-
Notifications
You must be signed in to change notification settings - Fork 55
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
DOF with holes does not read correctly with mpi-serial #1277
Comments
@edhartnett is there a test of read with holes in the repo? So far unable to reproduce this problem and think it may be netcdf version. @gold2718 the netcdf version is listed in env_mach_specific |
@gold2718 I am not able to reproduce the problem as you describe it above, although I can reproduce the original cam-se problem. |
I figured out a test case, the problem is in mpi-serial and I will report it there. |
Moved to MCSclimate/MCT#53 |
Moved to ESMCI/mpi-serial#11 |
I'm currently running into this on Hobart with a current (very close to master) version of CIME. I'm not sure how to get the PIO or NetCDF version in this situation.
The issue is doing a read_darray in serial mode seems to leave out the zero DOF entries so that the destination array has packed values.
With 1 task using MPI, things work fine and get the same answer as 48 tasks. The DOF arrays being passed to pio_initdecomp are the same but the read-in arrays are different.
As an example:
I have an array initially set to -1
The DOF is
1, 2, 3, 0, 4, 5, 0, 6, 7
The values on file are `1, 2, 3, 4, 5, 6, 7'
With holes, the correct read is:
array = 1, 2, 3, -1, 4, 5, -1, 6, 7
that is, the positive numbers were read into the correct places. In serial mode, I get:
array = 1, 2, 3, 4, 5, 6, 7, -1, -1
I see this behavior with both PIO1 (Nag) and PIO2 (PGI) on Hobart.
The text was updated successfully, but these errors were encountered: