Project

General

Profile

Activity

From 01/30/2018 to 02/28/2018

03/01/2018

10:45 PM Bug #1796: Support for partitions on netlrts-linux-x86_64-tcp builds
Please run this partitions test on v6.8.2 to see if the recent changes to charmrun caused this issue Sam White
08:38 PM Bug #1698 (Implemented): [inline] template entry methods do not work
https://charm.cs.illinois.edu/gerrit/#/c/3790/ Eric Mikida

02/28/2018

04:27 PM Bug #1699 (Implemented): [inline] entry methods should use perfect forwarding in C++11
https://charm.cs.illinois.edu/gerrit/#/c/3788/ Sam White
03:38 PM Feature #1808 (Implemented): Direct API for the MPI layer
Nitin Bhat
03:30 PM Feature #1739 (Implemented): Improve PUP performance of std::vector<T> where T is PUPbytes
Make PUP STL check if the type is PUPbytes rather than just arithmetic: https://charm.cs.illinois.edu/gerrit/#/c/3785... Sam White
02:49 PM Feature #1739 (In Progress): Improve PUP performance of std::vector<T> where T is PUPbytes
Sam White
03:19 PM Bug #1698: [inline] template entry methods do not work
Is this only in the case where template arguments can't be inferred automatically? Eric Mikida
11:12 AM Feature #1805 (Implemented): Direct API for the GNI layer
Nitin Bhat
11:09 AM Feature #1810 (Implemented): Direct API - Support multiple user operation modes
Nitin Bhat

02/27/2018

04:16 PM Bug #1801: charmrun on multicore builds doesn't consume ++local
> I think we currently only consume things like `++ppn` in SMP builds, though I think we should instead consume them ... Evan Ramos
10:29 AM Bug #1801 (Merged): charmrun on multicore builds doesn't consume ++local
Sam White
03:37 PM Feature #1113 (Implemented): smart-build.pl should detect supercomputers with specialized softwar...
Evan Ramos
03:20 PM Feature #1814: Add CkLoop Split Execution
Seonmyeong Bak wrote:
> I think adding an API to wait for all the spawned tasks is enough for this issue with some c...
Ronak Buch
02:00 PM Feature #1814: Add CkLoop Split Execution
I think adding an API to wait for all the spawned tasks is enough for this issue with some changes to the non-sync ve... Seonmyeong Bak
10:30 AM Bug #1699 (In Progress): [inline] entry methods should use perfect forwarding in C++11
Sam White
10:29 AM Documentation #1583 (Merged): Check commented old template documentation in doc/charm++/inhertmpl...
Sam White
11:26 PM Bug #1684 (Merged): AMPI message sequence numbers are susceptible to overflow
Sam White

02/26/2018

05:12 PM Feature #1814: Add CkLoop Split Execution
Assigning it to Seonmyeong since he's in charge of CkLoop now, but feel free to triage it. Ronak Buch
05:10 PM Feature #1814 (New): Add CkLoop Split Execution
Based on the documentation and code samples, the only way to currently use CkLoop is by calling the CkLoop_Paralleliz... Ronak Buch
02:58 PM Bug #1811: Cross-array section reduction and multicast with callback set at contribute time
Is the callback version broken? The desired functionality is that one can pass any reduction target as the destinati... Eric Bohm
02:24 PM Bug #1811: Cross-array section reduction and multicast with callback set at contribute time
For the cross array section reduction to work the reduction client should be set first.
The example code in: ...
Raghavendra Kanakagiri

02/25/2018

05:49 PM Bug #1813 (Merged): Fix failed syncft builds and pamilrts-async build after CMA patch integration
Sam White
06:30 PM Bug #1813 (Implemented): Fix failed syncft builds and pamilrts-async build after CMA patch integr...
Fix: https://charm.cs.illinois.edu/gerrit/#/c/3771/ Nitin Bhat

02/24/2018

04:33 PM Bug #1813 (Merged): Fix failed syncft builds and pamilrts-async build after CMA patch integration
Few builds have failed during autobuild as they used a message header that didn't include the 'type' parameter.
<...
Nitin Bhat

02/23/2018

10:37 AM Feature #1655 (Merged): Enable use of shm transport for regular messages in LRTS
Nitin Bhat
09:07 AM Bug #1812 (Rejected): charmc option "-ld" to specify linker conflicts with passing "-ld" to link ...
Evan Ramos
08:35 AM Support #1674 (Resolved): Add 'ofi' target to autobuild
ofi non-SMP and SMP passed yesterday. The SMP build seems to oftenhang, so that should still be monitored and address... Sam White

