From 08/26/2018 to 09/24/2018
- 03:22 PM Cleanup #1980 (Implemented): Remove the old RDMA API (CkDirect/CmiDirect) from source code and li...
- 03:12 PM Feature #195 (Implemented): Modify old charm++/converse onesided/direct API and examples to new r...
- 02:33 PM Feature #195 (In Progress): Modify old charm++/converse onesided/direct API and examples to new r...
- Examples to be removed/modified:
- 01:33 PM Feature #1667: Direct API for nocopy operations on sender-side and receiver-side
- We could potentially use sendmsg and WSASend on netlrts-tcp as Juan did here, though I don't think it should be a pri...
- 01:28 PM Feature #1667 (Merged): Direct API for nocopy operations on sender-side and receiver-side
- Netlrts will not have a true "zero copy" implementation, it'll use the generic copy based implementation.
- 01:30 PM Feature #1919 (Merged): Zerocopy Direct API: Allow user to pass an arbitrary reference pointer in...
- Was merged on 07/03/2018, but the redmine issue was not updated.
- 01:29 PM Feature #1459 (New): Zero-copy send support for the netlrts machine layer
- 01:20 PM Bug #1960: tests/charm++/megatest hangs on netlrts-win-x86_64-smp build
- Last night it got through megatest but not simplearrayhello. So there is something peculiar going on in our windows ...
- 01:12 PM Feature #1834: Functionality to Broadcast large buffers using the nocopy API
- Since the broadcast API was initially implemented in the entry method style, the design scheme discussed before was n...
- 01:07 PM Feature #1657 (Implemented): CMA support for nocopy sends using the Entry Method API across proce...
- By implementing the Entry Method API internally using the Direct API, the Entry Method API internally gets CMA suppor...
- 01:04 PM Feature #1468 (Implemented): Enable pre-pinning memory for the zero-copy message sends through th...
- By implementing the Entry Method API internally using the Direct API, the Entry Method API can be used for passing pr...
- 08:36 PM Feature #1971 (Closed): Explore use of non-temporal stores in the RTS
- 07:51 PM Bug #1978 (Implemented): Segmentation fault in examples/charm++/arraysection and cleanup
- 2D section example: https://charm.cs.illinois.edu/gerrit/#/c/charm/+/4568/
- 05:59 PM Feature #13: CkCallback to section breaks
- There is a test case for which it seems to break.
A section proxy for a 2D array consisting of 1D elements [0, *]:...
- 12:29 PM Feature #13: CkCallback to section breaks
- CkCallback's to sections are still non-migratable, but at least work on the same PE they are created on now: https://...
- 10:25 AM Bug #1978: Segmentation fault in examples/charm++/arraysection and cleanup
- Yes, it works for me. Will add a 2D array example.
- 08:26 AM Bug #1978: Segmentation fault in examples/charm++/arraysection and cleanup
- The example builds and runs correctly now for me. Can you try it, and then add a 2D example?
- 08:20 AM Bug #1978 (In Progress): Segmentation fault in examples/charm++/arraysection and cleanup
- Cleanup here: https://charm.cs.illinois.edu/gerrit/#/c/charm/+/4566/
Still need a 2D example...
- 09:25 PM Bug #1978: Segmentation fault in examples/charm++/arraysection and cleanup
- Some things to clean up:
1. It doesn't compile on my Mac using GCC:...
- 09:18 PM Bug #1978 (Merged): Segmentation fault in examples/charm++/arraysection and cleanup
- Segmentation fault when the example in
charm++/arraysection is run.
On a different system setup it runs into an er...
- 10:21 AM Bug #1977 (Closed): Broadcast on a 2D array section segfaults
- CkMulticast library needs to be used for a section broadcast.
A message inherited from CkMcastBaseMsg needs to be i...
- 09:15 PM Bug #1977 (Closed): Broadcast on a 2D array section segfaults
- Segmentation fault when a section created using 2D array elements is either used in creating a callback or used for b...
- 10:30 AM Bug #1976 (New): ci files with extensive code blocks result in incorrect line numbering in synta...
- unclear exactly what trips up the translator, but I've found the charmxi emitted line numbers for errors to be garbag...
- 12:14 AM Support #1761: Update GPU Manager Tracing API
- New patch using CUPTI Activity API: https://charm.cs.illinois.edu/gerrit/#/c/charm/+/4560/
- 01:41 PM Feature #1972: Tests/examples for non-threaded Charm++ programs using Isomalloc
- We also need to document the interface in the manual somewhere too, once we have tests and examples.
