Ecosyste.ms: Timeline
Browse the timeline of events for every public repo on GitHub. Data updated hourly from GH Archive.
rfj82982 pushed 19 commits to UpdateTests xcompact3d/Incompact3d
- fix bug #294 197f56c
- Merge pull request #316 from xcompact3d/fix_ibm3 fix bug #294 (ibm out of memory problem with iibm=3) cd1f8d6
- Move stretching subroutines to a dedicated module and fix undefined ypraf values (#314) * Move stretching suborutine... afdf902
- Add DEBG macro to DEBUG builds DEBUG builds will add both the DEBG and DEBUG macros bb3183f
- Make CMake add -cpp 41e301a
- Apparently GNU doesn't support -g3, replace with -g 33f59ab
- The poisson module needed to use decomp_2d_mpi : nrank However refinement_matrice was already using the variable wit... bb2a111
- adding mpi_in_place for allreduce operations d7e54ad
- Merge pull request #317 from xcompact3d/pbartholomew08-patch-1 Add DEBG macro to DEBUG builds 1bfe7e5
- mpi_in_place implementation in allreduce 501011b
- finishing implementation of MPI_IN_PLACE 30e0c98
- small typo fixed in acl_source 15d27fa
- small bug for iterative Poisson solver a3c4fd0
- Refactor LMN/Poisson solver to compute rho0 explicitly The original code with rho0 computation "hidden" was not clea... a3a93b1
- Fix isnan function using ieee intrinsic be33fa9
- Merge pull request #318 from xcompact3d/fix-allreduce-with-inplace Fix allreduce with inplace c1cc41b
- Merge branch 'master' into 319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit 633935e
- Merge pull request #320 from xcompact3d/319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit ... 6108535
- Merge branch 'master' into UpdateTests 69fb139
rfj82982 deleted a branch xcompact3d/Incompact3d
319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit
rfj82982 closed a pull request on xcompact3d/Incompact3d
Refactor LMN/Poisson solver to compute rho0 explicitly
The original code with rho0 computation "hidden" was not clear, this change makes it explicit, and simplifies the logic. Closes #319
pbartholomew08 pushed 8 commits to 319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit xcompact3d/Incompact3d
- adding mpi_in_place for allreduce operations d7e54ad
- mpi_in_place implementation in allreduce 501011b
- finishing implementation of MPI_IN_PLACE 30e0c98
- small typo fixed in acl_source 15d27fa
- small bug for iterative Poisson solver a3c4fd0
- Fix isnan function using ieee intrinsic be33fa9
- Merge pull request #318 from xcompact3d/fix-allreduce-with-inplace Fix allreduce with inplace c1cc41b
- Merge branch 'master' into 319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit 633935e
rfj82982 closed an issue on xcompact3d/Incompact3d
Perform MPI_ALLREDUCE operations in place when possible
Several all-to-all MPI reductions are using a temporary variable : https://github.com/xcompact3d/Incompact3d/blob/55652680be067db9336161b2a9c873744c637811/src/Case-TGV.f90#L298 However, MPI ...
rfj82982 pushed 7 commits to master xcompact3d/Incompact3d
- adding mpi_in_place for allreduce operations d7e54ad
- mpi_in_place implementation in allreduce 501011b
- finishing implementation of MPI_IN_PLACE 30e0c98
- small typo fixed in acl_source 15d27fa
- small bug for iterative Poisson solver a3c4fd0
- Fix isnan function using ieee intrinsic be33fa9
- Merge pull request #318 from xcompact3d/fix-allreduce-with-inplace Fix allreduce with inplace c1cc41b
pbartholomew08 opened a draft pull request on xcompact3d/Incompact3d
Refactor LMN/Poisson solver to compute rho0 explicitly
The original code with rho0 computation "hidden" was not clear, this change makes it explicit, and simplifies the logic. Closes #319
pbartholomew08 pushed 1 commit to 319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit xcompact3d/Incompact3d
- Refactor LMN/Poisson solver to compute rho0 explicitly The original code with rho0 computation "hidden" was not clea... a3a93b1
pbartholomew08 created a branch on xcompact3d/Incompact3d
319-redesign-computation-of-rhomin-in-lmn-poisson-solver-to-make-it-explicit - Current CPU version of our solver for the Navier-Stokes equations
pbartholomew08 opened an issue on xcompact3d/Incompact3d
Redesign computation of rhomin in LMN Poisson solver to make it explicit
By burying this, it is easy to make changes that appear sensible but break the LMN, see: For consistency with calling convention set `rho0=rhomin` after ALLREDUCE and use `rho0` as...
mathrack created a review comment on a pull request on xcompact3d/Incompact3d
I think the new version is better, no need to fix in my opinion
slaizet pushed 1 commit to fix-allreduce-with-inplace xcompact3d/Incompact3d
- small typo fixed in acl_source 15d27fa
mathrack created a review comment on a pull request on xcompact3d/Incompact3d
As a side remark, MPI_Wtime is returning a double precision number. If accuracy is needed on the timer, the single precision builds should not store the result in a variable defined with `real(myty...
slaizet created a review comment on a pull request on xcompact3d/Incompact3d
I will have a look and fix the typo
mathrack created a review comment on a pull request on xcompact3d/Incompact3d
The other modifications are looking equivalent to the current code. This is probably the only one changing the behavior.
slaizet created a review comment on a pull request on xcompact3d/Incompact3d
possibly a typo on my side :-)
mathrack created a review comment on a pull request on xcompact3d/Incompact3d
This modification is fixing a minor issue ? The code was printing the local value before. Now it is printing the global value obtained after reduction.
slaizet opened a pull request on xcompact3d/Incompact3d
Fix allreduce with inplace
This is to deal with issue #287 I have listed 186 ALLREDUCE operations in the codebase. 181 of them are now using MPI_IN_PLACE for clarity. Only 5 of them are not using MPI_IN_PLACE. I have ...