02/22/2018

04:22 PM Bug #1812: charmc option "-ld" to specify linker conflicts with passing "-ld" to link libd.so
Also, don't you need to link with libdl, not libd? Jim Phillips
04:16 PM Bug #1812: charmc option "-ld" to specify linker conflicts with passing "-ld" to link libd.so
I think you can do this with "-ld++-option -ld". Jim Phillips
10:31 AM Bug #1812 (In Progress): charmc option "-ld" to specify linker conflicts with passing "-ld" to li...
https://charm.cs.illinois.edu/gerrit/3761 Evan Ramos
06:53 PM Bug #1812 (Rejected): charmc option "-ld" to specify linker conflicts with passing "-ld" to link ...
Any program that uses dlopen() will need to link with libd.so by passing `-ld` to the linker. However, charmc has opt... Evan Ramos
03:57 PM Feature #1113 (In Progress): smart-build.pl should detect supercomputers with specialized softwar...
https://charm.cs.illinois.edu/gerrit/3765 Evan Ramos
12:39 PM Documentation #1583 (Implemented): Check commented old template documentation in doc/charm++/inhe...
https://charm.cs.illinois.edu/gerrit/3763 Evan Ramos
12:38 PM Documentation #1583: Check commented old template documentation in doc/charm++/inhertmplt.tex aga...
Yes, that'd be good Sam White
12:26 PM Documentation #1583 (In Progress): Check commented old template documentation in doc/charm++/inhe...
It appears that all the information contained in inhertmplt.tex but commented out was successfully migrated to templa... Evan Ramos

02/21/2018

05:24 PM Bug #1811 (Merged): Cross-array section reduction and multicast with callback set at contribute time
Cross array sections and multicasts have basic testing in mega test.
Cross array section reductions have questiona...
Eric Bohm
03:10 PM Bug #1699: [inline] entry methods should use perfect forwarding in C++11
The reason is that currently without perfect forwarding the called function would receive an lvalue reference, which ... Nils Deppe
03:03 PM Bug #965: ampicc -swapglobals is broken for ld v2.24+
Removing v6.9 target as per Core meeting. The crash on new systems has been resolved but swapglobals does not functio... Evan Ramos
02:55 PM Bug #1801 (Implemented): charmrun on multicore builds doesn't consume ++local
https://charm.cs.illinois.edu/gerrit/3758 Evan Ramos
02:37 PM Bug #1797: ARM64 build and stability issues
If there is any need for help with testing, I personally own a Raspberry Pi 3 and have an ARM64 installation on an SD... Evan Ramos
02:01 PM Bug #1797: ARM64 build and stability issues
Someone in NAMD group should take on arm64 support. Ronak, please reassign as appropriate. Eric Bohm
02:07 PM Feature #1805 (In Progress): Direct API for the GNI layer
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3117/ Nitin Bhat
09:04 AM Feature #1805 (Merged): Direct API for the GNI layer
Nitin Bhat
02:07 PM Feature #1806 (In Progress): Direct API for pamilrts-bluegeneq
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3118/ Nitin Bhat
09:04 AM Feature #1806 (Merged): Direct API for pamilrts-bluegeneq
Nitin Bhat
02:06 PM Feature #1808 (In Progress): Direct API for the MPI layer
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3120/ Nitin Bhat
09:05 AM Feature #1808 (Merged): Direct API for the MPI layer
Nitin Bhat
02:05 PM Feature #1809 (In Progress): Direct API for the OFI layer
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3126/ Nitin Bhat
09:06 AM Feature #1809 (Merged): Direct API for the OFI layer
Nitin Bhat
02:05 PM Feature #1807 (In Progress): Direct API for the Verbs layer
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3119/ Nitin Bhat
09:05 AM Feature #1807 (Merged): Direct API for the Verbs layer
Nitin Bhat
02:04 PM Feature #1810 (In Progress): Direct API - Support multiple user operation modes
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3433/ Nitin Bhat
09:11 AM Feature #1810 (Merged): Direct API - Support multiple user operation modes
There are four user operation modes that need to be supported in this feature. These modes can be passed into CkNcpyS... Nitin Bhat
09:49 AM Feature #1804 (Implemented): Direct API - Add CMA support for netlrts, multicore and pami builds
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3433/ Nitin Bhat
09:02 AM Feature #1804 (Merged): Direct API - Add CMA support for netlrts, multicore and pami builds
CMA support for the Direct API is added in the LRTS API implementation (https://charm.cs.illinois.edu/redmine/issues/... Nitin Bhat
08:43 AM Feature #1803 (Implemented): Direct API for the LRTS layer
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/3116/ Nitin Bhat

02/20/2018

05:40 PM Cleanup #292 (Implemented): charmc: Inconsistent handling of distinction between C and C++ linking
https://charm.cs.illinois.edu/gerrit/515 Evan Ramos
05:28 PM Cleanup #292: charmc: Inconsistent handling of distinction between C and C++ linking
Yeah I think so Sam White
05:22 PM Cleanup #292: charmc: Inconsistent handling of distinction between C and C++ linking
This issue description is solely about OPTS_LD(XX), while the linked patch also affects CMK_CXX_OPTIMIZE and CMK_CXX_... Evan Ramos
05:11 PM Feature #1704 (Implemented): Add a pamilrts-linux-ppc64le build target
Nitin Bhat
03:53 PM Feature #1803 (Merged): Direct API for the LRTS layer
Nitin Bhat
03:27 PM Feature #1802 (Merged): Direct API for the generic layer (non-rdma based default implementation)
Gerrit: https://charm.cs.illinois.edu/gerrit/#/c/2967/ Nitin Bhat
12:56 PM Feature #1655 (Implemented): Enable use of shm transport for regular messages in LRTS
This patch supports using CMA for regular messages. However, since the advantages of using CMA over Charm's regular n... Nitin Bhat
11:32 AM Feature #1420: Lockless queue build option --enable-lockless-queue (off by default)
For whoever is going to evaluate the performance against the Boost Queue, they should also evaluate against moodycame... Ronak Buch
10:34 AM Bug #1711 (In Progress): syncft tests: unclear failure
Juan Galvez

