Project

General

Profile

Activity

From 01/20/2016 to 02/18/2016

02/19/2016

11:08 PM Bug #987 (Merged): Entry methods with no parameters can't be called with CkEntryOptions, can't ta...
Most entry method declarations lead to generated methods on the corresponding CProxy object that include an optional ... Phil Miller

02/18/2016

01:42 PM Revision 621d3fdd (charm): AMPI: MPI_Igatherv, MPI_Iscatterv, MPI_Iallgatherv, and MPI_Ialltoallv
support
Change-Id: I508895ec68cbbfe9eb73b1eb4a96921e900b1bbe
Sam White
01:41 PM Revision 5c8dc593 (charm): AMPI: Fix linker error on Windows builds
Change-Id: I895c63bf42feb9c5e941c20a233f35d9f5d0958f Sam White
01:28 PM Revision d1074fe3 (charm): AMPI: MPI_Igather and MPI_Iscatter support
Change-Id: Ia93ee0830655129d540757eae80567b3ec92f968 Sam White
11:42 AM Bug #935 (Merged): Memory affinity autodetection doesn't work on RHEL 5
Eric Bohm
11:10 AM Revision d8fc771a (charm): Bug #935: Memory affinity autodetection doesn't work on RHEL 5
A very slightly modified version of Tom Coles' patch (see redmine
issue), uses numa.h API to set CmiNumNUMANodes and ...
Eric Bohm
11:01 AM Feature #982: Gather reduction operation using Tuple/Set reducers
this is not a bug, it is a new feature request. Eric Bohm
08:25 AM Revision 3630addc (charm): AMPI: MPI_Ibarrier and MPI_Ibcast support
Change-Id: Idad71f0904d91a7b21963055567cc9e416c64092 Sam White
12:35 AM Revision 686dc58f (charm): AMPI cleanup: make MPI_Info_* routine parameters const
Change-Id: I6bc2fe402dce3ae1f36d54dd0811e6260a6388d0 Sam White
10:40 PM Bug #941 (Rejected): Metabalancer initialization is not thread safe
Phil Miller
10:34 PM Feature #400 (Merged): In partitioned runs, abort message should include partition ID
All subtasks had implementations merged before 6.7.0 was released! Phil Miller
10:33 PM Feature #405: charmrun_abort should print partition ID in partitioned runs
Fix was actually merged before 6.7.0. Phil Miller

