 00:47:22 +0000 (00:47 +0000)]
BGQ: fixing a typo pointed by Phil in previous commit

4 years agoBGQ: clean exit - this seems to be working fine
Nikhil Jain [Mon, 4 Mar 2013 00:28:47 +0000 (00:28 +0000)]
BGQ: clean exit - this seems to be working fine

4 years agoBGQ: enabling and debugging optimized code for List Send
Nikhil Jain [Mon, 4 Mar 2013 00:27:47 +0000 (00:27 +0000)]
BGQ: enabling and debugging optimized code for List Send

4 years agoBGQ: change threads per context to make it consistent with pami
Nikhil Jain [Sun, 3 Mar 2013 19:01:05 +0000 (13:01 -0600)]
BGQ: change threads per context to make it consistent with pami

4 years ago#62: Add explicit CkArgMsg to example mainchare's constructor declaration, since...
Phil Miller [Sun, 3 Mar 2013 08:04:37 +0000 (02:04 -0600)]
#62: Add explicit CkArgMsg to example mainchare's constructor declaration, since its omission is deprecated

4 years agoadd gz-no-trace for generated regular log files (not compressed)
Yanhua Sun [Sun, 3 Mar 2013 06:00:47 +0000 (00:00 -0600)]
add gz-no-trace for generated regular log files (not compressed)

4 years agoFixing bug on QD with message-logging (reported by Jonathan).
Esteban Meneses [Sat, 2 Mar 2013 23:28:10 +0000 (17:28 -0600)]
Fixing bug on QD with message-logging (reported by Jonathan).

4 years agoBGQ: memalloc fix for PAMI LRTS
Nikhil Jain [Sat, 2 Mar 2013 06:49:50 +0000 (06:49 +0000)]
BGQ: memalloc fix for PAMI LRTS

4 years agoBGQ: Bug fix for limiting memory allocated on BGQ
Nikhil Jain [Sat, 2 Mar 2013 05:38:44 +0000 (05:38 +0000)]
BGQ: Bug fix for limiting memory allocated on BGQ

4 years agoA whole lot of FAQ updating
Eric Bohm [Sat, 2 Mar 2013 01:14:50 +0000 (19:14 -0600)]
A whole lot of FAQ updating

4 years agouse optimized memory allocator
Yanhua Sun [Fri, 1 Mar 2013 23:39:36 +0000 (23:39 +0000)]
use optimized memory allocator

4 years agoMerge branch 'charm' of charmgit:charm into charm
Nikhil Jain [Fri, 1 Mar 2013 20:27:50 +0000 (20:27 +0000)]
Merge branch 'charm' of charmgit:charm into charm

4 years agoBGQ: Correction in queue creation
Nikhil Jain [Fri, 1 Mar 2013 20:27:31 +0000 (20:27 +0000)]
BGQ: Correction in queue creation

4 years agolocmgr: remove unnecessary casts to CkArrayIndex
Ramprasad Venkataraman [Tue, 26 Feb 2013 23:18:01 +0000 (17:18 -0600)]
locmgr: remove unnecessary casts to CkArrayIndex

4 years agolocmgr: minor reindent
Ramprasad Venkataraman [Tue, 26 Feb 2013 21:40:30 +0000 (15:40 -0600)]
locmgr: minor reindent