02/19/2018

01:20 PM Bug #1711: syncft tests: unclear failure
The autobuild failures are caused by this commit: "Charmrun: Distribute PEs among hosts using two phases of communic... Juan Galvez
09:56 AM Documentation #1156 (Merged): Add example showing refnum's on callbacks
Sam White
08:44 AM Bug #1799 (Merged): Charm++ program segfaults when an element from a nodegroup proxy is used for ...
Ronak Buch

02/18/2018

09:01 PM Bug #1801 (Merged): charmrun on multicore builds doesn't consume ++local
When using a multicore build, charmrun doesn't consume '++local'. I think that it should, since it makes perfect sens... Sam White

02/17/2018

12:03 AM Bug #1800 (Merged): Section reduction contribute() should accept std::vector
We added a new overload for contribute() which accepts a std::vector instead of taking a length and a buffer address.... Sam White
10:49 PM Feature #967 (Merged): AMPI support for MPI-3's MPI_Comm_create_group
Sam White
10:49 PM Feature #1018 (Merged): AMPI support for MPI-3's distributed graph interface
Sam White
10:49 PM Cleanup #1103 (Merged): Refactor AMPI virtual topologies
Sam White
07:59 PM Bug #1220: AMPI: Support tlsglobals with dynamically linked objects
Yeah -swapglobals doesn't work in SMP mode Sam White
07:54 PM Bug #1220: AMPI: Support tlsglobals with dynamically linked objects
Sam White wrote:
> I think it's probably worth seeing how much the data structures in question have actually changed...
Evan Ramos
07:52 PM Bug #1220: AMPI: Support tlsglobals with dynamically linked objects
I think it's probably worth seeing how much the data structures in question have actually changed over time. For inst... Sam White
06:47 PM Bug #1220: AMPI: Support tlsglobals with dynamically linked objects
I have investigated this issue in depth and have come to some conclusions. In summary, tlsglobals works by swapping t... Evan Ramos
06:00 PM Bug #965: ampicc -swapglobals is broken for ld v2.24+
It looks like swapglobals outright crashes with more up-to-date systems:... Evan Ramos

02/16/2018

02:40 PM Cleanup #895: Refactor charmrun interface code in RTS to deduplicate netlrts/verbs
No, there is still code duplication between netlrts and verbs. They strongly parallel each other for the majority of ... Evan Ramos
11:59 PM Cleanup #895: Refactor charmrun interface code in RTS to deduplicate netlrts/verbs
Evan, did your work on charmrun address this? Sam White
02:27 PM Projections Bug #459: inline and local entry method can not be tracked back
It does but the issue is https://charm.cs.illinois.edu/redmine/issues/937 doesn't work in case when inline entry meth... Seonmyeong Bak
11:57 PM Projections Bug #459: inline and local entry method can not be tracked back
Does the related bug #937 fix this entirely? Sam White
08:27 AM Bug #1711: syncft tests: unclear failure
This has failed the past 3 nights in autobuild. Sam White
08:12 AM Bug #1275 (Implemented): DistributedLB: Objects not migrating after strategy runs
Sam White
12:09 AM Bug #1263: AVX512 Support
Are there users of this library? If not, we might want to remove it. Sam White
12:04 AM Cleanup #1034 (Closed): Constant comparison warnings from xlC compiler
He said these warnings have been taken care of or the code they were in was removed. Sam White
12:03 AM Bug #1037: megacon link failure
Is this still an issue? Close if not. Sam White
12:02 AM Support #1079 (Closed): Remove deprecated machine layers and retired machines from Autobuild
Sam White
12:02 AM Bug #1063 (New): Error in buddy assignment code for checkpointing
Sam White
12:00 AM Bug #937: [local] entry methods don't set tracing event dependencies
Does this also fix bug #459? Sam White
12:00 AM Bug #978: Compare the performance of Harshit's accel patch to GPU manager
Is this still relevant? If not, close the issue. Sam White
11:57 PM Projections Feature #505: show communication threads in usage profile
Bump. This was marked Implemented 8 months ago but never merged. Sam White
11:41 PM Feature #321 (Closed): default, efficient sections
This seems too broad and ill-described to be an issue. Sam White
11:40 PM Cleanup #292: charmc: Inconsistent handling of distinction between C and C++ linking
A partial fix, though probably quite rusty at this point: https://charm.cs.illinois.edu/gerrit/#/c/515/ Sam White
11:38 PM Cleanup #418 (Closed): Investigate profusion of user-level thread libraries
We already merged support for Boost's uFcontext threads in 6.9.0. These are fast and portable and until ULTs are stan... Sam White
11:33 PM Bug #184 (Closed): PAMI LRTS performance using ppn 64 is worse than that with PAMI
This is really old and seems to not have been reproduced Sam White
11:32 PM Feature #22 (Closed): Stack aliasing on BGQ
We don't care about BGQ enough anymore to make this worth doing Sam White
11:30 PM Feature #13: CkCallback to section breaks
Is this issue still around? If so, we should target the fix to 6.9.0 or 6.9.1 Sam White
11:27 PM Feature #92: Topology information for Cray XC
Close this if no longer deemed relevant Sam White
11:24 PM Feature #190 (Closed): Build target(s) for Intel MIC
Sam White

