grantfirl
commented
5 months ago I think that this problem still exists. Although discovered on Cheyenne, which is now decommissioned, I think the underlying problem has not been addressed. I don't know who is familiar enough with the problem to fix it though.
Description
The RT tests on Cheyenne for cpld_control_p8, cpld_control_noaero_p8, cpld_control_nowave_noaero_p8 and atmaero_control_p8 with the default settings for both INTEL and GNU. The current solution for these failed tests on Cheyenne is to use more memory for each core. The possible cause for this issue probably hides in the physics code to process the new ice climatology data, e.g., subroutine SFCCYCLE.