02/17/2016

05:07 PM Revision 29907985 (charm): Add -c to generate the .o in FT test
Since Sam made default-to-aout as the default behavior with AMPI, we
need -c to tell charmc that it should only be co...
Xiang Ni
03:52 PM Feature #984: AMPI support for missing MPI-3 non-blocking collectives
MPI_Igatherv, MPI_Iscatterv, MPI_Iallgatherv, MPI_Ialltoallv, MPI_Ireduce_scatter:
https://charm.cs.illinois.edu/g...
Sam White
11:35 AM Feature #984: AMPI support for missing MPI-3 non-blocking collectives
MPI_Igather and MPI_Iscatter:
https://charm.cs.illinois.edu/gerrit/#/c/1075/
Sam White
09:16 AM Feature #984 (In Progress): AMPI support for missing MPI-3 non-blocking collectives
MPI_Ibarrier and MPI_Ibcast:
https://charm.cs.illinois.edu/gerrit/#/c/1074/
Sam White
10:21 PM Feature #984: AMPI support for missing MPI-3 non-blocking collectives
Missing non-blocking collectives (strike-thru means implemented since opening this issue):
-MPI_Ibarrier-
-MPI_Ib...
Sam White
07:59 PM Feature #984 (Merged): AMPI support for missing MPI-3 non-blocking collectives
AMPI currently only supports the 4 following non-blocking collectives, and is missing the rest:
MPI_Ireduce
MPI_I...
Sam White
03:45 PM Revision c864303d (charm): Makefile: consistently pass $OPTS to *all* recursive make invocations
Phil Miller
02:13 PM Bug #986 (Merged): Cray regularpages build is broken
The regularpages build for Crays was broken by this commit:
https://charm.cs.illinois.edu/gerrit/#/c/852/
The e...
Sam White
02:13 PM Revision 5cc4755f (charm): Don't shadow variable declarations in for loops, to avoid Intel compil...
Change-Id: Idaad3eb78d578d00ea1ba476717ca5aba7183b1a Phil Miller
12:55 PM Bug #981 (Merged): Application hangs when load balancer is called with dynamic deletion
https://charm.cs.illinois.edu/gerrit/1056 Phil Miller
12:54 PM Cleanup #568 (Merged): Charm++ integrate and integrate2 examples
Phil Miller
12:53 PM Bug #977 (Merged): Comm Thread Tracing Build Fails
https://charm.cs.illinois.edu/gerrit/#/c/1050/ Phil Miller
12:52 PM Bug #962 (Merged): multicore segfaults with +p129 due to fixed-size array
Phil Miller
12:34 PM Revision 67471daa (charm): Fix various minor spelling errors
Change-Id: I5a5f57cacff4d07dca516c89d88a7d1054745efd Ronak Buch
12:32 PM Revision ade72406 (charm): doc: remove --enable-qlogic from install.tex, remove +nodesize from
run.tex and cleanup run.tex
Change-Id: Iee3ccb374a8e102ccdfb0138cd9b284003e23895
Sam White
12:12 PM Feature #985 (Merged): Make AMPI_Gather and its variants use Tuple/Set Reductions
AMPI_Gather, AMPI_Scatter, and all of their variants currently use point-to-point messages internally.
This will be ...
Sam White
11:59 AM Bug #983: AMPI chkpt test fails during restart on less PEs than were checkpointed with
The crash only happens when restarting on less PEs than were checkpointed with (not with more). Sam White
07:32 PM Bug #983 (Merged): AMPI chkpt test fails during restart on less PEs than were checkpointed with
The checkpoint/restart example in tests/ampi/chkpt/ fails, seemingly when trying to restart on a different number of ... Sam White
11:39 AM Revision a65b4f46 (charm): Bug #977: Fix comm thread tracing
Change-Id: Ib1dcc3870eafc747cc88ca5fc9f44fec8e4048be Ronak Buch
10:05 AM Cleanup #12: Factor out massive duplication in reductions
Replicas of the reduction infrastructure in Charm++:
* src/ck-core/ckreduction.C (x2: CkReductionMgr, CkNodeReduct...
Phil Miller
01:48 AM Revision 53f75b46 (charm): Cleanup #568: rename integrate2 and add README to clarify its purpose
Revised defaults and makefiles so that all the integrate examples produce
the same results.
Renamed integrate2 to in...
Eric Bohm
01:39 AM Revision 07e2a189 (charm): LBDBManager: Check whether deleting an object lets the AtSync() barrie...
Change-Id: Ib77c2fcdbbfb868380716e7cc34e3a47875eb03a Phil Miller
01:38 AM Revision 26ea275e (charm): darwin: switch from libstdc++ to libc++ for full C++11 library conform...
Change-Id: I13624ba1e7e7162c65cc3bd1cb53c4dec81f2a52 Phil Miller
01:37 AM Revision 1c380a44 (charm): Cleanup: give names to Converse PIC method types
Change-Id: I50f3004bfeff6a450672d14ebbe41cc1ef63e95c Sam White
07:58 PM Bug #826 (Merged): AMPI non-blocking collectives hang
MPI_Ialltoall and MPI_Iallgather:
https://charm.cs.illinois.edu/gerrit/#/c/1064/
Sam White
07:44 PM Feature #886 (Implemented): Add option to toggle comm thread tracing
https://charm.cs.illinois.edu/gerrit/#/c/1069/1 Ronak Buch

02/16/2016

04:49 PM Cleanup #12: Factor out massive duplication in reductions
Potentially moving those over to Charmworks, assuming Ronak hasn't started in on it yet. Phil Miller
04:21 PM Bug #951 (Implemented): multicore segfaults with +partitions 2
There were two aspects of the bug
1) Whenever the macro CMK_CCS_AVAILABLE (for CharmDebug) is set, any call made t...
Nitin Bhat
12:52 PM Bug #802 (In Progress): CUDA examples broken on Blue Waters
As a result of Debug Day:
# I have an improved error checking patch that I'm going to clean up and submit, I'll post...
Michael Robson
12:32 PM Revision c499684e (charm): Add current version of traceBegin/EndComm
Change-Id: I59e56f9ca72141764d126a305bdd6e0517225e7c Ronak Buch
12:32 PM Revision 5037b930 (charm): Fix spelling errors
Change-Id: I5a5f57cacff4d07dca516c89d88a7d1054745efd Ronak Buch
10:13 AM Revision e04e8cc8 (charm): AMPI cleanup: give names to AMPI request types
Change-Id: I9f0ee6f810c2cc59b1447337ec78f9c2451a3946 Sam White
08:32 AM Revision e00660d6 (charm): AMPI #826: Add non-blocking IATAReq type to fix MPI_Iallgather and MPI...
Change-Id: I5bab36c0d800c36cb875f20b60a79eb54096ec44 Sam White
09:04 PM Bug #826 (In Progress): AMPI non-blocking collectives hang
Sam White
07:11 PM Bug #826: AMPI non-blocking collectives hang
Fixed MPI_Ireduce and MPI_Iallreduce. A simple test program for both of those is attached here.
https://charm.cs.i...
Sam White
08:38 PM Revision 446ee84e (charm): Bug #826: Fix AMPI_Ireduce and AMPI_Iallreduce to not hang
Change-Id: I04f4b1e33d5693d359ab35c84a7290e3b4ca3eba Sam White