02/15/2018

02:02 PM Bug #1799 (Implemented): Charm++ program segfaults when an element from a nodegroup proxy is used...
https://charm.cs.illinois.edu/gerrit/#/c/3726/ Ronak Buch
01:35 PM Bug #1799 (In Progress): Charm++ program segfaults when an element from a nodegroup proxy is used...
Ronak Buch
01:24 PM Feature #1420: Lockless queue build option --enable-lockless-queue (off by default)
I'll add a build option to make this lockless queue implementation available for those who want to test this.
One...
Seonmyeong Bak

02/14/2018

05:41 PM Bug #965: ampicc -swapglobals is broken for ld v2.24+
In Core we decided that we should do the following for this:
1. Add a configure check for the version of ld, if the ...
Sam White
05:23 PM Bug #937: [local] entry methods don't set tracing event dependencies
https://charm.cs.illinois.edu/gerrit/#/c/3723/ Seonmyeong Bak
05:14 PM Bug #937 (Implemented): [local] entry methods don't set tracing event dependencies
Seonmyeong Bak
03:25 PM Bug #1799: Charm++ program segfaults when an element from a nodegroup proxy is used for message p...
Originally saw this problem from another application using:... Dong Hun Lee
03:17 PM Bug #1799 (Merged): Charm++ program segfaults when an element from a nodegroup proxy is used for ...
In the attached example, when a callback of a reduction for a nodegroup is specified with an element of the nodegroup... Dong Hun Lee
03:08 PM Bug #1798: uFcontext thread support for 64-bit ARM systems
Yeah, I didn't consider ARM64 for boost context libraries. So, I expect it can be fixed quickly. Seonmyeong Bak
03:07 PM Bug #1798 (Merged): uFcontext thread support for 64-bit ARM systems
Jim Philips reported that current charm fails to build because uFcontext threads always picks the 32-bit ARM. There a... Sam White
02:53 PM Feature #1579 (Implemented): Mark CkAbort [[noreturn]]
Needs review. Matthias Diener
02:52 PM Bug #1797 (Merged): ARM64 build and stability issues
When building on ARM64, the multicore-arm build uses 32-bit version of boost-context library, causing build failure. ... Jim Phillips
02:50 PM Feature #1449: AMPI support for MPI_Win_allocate_shared
Not a release blocker Sam White
02:47 PM Bug #1689: CMK_TIMER_USE_PPC64 inaccurate with variable clock speeds
I talked to Bilge to get more information about the timers on POWER and if there was any chance that the timer can be... Ronak Buch
02:45 PM Feature #1730 (Merged): The RTS should set std::set_terminate
Matthias Diener
11:57 AM Feature #1730 (Implemented): The RTS should set std::set_terminate
Patch here: https://charm.cs.illinois.edu/gerrit/#/c/3720/ Matthias Diener
10:27 AM Bug #1796: Support for partitions on netlrts-linux-x86_64-tcp builds
I looked into this further via gdb and debug build. It looks like the process for the second partition hangs in CPU t... Karthik Senthil
08:27 AM Bug #1711: syncft tests: unclear failure
netlrts-linux-x86_64-syncft failed autobuild last night in tests/charm++/jacobi3d:... Sam White

02/13/2018

05:00 PM Bug #937: [local] entry methods don't set tracing event dependencies
Similar to the CthThread and CkLoop tracing issue. (eventID is set 0 )
For local methods (or inline), we may set a...
Seonmyeong Bak
03:31 PM Bug #937 (In Progress): [local] entry methods don't set tracing event dependencies
Seonmyeong Bak
08:25 AM Bug #1764 (Merged): Fix AMPI on mpi-windows builds
Matthias Diener
08:36 PM Feature #1582: DDT support for direct copy from a noncontiguous type to another noncontiguous type
MPI_Types library calls this "transpacking": http://www.mcs.anl.gov/mpitypes/files/2014/07/ross_mpitypes-europvmmpi-2... Sam White

02/12/2018

04:49 PM Bug #1764 (Implemented): Fix AMPI on mpi-windows builds
Patch here: https://charm.cs.illinois.edu/gerrit/#/c/3713 Matthias Diener
01:19 PM Feature #1731: Complete spack installation script
We should discuss if it makes sense to put the package in charm++'s repository, or if we just update the package on s... Matthias Diener
08:44 AM Documentation #1156: Add example showing refnum's on callbacks
The patch above just needs a little work Sam White
08:41 AM Documentation #1583: Check commented old template documentation in doc/charm++/inhertmplt.tex aga...
This needs to be reassigned Sam White

02/11/2018

01:06 PM Bug #1694: Projections shows garbage for indices of 4d, 5d, 6d array elements
Projections needs the corresponding support to display >3 dimensional array indices. Sam White
01:02 PM Bug #1035: Idle PEs compete with comm thread for node queue lock
Since there hasn't been any movement on this recently and it doesn't involve any user-visible API changes, retargetin... Sam White
12:58 PM Bug #1665 (Implemented): DDT needs to reference count and free its type objects
Sam White
12:23 PM Bug #1764 (In Progress): Fix AMPI on mpi-windows builds
Sam White
10:47 AM Bug #1764: Fix AMPI on mpi-windows builds
This seems to be caused by the hwloc patches. In @tmp/conv-mach-opt.sh@, the following lines are added by hwloc confi... Matthias Diener
11:06 PM Feature #1766 (Merged): Add an ampirun script as drop-in replacement for mpiexec
Matthias Diener
10:59 PM Bug #1711: syncft tests: unclear failure
I believe Juan gave an update on these in Core a week or two ago, that he couldn't reproduce them? Sam White
10:55 PM Bug #1689: CMK_TIMER_USE_PPC64 inaccurate with variable clock speeds
That patch was merged, so should this issue be marked 'merged'? If not, maybe retarget to 6.9.1 Sam White
10:53 PM Bug #1671: Verbs memory pool may leak pinned memory when message is deleted on a PE different fro...
We don't really have a pressing need for this in 6.9.0, so deferring Sam White

02/10/2018

11:43 AM Bug #1699 (Feedback): [inline] entry methods should use perfect forwarding in C++11
Nils, can you answer Eric's questions above? Sam White
11:39 AM Feature #1357 (Merged): Add AMPI RMA routine tests to megampi
Sam White
11:16 AM Bug #1696 (Merged): Allow default arguments in group entry methods
Sam White

02/09/2018

05:26 PM Bug #1786 (Merged): Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Sam White
01:08 AM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
https://charm.cs.illinois.edu/gerrit/#/c/3706/ Seonmyeong Bak
01:07 AM Bug #1786 (Implemented): Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Seonmyeong Bak
01:07 AM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
This issue happened because mpi-crayxe/mpi-crayxc set JCONTEXT in conv-mach-smp.h while any other targets doesn't set... Seonmyeong Bak
03:01 PM Feature #1712: Avoid intermediate ctx to scheduler in case of ULTs
I think only function checking the queues in the converse queue efficiently is enough for your purpose. If all the qu... Seonmyeong Bak
02:14 PM Feature #1712: Avoid intermediate ctx to scheduler in case of ULTs
I hadn't thought of that, but it might be good. I meant that if a ULT is running and is about to suspend, then we sho... Sam White
02:07 PM Feature #1712: Avoid intermediate ctx to scheduler in case of ULTs
Maybe, the current implementation tries to follow the priority order of the converse queues.
If we want to process ...
Seonmyeong Bak
09:56 AM Feature #1712: Avoid intermediate ctx to scheduler in case of ULTs
Seonmyeong, can you take a look at this? You have more knowledge of our ULTs and Csd module. The idea here is just to... Sam White
10:04 AM Bug #1634: HDF5 issues in AMPI
Note that Matthias added support for ROMIO compilation as a shared object and confirmed that we can now build HDF5 as... Sam White
09:59 AM Cleanup #1034: Constant comparison warnings from xlC compiler
A Charmworks intern is looking into this. Sam White
10:33 PM Feature #1018: AMPI support for MPI-3's distributed graph interface
https://charm.cs.illinois.edu/gerrit/#/c/1310/ Sam White

02/08/2018

03:07 PM Cleanup #536 (Implemented): Data Races in SMP PCQueue
Implemented in patch series concluding with: https://charm.cs.illinois.edu/gerrit/418
Merge will need to wait for ...
Evan Ramos
02:59 PM Bug #1789 (Implemented): ++oneWthPerSocket doesn't work on Darwin
https://charm.cs.illinois.edu/gerrit/3703 Evan Ramos
09:18 AM Bug #1796 (New): Support for partitions on netlrts-linux-x86_64-tcp builds
The test appears to finish but doesn't actually exit.... Sam White

