| |||||||||||||
Oracle Broadens Support for Open-source R Analytics (PCWorld)Oracle Broadens Support for Open-source R Analytics (PCWorld)Posted Aug 27, 2012 15:26 UTC (Mon) by itoen (guest, #86424)Parent article: Oracle Broadens Support for Open-source R Analytics (PCWorld)
It's debatable whether linking against MKL and ACML will make much of a difference for R programs that are not spending their majority of time in matrix multiplication or matrix decompositions (ie. non-trivial code that does a lot of data manipulation). A big problem with R is that it's an interpreted language. There is a "compiler" for it, but it merely compiles the code into form that is a bit faster to interpret. The issue is so pronounced that elaborate bridges between R and C++ have started to become popular, eg. the Rcpp package, allowing bidirectional links with C++ matrix libraries such as Armadillo. If Oracle is serious about speeding up R, they should really look into making a JIT for R, rather than making noise about trivialities such as swapping out standard LAPACK for MKL.
(Log in to post comments)
|
|||||||||||||