Experimenting with Rblas.dll
On Tue, 12 Nov 2002, Kenneth Cabrera wrote:
Dear R-List: I was experimenting with ATLAS to compile Rblas.dll again. My processor is a PIII (x86 Family 6 Model 8 Stepping) in W2K. I compile R from the sources with all the recommended tools, no problem. I remake Rblas.dll from sources with cygwin.
^^^^^^^^^^^ Wrong compiler, if you really used cygwin. On small problems very small changes in the compiler or the problem can affect the speed comparisons. With large problems ATLAS wins hands down. Very few things people do with R are actaully affected significantly by the BLAS speed.
When I use the CRAN compiled version (rw1061.exe) its performance (I use kriging function on GeoR) is very similar to the one I compiled. I then recompile my version using the Rblas I built, now the performance was down in relation with the CRAN compiled version. I experiment also with the CRAN /ATLAS Rblas.dll precompiled version. But the thing that I don't understand is that if I use the Rblas.dll form the CRAN version (the one in the rw1061.exe version) on my compiled version, it is faster!!! than the CRAN version and of course my compiled version both with and without the Rblas recompiled. I wonder ... why? I would like to know more about your own experiences in this issue. Thank you very much! Kenneth Cabrera -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
Brian D. Ripley, ripley at stats.ox.ac.uk Professor of Applied Statistics, http://www.stats.ox.ac.uk/~ripley/ University of Oxford, Tel: +44 1865 272861 (self) 1 South Parks Road, +44 1865 272860 (secr) Oxford OX1 3TG, UK Fax: +44 1865 272595 -.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.- r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html Send "info", "help", or "[un]subscribe" (in the "body", not the subject !) To: r-help-request at stat.math.ethz.ch _._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._