Difference between revisions of "WRF performance"

 
Line 13: Line 13:
 
In the legend, "T" indicates the number of threads (OpenMP).
 
In the legend, "T" indicates the number of threads (OpenMP).
  
−
One can see for this configuration, the best option is to run with 432 processors, 2 OpenMP threads and 1 I/O group with 2 tasks. This gives the best walltime at one of the lowest cost in SU. Unfortunately, it seems simulations with I/O quilting aren't very stable (this doesn't show in those short tests) so please choose the configuration you think is best for you.
+
One can see for this configuration, the best option is to run with 624 processors and 2 OpenMP threads. This gives the best walltime at one of the lowest cost in SU. I/O quilting does not seem to be worth using in this case.
  
−
[[File:WRF scaling v4.1.3.png|1000px|center]]
+
[[File:WRF_v4.1.3_SU_per_model_day.png|1000px|center]]
 +
[[File:WRF_v4.1.3_Walltime_per_model_day.png|1000px|center]]

Latest revision as of 22:24, 22 January 2020

Configuration

We tested the performance of WRF with the CORDEX domain with a nest over NSW, NARCliM project setup.

It is possible to look at WRF performance with other domains as long as those are used by a group of people and are used over a significant period of time if possible. Please contact cws_help@nci.org.au to discuss the possibility.

The configuration tested can be found at /g/data/sx70/wrf-scaling.

Results

Below is a chart showing the walltime (WT) and the SU cost per model day for different number of processors, threads and I/O quilting options.

The chart shows the data for Gadi and some for Raijin as a comparison.

In the legend, "T" indicates the number of threads (OpenMP).

One can see for this configuration, the best option is to run with 624 processors and 2 OpenMP threads. This gives the best walltime at one of the lowest cost in SU. I/O quilting does not seem to be worth using in this case.

WRF v4.1.3 SU per model day.png
WRF v4.1.3 Walltime per model day.png