Changes needed for GFS Update 2017?

[meteoadriatic]

But what can be the effect of this parameter on F on model performance??

Not much but I don't understand why it causes errors at all.

[JMeans]

meteo60, are you sure that's what fixed your problem? I thought that setting those to "F" was just a temporary workaround, and now that the new version is out that wasn't necessary. Either way, I still keep having problems with crashes, and I have both parameters set to F. It's driving me crazy. I'm tempted to start messing with the time step, just to try something.

[meteo60]

Yes I am sure that the parameter on F solves the problem for my configuration, I have done tests many times and some runs never work with T whereas with F they work. I specify that when the bug occurs with T it is at the very beginning of the run, less than 5 seconds after the start.
I'm on ARW with an adaptive timestep.
Have you modified real_run.conf in the folder of runs/xxx/conf/ems_run?

On a run that worked as well with T or F I could notice some differences on the output of the model on all parameters but these are not very big differences

[meteo60]

Robert has updated UEMS, the latest version is the 15.99.8. Has anyone tested it?

[meteosciez]

Robert has updated UEMS, the latest version is the 15.99.8. Has anyone tested it?

Running UEMS on CentOS 6.9, I have an "Illegal instruction (signal 4) " error with real_arw.exe of version 15.99.8 !
I put back the previous versions (15.99.5) of real_arw.exe and wrfm_arw.exe , and all is OK now.

[meteo60]

I updated to 15.99.8 and for me it's ok