02/15/2016

04:58 PM Feature #982 (Merged): Gather reduction operation using Tuple/Set reducers
This is CkReduction::set with source identifiers per contribution.
We could have 2 arrays with the same number of ...
Sam White
04:53 PM Bug #597 (In Progress): AMPI pingpong example segfaults --with-production
Nitin Bhat
04:51 PM Bug #951 (In Progress): multicore segfaults with +partitions 2
Nitin Bhat
04:50 PM Bug #863: netlrts-smp hangs or rounds up pes on startup if ppn does not divide pes
Error message is thrown and charm exits when the number of PEs per node cannot divide the number of PEs.
Nitin Bhat
04:49 PM Bug #863 (Implemented): netlrts-smp hangs or rounds up pes on startup if ppn does not divide pes
Error message is thrown and charm exits when the number of PEs per node cannot divide the number of PEs.
Nitin Bhat
02:48 PM Bug #863: netlrts-smp hangs or rounds up pes on startup if ppn does not divide pes
https://charm.cs.illinois.edu/gerrit/#/c/1052/1 Bilge Acun
04:18 PM Bug #667 (In Progress): charm++/sdag/migration test fails with randomized queues
Relate with issue 665: https://charm.cs.illinois.edu/redmine/issues/665
With randomized queue, broadcast messages ar...
Xiang Ni
04:10 PM Feature #831 (Implemented): Reject duplicate command line arguments that don't make sense to repeat
When a user inserts duplicate arguments for the same option, charm RTS abrots the app and print the message saying th... Seonmyeong Bak
03:53 PM Feature #963 (In Progress): Runtime flag to silence runtime output
Juan Galvez
03:40 PM Feature #109: Test and merge section ID and manager work
ramv/com-kliqs is the branch that contains the code. I believe we delayed this due to the design collisions with 64b... Eric Bohm
03:33 PM Revision 37ba66a7 (charm): Bug #962: Do not free inProgress array
We do not need to free the array, since we are calling exit right after.
Also, freeing this was causing mysterious ha...
Bilge Acun
03:31 PM Support #972 (Closed): Investigate nightly autobuild failures
Error was caused by commit for bug 962. But we don't know the cause. Removing the free(inProgress) at the end of Conv... Juan Galvez
03:07 PM Support #972: Investigate nightly autobuild failures
The issue is that when autobuild runs charm++ tests, some tests processes finish but the process remains in memory.
...
Juan Galvez
11:07 AM Support #972 (In Progress): Investigate nightly autobuild failures
Some autobuild tests return 'Test on remote host', for example on integrity with net-linux build.
I replicated the s...
Juan Galvez
03:29 PM Support #637: update projections
CkLoop issues fixed in https://charm.cs.illinois.edu/gerrit/#/c/512/. Ronak Buch
03:28 PM Support #637 (Closed): update projections
Ronak Buch
02:58 PM Feature #620 (In Progress): pe threads sleep on condition variables
Bilge Acun
02:31 PM Feature #620: pe threads sleep on condition variables
Gerrit patch: https://charm.cs.illinois.edu/gerrit/#/c/1058/
Bilge Acun
02:55 PM Feature #975 (In Progress): OFI Layer
Progress of fixes: https://charm.cs.illinois.edu/gerrit/#/c/1059/ Bilge Acun
10:18 AM Feature #975 (Merged): OFI Layer
Integration of OFI layer implemented by Intel. Test and make it work on the Cisco cluster.
The initial patch from ...
Bilge Acun
02:50 PM Bug #802: CUDA examples broken on Blue Waters
When running the same hello program on blue waters using the nodeGPU patch we also get an error, albeit a different o... Michael Robson
02:47 PM Bug #935 (Implemented): Memory affinity autodetection doesn't work on RHEL 5
A very slightly modified version of Tom Coles' patch is pushed as gerrit https://charm.cs.illinois.edu/gerrit/#/c/106... Eric Bohm
02:05 PM Bug #981: Application hangs when load balancer is called with dynamic deletion
When an object is deleted, the barrier condition may get satisfied. This is not checked currently. Harshitha Menon
02:03 PM Bug #981 (Merged): Application hangs when load balancer is called with dynamic deletion
Harshitha Menon
02:02 PM Revision dff06b3b (charm): LBDBManager: Check whether deleting an object lets the AtSync() barrie...
Phil Miller
01:50 PM Cleanup #980 (New): Cleanup examples/charm++/cuda/hello
NVCC hardcoded here
Check other examples to ensure this isn't the case as well
This is also tangentially relate...
Michael Robson
01:38 PM Bug #806 (Implemented): CkReduction::set breaks CkMulticast section reductions
Fixed in https://charm.cs.illinois.edu/gerrit/#/c/1055/1 Eric Mikida
12:35 PM Bug #806: CkReduction::set breaks CkMulticast section reductions
