search

environmentalinformatics-marburg / remote

empirical orthogonal teleconnections in R
GNU General Public License v3.0
1 stars 4 forks source link

speed up the calculation when using function eot() #6

Open Mew-YL opened 3 years ago

Mew-YL commented 3 years ago

For downscaling purpose, I have about 4GB MODIS NDVI data and try to use the eot to downscale GIMMS NDVI data, but the calculation speed is too slow with such large dataset( only used 400MB MODIS NDVI data, I already waited for 5 hours ) and I wonder is there any possible ways to accelerate the calculation?

tim-salabim commented 3 years ago

I don't think there's an easy way to speed things up. eot() is already programmed in C++

Mew-YL commented 3 years ago

Can eot() run in parallel processing? just as mentioned in https://strimas.com/post/processing-large-rasters-in-r/

tim-salabim commented 3 years ago

No, parallel processing is not implemented. It has been a long time since we worked on this, but IIRC parallel processing is not feasible, because every pixel in x needs to be regressed against each pixel in y.