02/07/2018

05:13 PM Bug #1793 (Merged): Isomalloc breaks in hwloc on multicore-darwin-x86_64
Even though the underlying problem is still there, I'm going to mark this 'Merged' for practical purposes. We'd like ... Sam White
02:27 PM Bug #1793 (In Progress): Isomalloc breaks in hwloc on multicore-darwin-x86_64
workaround: https://charm.cs.illinois.edu/gerrit/#/c/3698/ Sam White
01:53 PM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
It doesn't help Sam White
01:47 PM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
Does this patch help? https://charm.cs.illinois.edu/gerrit/3606
EDIT: Probably not, looking again.
Evan Ramos
01:26 PM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
This is the offending commit: https://charm.cs.illinois.edu/gerrit/#/c/3144/
I'll see if there's an easy fix and o...
Sam White
01:13 PM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
Yeah, I'll submit a patch to do that. I'm doing a git bisect now to see what broke it too. Sam White
12:43 PM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
It sounds like we should always change isomalloc to os-isomalloc on Darwin. Evan Ramos
09:51 AM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
This is doesn't happen on multicore-linux-x86_64{-clang}. It does reproduce on multicore-darwin-x86_64-gfortran-gcc. Sam White
09:37 AM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
'-memory os-isomalloc' still works. charmc already changes isomalloc to os-isomalloc on Clang non-SMP, so I think we... Sam White
05:12 PM Bug #1792 (Merged): AMPI failing occasionally during migration in PUP of MPI_Info objects
Sam White
02:56 PM Bug #1786 (In Progress): Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Sam White
02:55 PM Bug #1332: assumes remote shell is OpenSSH
Removing version tag because of the minor nature of this issue. Waiting on feedback before proceeding with implementa... Evan Ramos
02:54 PM Feature #789: Charmrun should test for SSH failures when node programs fail to launch
Removing version tag because I consider this issue satisfied, but am waiting on feedback before marking it closed. Evan Ramos
02:52 PM Feature #1173: Automatic process launching, thread spawning, and hardware binding
Removing the version tag from this meta-task since the remaining subtasks have not been discussed for inclusion with ... Evan Ramos
02:30 PM Cleanup #1311 (Closed): Align XL-specific conditional compilation TRAM to relevant versions
Ronak Buch
02:05 PM Feature #1730 (In Progress): The RTS should set std::set_terminate
Matthias Diener
01:52 PM Cleanup #1795 (New): Remove cc-cray 128 bit defines
In 191caa0430b1e350f1ea404c6917b92b5444742a, defines were added to disable 16 byte integers on Cray compilers since t... Ronak Buch
12:09 PM Feature #1766 (Implemented): Add an ampirun script as drop-in replacement for mpiexec
This is implemented and needs some reviews. Matthias Diener
11:45 AM Feature #1731 (In Progress): Complete spack installation script
Sam White
11:45 AM Feature #1731: Complete spack installation script
https://charm.cs.illinois.edu/gerrit/#/c/3638/ Sam White
09:41 PM Documentation #1491 (Feedback): Update GPUManager documentation
Completed first draft. Jaemin Choi

02/06/2018

11:28 AM Bug #1792 (Implemented): AMPI failing occasionally during migration in PUP of MPI_Info objects
Fix here: https://charm.cs.illinois.edu/gerrit/#/c/3637/
Our internal handling of MPI_Info's is pretty ugly, based...
Sam White
11:18 AM Bug #1792: AMPI failing occasionally during migration in PUP of MPI_Info objects
This is showing up in some of the Jenkins builds for gerrit, but it only fails ~33% of the time on my machine. Sam White
11:13 AM Bug #1792 (Merged): AMPI failing occasionally during migration in PUP of MPI_Info objects
On multicore-darwin-x86_64, in tests/ampi/megampi/:... Sam White
11:17 AM Bug #1793: Isomalloc breaks in hwloc on multicore-darwin-x86_64
I suspect the ordering of initialization between Converse memory modules and hwloc processs launch stuff was inverted... Sam White
11:16 AM Bug #1793 (Merged): Isomalloc breaks in hwloc on multicore-darwin-x86_64
On a multicore-darwin-x86_64 build, in tests/ampi/megampi/. You need to remove "-Wl,--allow-multiple-definition" from... Sam White

02/03/2018

11:56 AM Support #1674: Add 'ofi' target to autobuild
Added SMP target to @system_list@ and created @ofi-smp@ folder along with @instead_test.sh@ on Bridges. Jaemin Choi
10:58 AM Support #1674: Add 'ofi' target to autobuild
I added +isomalloc_sync and it passed last night. ALl that is need now is to add an SMP target. Sam White

02/02/2018

08:11 AM Support #1674: Add 'ofi' target to autobuild
We still need +isomalloc_sync. The tests ran last night but failed in an AMPI test that needs that flag. Sam White