Found the source of the error. It appears messages are not being correctly marked as leaves vs interior messages whic... Eric Mikida
12:03 PM Bug #806: CkReduction::set breaks CkMulticast section reductions
Removing the size check above is all that is needed to fix the issues with concatenation. The size was already correc... Eric Mikida
01:27 PM Cleanup #512: Remove duplication of conv-mach files
I fixed the special.sh duplication so there is only:
./common/special.sh (for CUDA builds)
./net-linux-cell/speci...
Michael Robson
01:25 PM Bug #788: Out of tree build broken for CUDA
https://charm.cs.illinois.edu/gerrit/#/c/1054/ Michael Robson
01:24 PM Bug #788 (Implemented): Out of tree build broken for CUDA
Michael Robson
01:08 PM Bug #979 (Closed): charmxi: mismatched module braces leads to confusing error messages.
The following code sample raises a very obscure error message:
@
mainmodule mod {
mainchare mchare {
entr...
Ralf Gunter Corrêa Carvalho
12:37 PM Feature #886: Add option to toggle comm thread tracing
For the sake of clarity, I'd like to note here that traceBegin and traceEnd do indeed enable and disable comm thread ... Ronak Buch
10:29 AM Feature #886 (In Progress): Add option to toggle comm thread tracing
Ronak Buch
12:19 PM Revision b2e23e2d (charm): Add message pool as well as projection events
Change-Id: Icdde390c05d80f84fe1e1043cbd34ed58dc50da7 Harshitha Menon
12:10 PM Cleanup #568 (Implemented): Charm++ integrate and integrate2 examples
https://charm.cs.illinois.edu/gerrit/#/c/1051/
integrate2 (now integratePUPable) is just the integrate example ext...
Eric Bohm
12:02 PM Revision 3424ccf2 (charm): Bug #977: Fix comm thread tracing
Change-Id: Ib1dcc3870eafc747cc88ca5fc9f44fec8e4048be Ronak Buch
11:49 AM Bug #883 (Closed): wave2D example gives segmentation fault on darwin builds (net & netlrts & mult...
Xiang Ni
11:49 AM Bug #883 (Resolved): wave2D example gives segmentation fault on darwin builds (net & netlrts & mu...
Making scalable start by default accidentally fixed the issue. The fix for this is the same as https://charm.cs.illin... Xiang Ni
11:46 AM Bug #978 (In Progress): Compare the performance of Harshit's accel patch to GPU manager
Harshit has a patch for accel: https://charm.cs.illinois.edu/gerrit/#/c/824/3
This needs to be tested and compared...
Prateek Jindal
11:38 AM Feature #2 (Merged): Reduce memory usage of reductions
Phil Miller
10:19 AM Feature #2 (Resolved): Reduce memory usage of reductions
Ronak Buch
11:36 AM Feature #405: charmrun_abort should print partition ID in partitioned runs
Assign back to Yanhua for attribution sake Phil Miller
11:36 AM Feature #405 (Merged): charmrun_abort should print partition ID in partitioned runs
Phil Miller
11:34 AM Feature #405 (Closed): charmrun_abort should print partition ID in partitioned runs
Already been merged earlier : https://charm.cs.illinois.edu/gerrit/#/c/113/ Vipul Harsh
11:02 AM Bug #977 (Merged): Comm Thread Tracing Build Fails
As noted in https://charm.cs.illinois.edu/gerrit/#/c/894/, the change breaks the build when comm thread tracing is en... Ronak Buch
11:01 AM Revision 0207a380 (charm): Added user trace events for the taskqueue Q
Change-Id: Ic646f2c1d872ccf579a47feafb283d9308984ecd Seonmyeong Bak
10:34 AM Projections Feature #976 (In Progress): Tool for analyzing network message latencies
Keshav Santhanam
10:28 AM Bug #696 (Closed): Stampede icc Shared Libraries Not Found
Ronak Buch
10:18 AM Feature #812: Merge Accel Branch
Needs to be performant before merging, waiting on cpu/gpu tagging or better kernel generation Michael Robson
10:15 AM Feature #861: Add Support for CrayPat
Blocked on getting information from Cray. Ronak Buch
11:28 PM Feature #973: multicore: spawn a thread per core by default
I agree that this will be fairly easy to implement once hwloc is integrated. However, changing a default behavior li... Eric Bohm
09:17 PM Feature #973: multicore: spawn a thread per core by default
Please do not change defaults (or anything else other than debugging assertions) based on the build flags (like --wit... Jim Phillips
07:30 PM Feature #973: multicore: spawn a thread per core by default
Assign to Eric B initially, since he's working on process launch configuration stuff intensively. If you agree that t... Phil Miller
07:29 PM Feature #973 (Merged): multicore: spawn a thread per core by default
The multicore builds of Charm++ don't have any big tradeoff space between process and thread count, or diversity in n... Phil Miller

