track cpu and memory usage

[Ohm-Np]

% of CPU and memory usage

Variables

• populaton count (raster)
• ecoregions (vector)

1. track for one hour (60601)
2. plot the graphs

[Ohm-Np]

To add:

• track while processing via mapme.biodiversity
• track while processing via stand-alone script

Also

• track while carrying out batch processing
• track while using single core (sequential)
• track while using multiple cores (parallel)

[Ohm-Np]

Example plot:

Inference from the plot

During the 1 minute (60 secs) of monitoring CPU and memory usage, the processing using 8 cores (worldpop raster for WDPA) seem to consume almost 88/90% of CPU usage while memory consumption has been pretty low which is ~1% of total memory.

[Ohm-Np]

Example GFW

Inference from the plot

As we can see, the gfw processing for the first 100 polygons of Latin America WDPA uses almost 90 percent of CPU at first but as the chunks get divided among the different cores, it starts to consume less CPU and remains constant at around 50%. However, the memory usage seems to fluctuate a lot during the processing time and reaches upto the maximum of ~30% at time.

Another plot using single core

Here we can see that, using single core in comparison to 14 cores has more CPU usage as well as more memory usage.

From the above two graphs, we can see that the parallel processing approach actually helps reduce memory issues in the R server.