D4Science Infrastructure

Tests have started for this huge computation.
I expect the computation time to be exponential descendant.
A linear estimate of the computation time is 4 days.

As it happened also in other cases (e.g. the Length-Weight algorithm), the effect of parallelising an R script is to reduce the computational time more than linearly.
With the latest input provided by FAO, the computation time of the sequential run is around 30 days.
Using 60 nodes, instead, the lower usage of memory and disk has the effect to reduce the computation time to 15h and 20 minutes.

Thus, the time reduction with respect to the sequential case is 97.8%

I have run the computation two times to double-check the time and the output.

The execution produces the following output files:
Main output: http://goo.gl/bJ1ZRx
Auxiliary output: http://goo.gl/slPIwA

In the list of the 5565 input species, there are 49 species on which the script crashes. This requires further investigation by FAO, in order to produce a patch or to discard these species.
The list of the 49 species records is here: http://goo.gl/kAPPaA

I will update this ticket as soon as FAO will have answered on how to proceed for the species without output.

The results have been sent to Yimin Ye of FAO.