02/01/2018

05:43 PM Bug #965: ampicc -swapglobals is broken for ld v2.24+
I have created versions of Phil's patch for all affected releases of ld, and also expanded it to cover i386.
I tes...
Evan Ramos
02:07 PM Bug #965: ampicc -swapglobals is broken for ld v2.24+
Upon closer inspection, the change in question did not make its way into a release until 2.24. Evan Ramos
05:05 PM Bug #1653 (Merged): NeighborLB segfaults during startup in SMP/multicore builds
Sam White
04:42 PM Bug #1791 (Merged): ddt dependence missing in AMPI's makefile
Matthias Diener
02:58 PM Bug #1791: ddt dependence missing in AMPI's makefile
... except that ddt.o does not get rebuilt when ddt.h changes.
Patch here: https://charm.cs.illinois.edu/gerrit/#/c...
Matthias Diener
02:50 PM Bug #1791: ddt dependence missing in AMPI's makefile
This seems to work for me:... Matthias Diener
02:26 PM Bug #1791 (Merged): ddt dependence missing in AMPI's makefile
ddt was moved into AMPI, but there is no dependence on ddt or its header, so if you update DDT and re-make AMPI the b... Sam White
04:40 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Running after compiling with -fsanitize=address gives:
Domain decomposition...SFC Peano-Hilbert
Created 128 piece...
Thomas Quinn
04:18 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
param, data and job file are in:
https://drive.google.com/open?id=1AU7P4hgMrVnRpLBDbFoP4h-wexKxDjwA
Thomas Quinn
02:30 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Try to compile charm with -DCMK_NOT_USE_TLS_THREAD=1.
Tom, can you share the input data so that I can replay your...
Seonmyeong Bak
09:19 AM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
This seems to be a memory corruption issue: I tried a smaller problem, and I got malloc failures:
free(): invalid po...
Thomas Quinn
04:37 PM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
Here's my understanding of the current hwloc-based command line arguments. For a concrete example, let's say we have ... Sam White
12:02 PM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
... Evan Ramos
11:54 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
But if my MacBook has a single socket with 8 PUs, why shouldn't "++processPerSocket 1 ++oneWthPerSocket" launch 1 pro... Sam White
11:43 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
I can see it on Linux with the correct arguments.
What is happening is that one thread per socket is not enough fo...
Evan Ramos
11:42 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
... Sam White
11:37 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
I see the same problem on Darwin. I can't reproduce it on Linux.
I thought this patch would have prevented this er...
Evan Ramos
11:13 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
What happens if you change `++np 1` to `++processPerSocket 1`? Evan Ramos
09:31 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
I'm seeing the same thing on Linux... netlrts-linux-x86_64-smp Sam White
09:16 AM Bug #1789: ++oneWthPerSocket doesn't work on Darwin
Same thing in standalone mode. This is all on a netlrts-darwin-x86_64-smp build. Sam White
09:15 AM Bug #1789 (Merged): ++oneWthPerSocket doesn't work on Darwin
... Sam White
03:30 PM Bug #1790 (Implemented): +oneWthPerHost not implemented in standalone builds
https://charm.cs.illinois.edu/gerrit/3608 Evan Ramos
11:38 AM Bug #1790: +oneWthPerHost not implemented in standalone builds
Yes, but we should accept that argument since it does make sense, and because we should always parse and consume all ... Sam White
11:11 AM Bug #1790: +oneWthPerHost not implemented in standalone builds
I specifically did not add oneWthPerHost to standalone builds (which are all multicore has) because host count is not... Evan Ramos
09:26 AM Bug #1790 (Merged): +oneWthPerHost not implemented in standalone builds
... Sam White
03:28 PM Bug #1567 (Merged): [aggregate] entry methods should refuse to accept CkEntryOptions, since the c...
Eric Mikida
11:19 AM Feature #1181: Support automated process launch on a single host for standalone runs
https://charm.cs.illinois.edu/gerrit/3558
https://charm.cs.illinois.edu/gerrit/3446
Evan Ramos
11:39 PM Feature #1181 (Merged): Support automated process launch on a single host for standalone runs
Sam White
11:18 AM Bug #1174 (Merged): Use hwloc data from compute host, rather than assuming they're identical to t...
Merged: https://charm.cs.illinois.edu/gerrit/3446 Evan Ramos
11:40 PM Bug #1174: Use hwloc data from compute host, rather than assuming they're identical to the host r...
Please post which gerrit patch this was implemented in and then mark 'merged' Sam White
11:18 AM Feature #973: multicore: spawn a thread per core by default
https://charm.cs.illinois.edu/gerrit/3565 Evan Ramos
11:41 PM Feature #973 (Merged): multicore: spawn a thread per core by default
++autoProvision provides this behavior, but we decided that the default of no +p option should remain +p1 Sam White
11:17 AM Feature #1178: Support automated launch/spawn/bind when using charmrun ++mpiexec
https://charm.cs.illinois.edu/gerrit/3446 Evan Ramos
11:43 PM Feature #1178 (Merged): Support automated launch/spawn/bind when using charmrun ++mpiexec
Sam White
11:17 AM Feature #1176 (Merged): Detect unsupported non-uniformity of processes/threads in charmrun, and e...
Merged: https://charm.cs.illinois.edu/gerrit/3446 Evan Ramos
11:44 PM Feature #1176: Detect unsupported non-uniformity of processes/threads in charmrun, and error
Please post which gerrit patch this was implemented in, and then marked 'merged'. Sam White
11:17 AM Feature #1180 (Merged): Support automated thread spawn/bind for standalone runs
Merged: https://charm.cs.illinois.edu/gerrit/3558 Evan Ramos
11:45 PM Feature #1180: Support automated thread spawn/bind for standalone runs
Please post which gerrit patch this was implemented in, and marked 'merged'. Sam White
08:44 AM Bug #1035: Idle PEs compete with comm thread for node queue lock
One proposed solution: instead of having one multi-producer/single-consumer queue per PE that the comm thread and all... Sam White
08:32 AM Bug #802: CUDA examples broken on Blue Waters
Bump, what is the status of this issue? Sam White
08:27 AM Bug #1785 (Closed): The manual isn't building correctly / updating online
Marking 'Closed' so redmine will remove it from the 'Open' category Sam White
11:42 PM Support #126 (Closed): Document process launching arguments with aim of cross-machine rationaliza...
hwloc launch parameters obviate the need for this Sam White
07:34 PM Cleanup #1311: Align XL-specific conditional compilation TRAM to relevant versions
I think this has already been fixed in https://charm.cs.illinois.edu/gerrit/#/c/3190/, but my latest changes fix impr... Ronak Buch