02/12/2016

04:44 PM Revision e6754908 (charm): Charm's port for newer IBM systems
This patch contains Sameer's port for running Charm on top of newer IBM systems.
The base layer is PAMI, which should...
Sameer Kumar

02/11/2016

04:31 PM Revision b096827f (charm): Added listener for reporting local Array Indices
Change-Id: I7673c4e856ecc24e38dc241359a699ca74cfdf4e Nitin Bhat
04:24 PM Support #972 (Closed): Investigate nightly autobuild failures
Nightly autobuild has been failing since February 5th, figure out what recently merged patch has caused this Michael Robson
04:21 PM Bug #968: charm++ programs fail to run on BlueWaters due to craype-hugepages8M
After talking with Blue Waters support apparently the craype-hugepages8M module has been having problems.
Also, I ...
Michael Robson
04:20 PM Feature #971 (New): Improve malloc implementation
From core discussion:
* Should we lift the ugni layer's implementation of mempool for use everywhere?
* PAMI layer'...
Michael Robson
04:08 PM Bug #879: segfaults in megatest groupcast on little-endian Power8
Phil Miller wrote:
> It may also be an indication of undefined behavior. Perhaps compile with @-fsanitize=undefined@...
Michael Robson
02:43 PM Bug #879: segfaults in megatest groupcast on little-endian Power8
Inside ... Michael Robson
02:26 PM Bug #879: segfaults in megatest groupcast on little-endian Power8
It may also be an indication of undefined behavior. Perhaps compile with @-fsanitize=undefined@? Phil Miller
02:06 PM Bug #879 (In Progress): segfaults in megatest groupcast on little-endian Power8
Nikhil and I encountered this same bug this morning while working on the new PAMI layer. We're pretty sure it's a com... Michael Robson
03:27 PM Bug #964 (Closed): multicore icc build failed on stampede
Seonmyeong Bak
03:13 PM Bug #964 (In Progress): multicore icc build failed on stampede
Ronak Buch
11:46 AM Bug #805: SDAG tracing counts time waiting for a when to be satisfied as overhead
Ping, again. Is there actually fixed code for this sitting in a repository somewhere, or is that still a work in prog... Phil Miller
11:44 AM Bug #936 (Merged): miniAMR crashes with BigSim
Phil Miller

02/10/2016

05:49 PM Bug #943 (In Progress): Rename AMPI extensions to keep MPI namespace sanctity
Sam White
03:40 PM Revision e7460ecb (charm): AMPI: add MPI_Comm_set_name and MPI_Comm_get_name
Change-Id: Ibaade1622bd70d628a2b7c730f566cf8dfa0ef34 Sam White
03:01 PM Revision 43ca1598 (charm): AMPI: return MPI_ERR_ARG in MPI_Group_range_{incl,excl} when stride is 0
Change-Id: Ib7cc036519d87cc6bfe07e061071885022ab6eba Sam White
09:44 AM Cleanup #913 (Closed): Factor out repeated logic in AMPI_Test* and AMPI_Wait*
Some of this was addressed by refactoring the MPI_Test* routines (below) and on closer inspection each of these has s... Sam White
09:38 AM Feature #947: AMPI support for MPI-2's MPI_Type_* routines
All new datatype support should be implemented in CkDDT, not AMPI, so these changes were reverted for that reason and... Sam White
09:35 AM Feature #938 (Merged): Add '-default-to-aout' flag to AMPI compilers by default
Phil Miller
11:38 PM Revision 2dfac815 (charm): AMPI: return MPI_COMM_NULL instead of aborting when creating an empty ...
Change-Id: I5accb9abf1210a8cb13bb276040b926eaf4d62b6 Sam White
11:37 PM Revision eb77e9e5 (charm): AMPI: support MPI_SIMILAR in MPI_Comm_compare
Change-Id: I33b751c739d6f3ed6948a7b2ed4edba492eb216e Sam White
11:37 PM Revision 32304e6a (charm): AMPI: add missing BigSim guard to trace call in MPI_Barrier
Change-Id: Iea3a66ec06b68e943d2a8495bd3268d4a0a88ee8 Sam White
11:36 PM Revision 2636ced4 (charm): AMPI: remove definitions of MPI_ORDER_C and MPI_ORDER_FORTRAN
to keep ROMIO happy.
Add these back when they are needed by
MPI_Type_create_subarray and MPI_Type_create_darray.
Ch...
Sam White
11:35 PM Revision b656f9e8 (charm): AMPI: support for MPI_Request_get_status
Change-Id: I4107e051e4f30736ffe0241670e786c918760412 Sam White
11:35 PM Revision ff5550c6 (charm): Feature #938: Add -default-to-aout to AMPI compiler flags by default
so AMPI compilers can be used plainly with autoconf and cmake
Change-Id: I8e97ecf76e707a835e53244a0127b6f590e36309
Sam White
10:58 PM Revision 03cc0478 (charm): doc: document the +LBTestPESpeed runtime option
Change-Id: Ic8c0499bb1ef2501a9b50a7e2a92ad5b9b5b4a42 Sam White
10:57 PM Revision 001da651 (charm): Make the Refiner used in refine LBs rate aware
Change-Id: I0cdd0844e08a4981be75c630ddaeb3374d221d19 Eric Mikida

