[DO NOT MERGE] : WIP: Dump strategy branch in sync with devel #144
Conversation
…NL ATS-1 and CTS-1 machines. Document how to use these two scripts.
…ment starts out as the Cray default of PrgEnv-intel. This change checks for the case where the user has modified their module environment and swaps it back to the case assumed by the build script.
…le sort implementation. Add build script support for a few more CMake variables that were missing and should be availble to users of the build scripts.
…nodes and one for KNL nodes.
Codecov Report
@@ Coverage Diff @@
## devel #144 +/- ##
==========================================
- Coverage 80.52% 76.19% -4.33%
==========================================
Files 123 125 +2
Lines 8543 9104 +561
Branches 1437 1482 +45
==========================================
+ Hits 6879 6937 +58
- Misses 1031 1534 +503
Partials 633 633
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
|
Brought branch current with devel: CI test failing: particle_annotation_restart (Failed) With the skipped test enabled, the same tests fail, as mentioned previously. |
Fixes issue lanl#146 where vectorized hydro_p has an extra momentum update.
src/vpic/dump_strategy.h
Outdated
| #endif | ||
| //#define CHUNK_FLAG 1 | ||
|
|
||
| //#define METADATA_COLL_WRITE 1 |
There was a problem hiding this comment.
Why is this commented out? It should get better performance when enabled.
| // if(!mpi_rank) | ||
| // printf("Write original files /w HDF5! \n"); | ||
| // } | ||
| file_exist_flag = 0; |
src/vpic/dump_strategy.h
Outdated
| //#define HAS_HYDRO_COMP 1 | ||
| //#define HAS_PARTICLE_COMP 1 | ||
|
|
||
| //#define HAS_INDEPENDENT_IO 1 |
There was a problem hiding this comment.
Why is this commented out, so IO is always collective?
|
This implementation seems to have a minimum version of HDF5 requirement, so there should be a check for this in CMake. |
This is Bin's update_strategy work which is in sync with devel. This is proposed to replace the current dump_strategy branch.
A couple of issues:
(1) the perform_restore test was disabled (like in the current dump strategy branch); we might want to enable that again.
(1a) buffered_particle test was disabled because buffered_particle_dump_restart fails, need to fix.
(3) UNVOXEL is not in devel.
ctest passes (minus tests noted)