01/31/2018

04:30 PM Bug #1699: [inline] entry methods should use perfect forwarding in C++11
After some more exploration, I'm not sure why this change is needed. Where is the extra copy you are referring to hap... Eric Mikida
06:28 PM Bug #1699: [inline] entry methods should use perfect forwarding in C++11
Nevermind. The restriction I mentioned was due to a different change for marshalling objects without default construc... Eric Mikida
06:12 PM Bug #1699: [inline] entry methods should use perfect forwarding in C++11
Nils, unless I'm misunderstanding something, the above example would still either require that the objects function t... Eric Mikida
03:37 PM Bug #1785: The manual isn't building correctly / updating online
This was caused by /web on panache not being mounted properly for some reason. Running sudo mount /web fixed it. Ronak Buch
03:18 PM Bug #1785 (Resolved): The manual isn't building correctly / updating online
Matthias Diener
03:18 PM Bug #1785: The manual isn't building correctly / updating online
The issue has been fixed on panache. The above patch was not necessary. Matthias Diener
03:16 PM Support #1674: Add 'ofi' target to autobuild
Actually the problem this time doesn't seem to be caused from @+p1@; the command that causes the hang is @../../../bi... Jaemin Choi
02:25 PM Support #1674: Add 'ofi' target to autobuild
The issue of "There seems to be an issue with the OFI build that @+p1@ passed to an application is regarded as @argv[... Jaemin Choi
09:19 AM Support #1674: Add 'ofi' target to autobuild
Once we get the non-SMP build running, we'll want to add a second target that is SMP Sam White
02:42 PM Feature #1039 (New): reject pemap/commap with duplicate or too few cpus
Evan Ramos
09:46 AM Feature #1039: reject pemap/commap with duplicate or too few cpus
I'd say we can close this issue or push it off to a future release. The new ++oneWthPer* routines should obviate the ... Sam White
02:04 PM Bug #1696 (Implemented): Allow default arguments in group entry methods
Eric Mikida
01:11 PM Bug #1567 (Implemented): [aggregate] entry methods should refuse to accept CkEntryOptions, since ...
From what I can tell, since we have C++11 support, using the delete keyword seems to be the more standard way of acco... Eric Mikida
12:17 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
We're seeing a possibly related topo issue with larger scale runs in OpenAtom on BlueWaters, still trying to narrow d... Eric Bohm
09:33 AM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
It would be good to run this under Valgrind and see if anything comes up from the Cth* or *context routines Sam White
11:33 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
We don't have access to Piz Daint, but Edison at NERSC is similar. Note the use of 'PrgEnv-gnu', 'gcc/5.3.0', and 'cr... Sam White
09:42 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Machine: "Piz Daint".
Modules:
Currently Loaded Modulefiles:
1) modules/3.2.10.6
2) eproxy/2.0.16-6.0.4.1_3.1...
Thomas Quinn
08:49 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
It's weird that this would show up in mpi-crayxc-smp but not gni-crayxc-smp. What machine is this on? It would be nic... Sam White
08:41 PM Bug #1786: Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
Change https://charm.cs.illinois.edu/gerrit/#/c/3482/
seems to have introduced this bug.
Thomas Quinn
12:04 PM Documentation #1787: AMPI+OpenMP documentation & example
This will be assigned to a Charmworks intern who is starting work on AMPI+OpenMP now. Sam White
12:04 PM Documentation #1787 (New): AMPI+OpenMP documentation & example
The modified LLVM OpenMP runtime that we package with Charm++ is documented in the Charm++ manual section 22.2, but n... Sam White
11:24 PM Bug #1501 (Merged): charmxi fails to deal with rvalue references
Eric Mikida
11:23 PM Bug #676 (Merged): Compiling ci files that use "typename T" in argument list fails
Eric Mikida
11:23 PM Bug #1646 (Merged): Support use of std::array in .ci files
Eric Mikida
11:22 PM Feature #1532 (Merged): Example for using Google benchmark library with Charm++
Eric Mikida
11:21 PM Bug #1782 (Merged): charmrun ++local dies without error message on non-existent program
Eric Mikida
10:48 PM Bug #1227: Support template entry methods in generated code from TRAM [aggregate] entry method at...
Deferring. The current workaround still exists, this feature is not heavily used, and would require a decent amount o... Eric Mikida

01/30/2018

03:58 PM Bug #1786 (Merged): Assertion "thisDim < thatDim" failed in file cklocation.C line 2880
So far, this only happens when I @"build ChaNGa mpi-crayxc smp -O2"@. (A gni-crayxc smp build works.)
A print state...
Thomas Quinn
03:15 PM Bug #1785: The manual isn't building correctly / updating online
Untested patch here: https://charm.cs.illinois.edu/gerrit/#/c/3591/ Matthias Diener
02:36 PM Bug #1785: The manual isn't building correctly / updating online
The build error is:... Matthias Diener
01:44 PM Bug #1785 (Closed): The manual isn't building correctly / updating online
The Charm++ manual and other manuals are not building correctly since 10 days ago according to Jenkins. This means th... Sam White
02:29 PM Bug #1699 (In Progress): [inline] entry methods should use perfect forwarding in C++11
Yeah I forgot to mark in progress. Have the implementation (without C++11 guards as mentioned) just had to test it an... Eric Mikida
01:53 PM Documentation #1781: Clean up incorrect quotation marks in the manuals
What software is generating the online manuals from LaTeX (and doesn't know how to handle quotes)? Jim Phillips
12:29 PM Support #1674: Add 'ofi' target to autobuild
The build works, but then the jobs are pretty consistently timing out for whatever reason now:... Sam White
 

Also available in: Atom