02/09/2016

05:53 PM Revision ca012059 (charm): adding power reading, tab correction and some other minor progress
Change-Id: Idd4d8aef57957f327bd938097d24e650960b3577 Bilge Acun
01:02 PM Revision 7417bf5d (charm): charmxi: correct mismatched appwork begin/begin tracing on Group [loca...
Change-Id: I2218c8ab8b869998ed586ad6960cff7b9b2da538 Phil Miller
01:01 PM Revision 471e39c4 (charm): charmxi: fix some egregious whitespace problems (indentation, mis-matc...
Change-Id: Idd3f0619b84d755e42b8218f734ab8faa09637ae Phil Miller
01:01 PM Revision 41f84559 (charm): charmxi: refactor entry method error checking to happen in standard fa...
Change-Id: I36d7486b62ea885a31c04b36c77fd4d3b2cef28f Phil Miller
12:05 PM Revision d5156564 (charm): Single instance of GPU Manager, shared across all PEs within a node
Change-Id: I15d0808a4d6f38fbacfe30cb0642dc6d56cc944c Harshit Dokania
12:05 PM Revision bef3b93a (charm): exitHybridAPI(): delete extra CmiDestroyLock().
Thomas Quinn
12:05 PM Revision 84c70435 (charm): nodeGPU: refactor recursive function and extend lock
Redmine Admin
12:03 PM Revision 4e9f73c5 (charm): Single instance of GPU Manager, shared across all PEs within a node
Change-Id: I15d0808a4d6f38fbacfe30cb0642dc6d56cc944c Harshit Dokania

02/08/2016

03:57 PM Feature #934 (Implemented): Isomalloc: refactor to associate migratable allocations with entities...
https://charm.cs.illinois.edu/gerrit/#/c/1046/ Phil Miller
03:52 PM Revision 67ecd024 (charm): Isomalloc #934: Move usage to an explicit BlockList interface, to avoi...
This enables code to use Isomalloc-managed migratable heaps without associating
them with a corresponding migratable ...
Phil Miller
10:05 AM Revision 89cff717 (charm): NOMERGE: Isomalloc #934: Start moving everything to the BlockList inte...
Change-Id: I2cdf750fc8a95c3fbbf4f0e5d69d38ebfbab9a24 Phil Miller
09:32 PM Revision 7947a706 (charm): Bug #903: Only free from one thread in smp mode
Fixing the bug that Commit bed4af1a6423921ed52bb06cac281788f97f2523
has introduced
Change-Id: Ief389b80620d3ee4b3487...
Bilge Acun

02/06/2016

06:53 PM Feature #969 (In Progress): AMPI support for collectives on inter-communicators
For MPI-2, the following intracommunicator collective operations also apply to intercommunicators:
• MPI BCAST
• ...
Sam White

02/05/2016

05:05 PM Revision 17532e95 (charm): AMPI: add MPI_Comm_create_group routine
Change-Id: Ic6fae9768dd0a69d4cfca5073d094d65f1e2829d Sam White
12:10 PM Revision c6aa58fa (charm): merging
Change-Id: Iacaa2bc1e4476c6ff88879b10fdd128d5e5f384f Sam White
12:02 PM Revision c9ac3f4e (charm): AMPI: support MPI_CONGRUENT in MPI_Group_compare
Change-Id: Idb0eba4db5473597db30b67e4886b441b8608c41 Sam White

02/04/2016

05:50 PM Revision f835f2a0 (charm): Revert "AMPI: support for MPI_Type_create_subarray"
Missing support for MPI_Type_create_darray has broken ROMIO's
build, and support for the Subarray type should be adde...
Sam White
05:26 PM Revision 9e17c278 (charm): AMPI: add MPI_NO_OP definition
Change-Id: I94735e6da739e89e7436d9ca9c946137709fc68a Sam White
05:15 PM Bug #968 (Upstream): charm++ programs fail to run on BlueWaters due to craype-hugepages8M
When running the ... Michael Robson
03:04 PM Feature #920: Per-GPU Node Groups
Discussion concluded this should be something like 2D NodeGroups - add within-node indexing of multiple instances. Phil Miller
02:59 PM Revision 56a3028c (charm): ampif: add MPI_REPLACE to Fortran MPI_Op list
Change-Id: Ia6977e753b37c3817854e206ab2f7cd9eb28feb5 Sam White
02:50 PM Bug #863: netlrts-smp hangs or rounds up pes on startup if ppn does not divide pes
Ping. Jim Phillips
02:32 PM Revision 048f77ad (charm): AMPI: add MPI_Get_library_version
Change-Id: Idfeed7ea110f05b82e3b5fdc4c58b2583454154d Sam White
02:28 PM Revision c2e09874 (charm): Add headers
Change-Id: Idf70b07515dfae4e29f033d69d7e4756a207fd07 Ronak Buch
02:03 PM Revision bed4af1a (charm): Bug #962: dynamically allocate inProgress array size
Change-Id: If1cec66d37afe3613e3cd000dbac72a38b01546a Bilge Acun
01:45 PM Revision c8b78149 (charm): Add lustre group lock
Change-Id: If96613e23a7ab72452bb67749291bc1a93fb39dd Ronak Buch
01:27 PM Revision 20796efd (charm): Make ++local execve error message show path of program instead of char...
Change-Id: I2d0d1fb7903490b2184e259d297acb277bffbf1e Jim Phillips
09:30 PM Revision 6e5ad96b (charm): AMPI: add MPI_NO_OP definition
Change-Id: I94735e6da739e89e7436d9ca9c946137709fc68a Sam White
09:16 PM Revision 58b52c95 (charm): ampif: add MPI_REPLACE to Fortran MPI_Op list
Change-Id: Ia6977e753b37c3817854e206ab2f7cd9eb28feb5 Sam White
07:45 PM Feature #967 (Merged): AMPI support for MPI-3's MPI_Comm_create_group
https://www.open-mpi.org/doc/v1.8/man3/MPI_Comm_create_group.3.php Sam White

