mpi: Support data assignment to SparseFunctions from local array with different distribution#2951
mpi: Support data assignment to SparseFunctions from local array with different distribution#2951FabioLuporini wants to merge 7 commits into
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #2951 +/- ##
==========================================
- Coverage 83.47% 83.05% -0.43%
==========================================
Files 249 249
Lines 52276 52631 +355
Branches 4503 4531 +28
==========================================
+ Hits 43638 43713 +75
- Misses 7880 8157 +277
- Partials 758 761 +3
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
FabioLuporini
force-pushed
the
sparse-data-mpi
branch
2 times, most recently
from
6ce82fb to
9a9b51c
Compare
FabioLuporini
force-pushed
the
sparse-data-mpi
branch
from
76378a5 to
0b87d5d
Compare
|
|
||
| def _mpi_advanced_1d_target(self, glb_idx, axis): | ||
| """ | ||
| Return the raw local ndarray addressed by ``glb_idx`` without |
There was a problem hiding this comment.
Nitpick: This seems out of step with the function name?
| """Raise the first error reported by any rank, on every rank.""" | ||
| if data._distributor.nprocs > 1: | ||
| errors = data._distributor.comm.allgather(error) | ||
| error = next((i for i in errors if i is not None), None) |
There was a problem hiding this comment.
Is it worth reporting the rank(s) that raised the error here?
| assert trace_data.flags.f_contiguous | ||
|
|
||
| rec.data_local[:, :] = 0 | ||
| rec.data[:, trace_index] = trace_data |
There was a problem hiding this comment.
Can you do the same with rec.data.coordinates?
There was a problem hiding this comment.
you cannot use advanced indexing like this here in objects that have more than 1 distributed dimension (explained in the notebook it's a current limitation)
but what would it be for?
2 participants
No description provided.