CGNR iterative solver with -no_reduced_fft fails miserably in MPI mode

yogevb / a-dda

Automatically exported from code.google.com/p/a-dda

0 stars 0 forks source link

CGNR iterative solver with -no_reduced_fft fails miserably in MPI mode #174

Open GoogleCodeExporter opened 9 years ago

GoogleCodeExporter commented 9 years ago

I have accidentally found this bug while running automatic tests. It is present 
down to at least v.0.78 (the earliest Windows MPI executable that is available).

However, the combination of parameters is rather weird to seriously worry about.

Original issue reported on code.google.com by yurkin on 24 Sep 2013 at 4:25

Blocked on: #177

GoogleCodeExporter commented 9 years ago

Issue 57 has been merged into this issue.

Original comment by yurkin on 25 Sep 2013 at 6:54

GoogleCodeExporter commented 9 years ago

r1263 adds an explicit exception for this case. However, making such a weird 
combination operational requires significant effort - rearranging of memory 
distribution memory distribution should be changed to have both x and gridX-x 
on the same processor. This is of low priority.

Original comment by yurkin on 25 Sep 2013 at 9:55

Added labels: Priority-Low
Removed labels: Milestone-1.3, Priority-Medium
Now blocked on: #177