Closed koheiw closed 4 months ago
I notice that zapsmall() takes very long time when the similarity matrix is large and dense, so moved this to C++. This and other changes in this branch makes handling of denser similarity matrices a lot more efficient.
zapsmall()
options("proxyC.threads" = 12) mt <- Matrix::rsparsematrix(50000, 10000, 0.001) profvis::profvis( out <- proxyC::simil(mt, use_nan = FALSE) )
@rcannood can you review?
I notice that
zapsmall()
takes very long time when the similarity matrix is large and dense, so moved this to C++. This and other changes in this branch makes handling of denser similarity matrices a lot more efficient.Master
This branch