02/03/2016

10:53 AM Cleanup #966 (Merged): cuda build fails unituitvely on blue waters when cudatoolkit isn't loaded
... Michael Robson

02/02/2016

05:23 PM Revision 9f5d77d8 (charm): Projection trace code added
Change-Id: I981c4f1895b24eaa36b20cfb52c01979e192f3a2 Seonmyeong Bak
04:58 PM Revision 287fd733 (charm): AMPI: add MPI_Get_library_version
Change-Id: Idfeed7ea110f05b82e3b5fdc4c58b2583454154d Sam White
03:42 PM Revision 2b1cd58e (charm): More cleanup, make it worth with more chares than AMPI ranks.
Change-Id: Ia316e8d832678b6a9c623b88bf3fa763e68e6b10 Sam White
11:04 AM Bug #962 (Implemented): multicore segfaults with +p129 due to fixed-size array
https://charm.cs.illinois.edu/gerrit/#/c/1033/ Bilge Acun
10:47 AM Revision c51832fa (charm): Pulled in changes from branch charm
Change-Id: If742cd5ff202ae6e09c2367bc8365291eb672cdd Sam White
10:34 AM Revision ec34f297 (charm): Cleanup up AMPI interop example in examples/charm++/AMPI-interop, adde...
Change-Id: Ifae0746f4d228ed046a500d10f53d9fbecf6ea9c Sam White

01/31/2016

11:24 AM Bug #964: multicore icc build failed on stampede
Try using a fresh copy of charm. It looks like your Make dependencies are out of date. Multicore-linux64-icc builds f... Sam White
10:06 AM Bug #965 (Closed): ampicc -swapglobals is broken for ld v2.24+
With gcc 4.8/ld 2.24, global variables are not listed in the relocation records. Thus preventing swapglobals from swa... Nikhil Jain

01/29/2016

01:39 PM Revision 8334287c (charm): Add ccd call for steal when still idle
Change-Id: I164689b85b072cb4a5ad8127f25a257adb297cba Harshitha Menon

01/28/2016

05:57 PM Revision 4769a20e (charm): nodeGPU: refactor recursive function and extend lock
Redmine Admin
04:19 PM Bug #964 (Rejected): multicore icc build failed on stampede
When building charm++ on stampede in the multicore version, it failed to be built.
- Error message ...
Seonmyeong Bak
03:35 PM Revision 17578a4b (charm): Feature #2: Make reductions streamable
If the reduction operation is streamable, compute the
results on the fly rather than waiting for all of the
messages ...
Ronak Buch
02:43 PM Bug #926: hangs on verbs error
This is partially resolved by the CkEnforce commit, but proper error messages are needed - now issue #960. Jim Phillips
02:27 PM Bug #879: segfaults in megatest groupcast on little-endian Power8
Ping. Jim Phillips

01/27/2016

03:51 PM Revision 52cc1e90 (charm): AMPIF cleanup: eliminate compiler warning for MPI_Op_create
Change-Id: I0c0940a05b1001ce66a575b289fbd0f9ba3db08b Sam White

01/26/2016

01:02 PM Revision 74db1f58 (charm): Use optimized recursive loop structure
Change-Id: I426eab363d76f4c52cc311ce313b187a39972ed4 Harshitha Menon
11:11 AM Bug #833 (Implemented): mpi smp build is locked to one core per node by default
Fix here: https://charm.cs.illinois.edu/gerrit/#/c/1030/
Juan Galvez
10:13 PM Support #932 (Resolved): Link to Installation page broken
Nikhil Jain
10:11 PM Revision 7ae87cec (charm): Charmrun: on failures, report node ID as well as host name
Change-Id: I73dcd953ae1d257bc188889d4db768ede5475b99 Phil Miller

01/25/2016

01:17 PM Revision 957eb62d (charm): added reduction
Change-Id: Id665b1466a6eeb9d8dc454e1be2134862bf9eb57 Harshitha Menon

01/24/2016

06:24 PM Bug #959 (Merged): AMPI built-in keyval attributes for mype, numpes, etc. are broken
Sam White

01/23/2016

10:51 AM Feature #963: Runtime flag to silence runtime output
A practical use case for this flag: HDF5's build system compiles and runs a simple program and then parses the output... Sam White

01/22/2016

05:24 PM Feature #963 (Merged): Runtime flag to silence runtime output
We should have a flag (++silent, ++silence, ++quiet, whatever) that suppresses runtime output such as all preamble in... Sam White
04:49 PM Bug #962: multicore segfaults with +p129 due to fixed-size array
Same issue in src/arch/verbs/machine.c Jim Phillips
04:39 PM Bug #962 (Merged): multicore segfaults with +p129 due to fixed-size array
In src/arch/netlrts/machine.c and src/arch/net/machine.c the inProgress array size is fixed at 128.
This size should...
Jim Phillips
01:12 PM Revision 7f4b0019 (charm): AMPI: add support for MPI_REPLACE operation
Change-Id: I144a6444a2d364704229adc084d81fd037f9ac72 Sam White
01:12 PM Revision 1a5e833f (charm): Fix a bug in how the chunks were created
Change-Id: I90c366e937799bb7bfa65abf784a5e689fc3cd2c Harshitha Menon
11:55 AM Bug #961 (Merged): AMPI support for cancelling recv requests/statuses
3 semi-related issues:
1) MPI_Test_cancelled is a no-op
2) MPI_Cancel incorrectly frees requests instead of marking...
Sam White

01/21/2016

05:45 PM Bug #959 (Implemented): AMPI built-in keyval attributes for mype, numpes, etc. are broken
Changed name of attributes to AMPI_MY_PE, AMPI_NUM_PES, AMPI_MY_NODE, and AMPI_NUM_NODES.
https://charm.cs.illinoi...
Sam White
02:47 PM Bug #959 (Merged): AMPI built-in keyval attributes for mype, numpes, etc. are broken
AMPI_KEYVAL_MYPE, AMPI_KEYVAL_NUMPES, AMPI_KEYVAL_MYNODE, AMPI_KEYVAL_NUMNODES are defined in AMPI for all communicat... Sam White
05:43 PM Revision c93f592c (charm): AMPI cleanup: add MPI_Test_cancelled to AMPIF
Change-Id: I29b254a8d9c31987023fcfd9b748799dcb8df2c5 Sam White
05:36 PM Revision e7eabb9e (charm): Bug #959: Rename and fix AMPI builtin keyval attributes
Change-Id: I294e93018416bdacdcd3a8e883b5fc365ced53b6 Sam White
05:32 PM Revision 0d1cd2b8 (charm): AMPI: bump MPI_VERSION and MPI_SUBVERSION to 2.2
Change-Id: I984a69ad9d110da5f8263999271cda825af05f4f Sam White
04:36 PM Revision 2b1bf178 (charm): Add CmiEnforce macro to check conditions in production builds
Implement CmiEnforce macro, patterned on CmiAssert, but always on.
For verbs/ibverbs substitute CmiEnforce for CmiAs...
Jim Phillips
04:34 PM Cleanup #960 (Merged): Replace CmiEnforce(0) calls in verbs and net layer
See [[https://charm.cs.illinois.edu/gerrit/#/c/966/]] for discussion. Copying some relevant bits here.
Phil: All o...
Michael Robson
03:20 PM Revision b5fae48c (charm): Feature #949 AMPI: Add MPI-2.2 compliant implementation of Spawn.
Calling Spawn is a user error if the MPI_UNIVERSE_SIZE attribute
of MPI_COMM_WORLD has a value equal to MPI_COMM_WORL...
Sam White
02:42 PM Bug #943: Rename AMPI extensions to keep MPI namespace sanctity
Consensus is that AMPI should have one extension routine for load balancing and all checkpointing schemes, with the u... Sam White
02:07 PM Bug #943: Rename AMPI extensions to keep MPI namespace sanctity
I was thinking of those attributes, yeah. We should make sure they're set as appropriate, rather than adding extra ro... Phil Miller
02:00 PM Bug #943: Rename AMPI extensions to keep MPI namespace sanctity
Thanks, I fixed the case inconsistency.
I don't think there is any standard MPI interface for getting an MPI rank'...
Sam White
01:22 PM Bug #943: Rename AMPI extensions to keep MPI namespace sanctity
Your @AMPI_Register_Just_migrated@ has a case inconsistency. 'J' should be lower case.
I think there are standard ...
Phil Miller
02:05 PM Feature #945 (Merged): AMPI compliant no-op implementation of MPI-2's Spawn and Spawn_multiple
Sam White
02:05 PM Feature #950 (Merged): AMPI support for MPI-2's Win related functions
Sam White
01:53 PM Revision 904294e5 (charm): Trace module api changes for pics.
Change-Id: I5feaaded3e6f677cf59e850032430b511852180b Yanhua Sun
01:19 PM Revision af3246bc (charm): ampif: make mpi_win_* functions take pointer arguments
Change-Id: Iebb71013afdb221f1ca4ccf8f6585b58c962ddb2 Sam White
01:19 PM Revision 06efd153 (charm): Feature #950: AMPI support for missing MPI-2 one-sided features.
Change-Id: I28510b5a6c7e991ebb1031ebe017e6132938fbec Sam White
01:18 PM Revision 2a257863 (charm): AMPI: refactor keyval and attribute support to support MPI_Win
and eventually MPI_Type objects, instead of assuming MPI_Comm.
Change-Id: I3b64e025bea220625d726cf319e6e1e4902a3184
Sam White
 

Also available in: Atom