Question about speeding up regional-scale simulations?

[mengqi]

Hi,

I am working on running over the Midwest (80°W to 105°W, 37°N to 49°N) at 1° spatial resolution, driven by the GSWP3 atmospheric forcing dataset. However, I noticed that running was a time-consuming process. The reason is that current simulations were used with a single processor.

This is my current info about processors:

<entry id="NTASKS">
<type>integer</type>
<values>
<value compclass="ATM">1</value>
<value compclass="CPL">1</value>
<value compclass="OCN">1</value>
<value compclass="WAV">1</value>
<value compclass="GLC">1</value>
<value compclass="ICE">1</value>
<value compclass="ROF">1</value>
<value compclass="LND">1</value>
<value compclass="ESP">1</value>

Thus, I was wondering if I should revise information above. Take an example,
./xmlchange NTASKS_ATM=4,NTASKS_LND=4,NTASKS_ICE=4,NTASKS_OCN=4,NTASKS_CPL=4,NTASKS_GLC=4

The first question is how to set a suitable value, such as 2? 4? or 5?

The second question is in addition to that, what else should I revise?

I am new at CLM5 regional-scale simulation. Could you give me some insights? Thanks!

 

[slevis]

I am not experienced at changing PE layouts, either. However, I have a suggestion:

Create a new case for one of our global resolutions that approximates the size of your regional in terms of the lat x lon (rows and colunns). Then look at the processor layout recommended for that case.