Andrew M. Bradley
Andrew M. Bradley
In possibly just non-scalar-accumulator cases, adjacent team parallel_reduce calls with no application-side data dependencies and thus no application-side team_barrier can lead to the wrong answer nondeterministically. This implies that there...
It's possible for a crash that occurs during restart writing to lead to inconsistent or incomplete rpointer files. While we can't salvage the restart files in general, we can at...
I was looking at an EAMxx history file and noticed a number of the global attributes are potentially misleading. For example, IC and topo files were used in the simulation,...
On two nodes of Frontier (16 GPUs total), the following configuration ``` COMPSET=F2010-SCREAMv1 RES=ne4pg2_ne4pg2 ``` produces the error ``` Kokkos::TeamPolicy< HIP > the team size is too large. Team size...
Compare NetCDF_Fortran_INCLUDE_DIR(S) in these two lines: https://github.com/E3SM-Project/scorpio/blob/3bcd9d37733c3ab0a0ed6d9488d60e028dd5ab9f/src/flib/CMakeLists.txt#L200 https://github.com/E3SM-Project/scorpio/blob/3bcd9d37733c3ab0a0ed6d9488d60e028dd5ab9f/src/flib/CMakeLists.txt#L205 Which is intended?
**Is your feature request related to a problem? Please describe.** view_reduction currently handles just a single Pack as accumulator. The result is that in SHOC, there are two cases where...
The version of EKAT that SCREAM uses has a Kokkos submodule reference that is not a tree. Github message: "This commit does not belong to any branch on this repository,...
See https://github.com/E3SM-Project/scream/issues/2070 for background on TeamVR. Once TeamVR is used in the Hommexx and EKAT tridiag files, the arch-specific code in those files can be removed.
Specific situation: REST_N=nminutes. EAMxx doesn't recognize this and errors out. More generally, CIME expects: ``` 'none', 'never', 'nsteps', 'nstep', 'nseconds', 'nsecond', 'nminutes', 'nminute', 'nhours', 'nhour', 'ndays', 'nday', 'nmonths', 'nmonth', 'nyears',...
In discussing EKAT issue 117, we came across two instances where divide-by-0 avoidance and clipping code lead to a discontinuous function: Example 1: ```F90 relvar(:,:) = 1.0_r8 relvarmax = 10.0_r8...