mvertens

@adagj @maritsandstad @MichaelSchulzMETNO @oyvindseland @gold2718 @TomasTorsvik @monsieuralok - I have made progress in tracking down the memory leak. I am using the following ``` --compset 1850_CAM70%LT%NORESM%CAMoslo_CLM60%SP_CICE%PRES_DOCN%DOM_SROF_SGLC_SWAV_SESP --res ne16pg3_ne16pg3_mtn14 ``` The...

@oyvindseland - thanks for the update. It's not only this file it turns out and the situation is totally perplexing. Here is where I am now. - I have run...

A few more details to add to the confusion. - I ran the same test with gnu rather than intel (in non-debug mode) and the results look different. I see...

Compiling with DEBUG mode using intel memory jumps like above still happen - just not at the same dates as with non-debug mode.

I managed to repeat the ne16 OsloAero run on derecho. - the memory jump for CH4_CHML occurs - but behaves different than on betzy It goes down again in the...

@adagj - I was not able to track this down after extensive work. Looking at some issues for CTSM on derecho- the following can me done to help track this...

@TomasTorsvik @mvdebolskiy @monsieuralok @gold2718 - we need a new pelayout for this configuration rather than P253.

My concern about the current implementation is that having clm continue to read in ndep in non-WACCM configurations might lead to cases where the ocean and land read in different...

Looking at the CTSM code in more detail I now realize that datm is not providing all the scenarios that CTSM needs. That said - CTSM should always receive ndep...