GitLab

Now redistributions always use Intracomms. Intercomms can be used still. Commented barriers in manager.

Simplified function for preparing data redistribution. New feature for intercomms where they communicate using intracomms when redistributing data. Other minor changes.

Minor bugfix when a node has some cores in use but not all and they will stay after a reconfiguration.

Mino refactor in Merge spawn + P2P redistribution to avoid possible blocking of some parents processes to send data waiting for other parents to start receiving it.

Refactored code scripts to use auxiliary scripts when performing similar steps. Those tasks are getting the number of cores, initial hostlist and initial number of processes. Trace and Valgrind scripts are not fully tested.

Minor bugfix in physical process distribution which did not take completely into account already existing processes for the Merge method.

Bugfix. There where two condition races that could happen if the main thread terminated earlier than the auxiliary thread when using the merge shrink.

Bugfix for Data Redistribution with Merge Expand. Array sizes for parents should always be the total amount of particular processes between both groups. Also a minor refactor in distribution.c

Bug found around Malleability merge shrink. Refactor needed so that procs which will become zombies has to first be confirmed by the user

Fast bugfix in Distribution check with requests. For some reason Testall fails in some cases. Bug encountered in results when shrink merge method used, where a Barrier solves the problem, does not make sense. Both problems should be studied.

Bugfix in realloc for results, where children had not the correct amount of iterations. Another fix for stage time storage.

Hotfix Merge Shrink. Code was trying to create a Mapping for new processes, but no process is created with this method

Hotfix in CommDist for Merge Shrink. Refactor in block distribution, removed not needed variable, added error checking in alltoall and minor changes.

Various bugfixes over the changes in ADR and SDR. Also a fix in results for asynch iterations. Some warning corrections.

WIP with some tests. Adapted check_redistribution to check all distributed data has been completed, not just one set. 4 Bugfixes when using the Merge shrink spawn method. States.h modified for more coherence with strategies values

Bugfix over large quantities (>2GB) for SDR and ADR. Now a set of arrays are allocated until is reached the expected quantity. 2 Minor bugfixes when init first group and when checking if the reconfiguration has ended. Eliminated unused variable in group struct

Refactored Manager and CommDist for non-blocking communications and MPI_Request usage. Two minor bug fixes. RMA asynchronous and Ibarrier strategy are the next steps to develop

Refactor for redistributing data with non-blocking funtions. Modified redistributing data constants names to -RED-. Fixed potential bug over requests as the number of requests used was not clear. WIP over CommDist and Manager to adapt non blocking calls. Seems to be working fine right now but more testing is required.

Configuration refactor to modify AT(Asynch Type) to RM(Redistribution method) and RS(Redistribution strategy)

Added synchronous MPI-RMA redistributions. Works over Baseline and Merge spawn, but has not been tested with all the strategies

Added functionality for asynchronous redistributions with threads only. Support for non blocking calls is not yet fully implemented for Merge method.

Fixed a memory issue related to communicating the results structure. A refactor is needed to send the structure as is expected. Minor bugfixes in Main.c

Hotfix - Some bugs when handling strings have been resolved. There is still an error when performing multiple reconfigurations

Added new functionality for stage processing. Now is possible to perform the stage until desired time stage is reached instead of using a number of operations

Hotfix - Now is possible to perform multiple reconfigurations. IO results need a refactor to show all results.

Hotfix&&WIP. Fixed errors when reconfiguring after the first group of processes. There is still an error related to initialization of stages