- 12:56 PM Bug #1975 (Merged): forced dynamic linking on Cray breaks build on Beagle
- 10:22 AM Feature #1742: Send large readonly variables using the Nocopy API to reduce memory footprint and ...
- This would be a good first use case for the zero copy broadcast API
- 06:08 PM Bug #1975 (In Progress): forced dynamic linking on Cray breaks build on Beagle
- 04:39 PM Bug #1975: forced dynamic linking on Cray breaks build on Beagle
- FYI, the workaround is "setenv PE_PKGCONFIG_LIBS cray-rca:$PE_PKGCONFIG_LIBS".
- 04:38 PM Bug #1975: forced dynamic linking on Cray breaks build on Beagle
- The dynamic linking failure on Beagle is an old fault with a known work-around from Blue Waters and Titan.
We still ...
- 04:38 PM Bug #1975: forced dynamic linking on Cray breaks build on Beagle
- Is adding @-lrca@ to the end of that @CMK_LIBS@ line sufficient to resolve the link failure?
- 04:16 PM Bug #1975 (Merged): forced dynamic linking on Cray breaks build on Beagle
- This change forces dynamic linking on Cray:
On Beagle at...
- 03:54 PM Support #1512: Measure test coverage
- Currently tracked over time here: http://charm.cs.illinois.edu/autobuild/coverage/rrdimages/
- 03:49 PM Feature #1974 (New): nocopy accelerated section multicast
- It should be possible to reduce the number of copies required to implement a section multicast.
Especially if the ...
- 03:13 PM Bug #1936: pose.decl.h is not found when using pose.h in the BigNetSim emulator
- BigNetSim should be deprecated as its functionality has been subsumed in Charades and Tracer.
- 02:29 PM Feature #1973 (New): GPU Messages
- Enable Charm++ to send RDMA-style messages to/from buffers on the GPU. Our initial plan is to use the RDMA syntax. Th...
- 03:15 PM Feature #1972 (New): Tests/examples for non-threaded Charm++ programs using Isomalloc
- https://charm.cs.illinois.edu/gerrit/4472 adds support for arbitrary Isomalloc contexts. We should have tests and exa...
- 05:16 PM Bug #684: Isomalloc PUP has invalid read memory errors
- My initial suspicion after briefly looking at this is that it results from a logic error somewhere in the mempool's h...
- 05:11 PM Bug #684: Isomalloc PUP has invalid read memory errors
- For reference, the line at @CmiIsomallocBlockListPup (isomalloc.c:2661)@ when this issue was created is: @pup_bytes(p...
- 04:27 PM Feature #1971: Explore use of non-temporal stores in the RTS
- Interesting article on non-temporal stores: https://vgatherps.github.io/2018-09-02-nontemporal/
- 04:26 PM Feature #1971 (Closed): Explore use of non-temporal stores in the RTS
- Non-temporal stores could be a useful performance optimization for things like in-memory checkpoints and tracing, whe...
- 12:44 PM Bug #1970 (Merged): libmemory- broken without production on BGQ
- 10:08 PM Bug #1970 (Implemented): libmemory- broken without production on BGQ
- 03:33 PM Bug #1035: Idle PEs compete with comm thread for node queue lock
- Is there a reproducible test program that exhibits this issue, so that we can get someone working on this.
- 10:36 AM Feature #1425: Virtualization-aware AMPI collectives
- Some notes: Charm++ already has virtualization-aware broadcast and reductions that do process-level aggregation and u...
- 10:33 AM Bug #1155: AMPI's non-blocking collectives are not sequenced
- C++17's std::variant should make implementing support for sequencing/matching of non-blocking collective messages eas...
- 10:06 AM Feature #1386: ckDestroy for Groups and NodeGroups
- Jozsef's original email to the Charm mailing list was IIRC more interested in how to best free the memory associated ...
- 05:03 PM Feature #1386 (Rejected): ckDestroy for Groups and NodeGroups
- For @CProxy_MyChareArray array;@, calling @array.ckDestroy();@ only serves to broadcast a call to @ArrayElement::ckDe...
- 03:38 PM Bug #1970: libmemory- broken without production on BGQ
- This must have been exposed by https://charm.cs.illinois.edu/gerrit/#/c/charm/+/4505/3/src/scripts/Makefile but I don...
- 11:25 PM Bug #1464 (Closed): CUDA example programs hang when run with 1 PE
- Does not appear in current build. May need to look at this if we ever use CUDA callbacks again, but closing for now.
- 11:23 PM Bug #1462 (Closed): Programs hang at startup with CUDA build
- No longer seeing the hangs/crashes.
Also available in: Atom