Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Which way is the more computationally efficient for running wf.exe with one way nesting: (a) by setting feedback to '0' or (b) using ndown.exe?

Ankan

Member
Hello everyone,
I want to run wrf.exe (WRF-Chem) with one-way nesting because of the limitations of my system (a linux server with 378 GB of RAM and 32 processors). I have gone through the WRF user guide, which explained how to do one-way nesting using 'ndown.exe'. Also, I found the following lines on the webpage (WRF Namelist.input Best Prac):
When feedback is off , it is equivalent to a one-way nested run, since nest results are not reflected in the parent domain.
So, my question is: Are both processes equally efficient as far as computational cost is concerned? Or is using 'ndown.exe' faster than setting 'feedback' to '0'?
Can anyone please guide me in this regard? That will be greatly appreciated.
Thanking you.
With regards,
Ankan
 
Ankan,
I am not a wrf-chem expert, and thus my answer is more general for WRF-ARW.
Personally I prefer to run WRF in 1-way nesting mode (i.e., feedback off). In this way the boundary of the child domain is more frequently updated than running ndown, and we expect better results. As for the computation cost, note that the run over the child domain is much more expensive than that over the parent domain. Thereby, we suppose the cost over the parent domain is not a big issue. Also, when your run ndown, you need to first run WRF over a larger domain (similar to the parent domain for 1-way nesting), which takes similar computation time to that for parent domain in 1-way nesting.
 
Hello @Ming Chen,
Thank you so much for your response. Actually I gave a WRF-Chem run and the simulation speed is very slow. That is why I wondered if using ndown.exe can make simulation faster than just setting feedback to '0'. That means, you are suggesting that both way simulation time will be almost same. Right?
Thanking you.
With regards,
Ankan
 
The choice between using feedback = '0' or ndown.exe for running wf.exe with one-way nesting depends on the specific needs and requirements of the simulation. Setting feedback = '0' would prevent the output from a finer-resolution simulation from being used to refine the coarser-resolution simulation, effectively turning off the feedback loop between the two resolutions. This would reduce vsco mod apk the computational load on the system, as there would be less data transfer between the two simulations. However, it would also limit the accuracy of the results, as the coarser-resolution simulation would not be able to benefit from the more detailed information provided by the finer-resolution simulation.
 
Last edited:
Top