[R-SIG-Mac] R multi-cpu and 64 bit

[Simon Urbanek]

David,

On Nov 2, 2006, at 7:25 AM, David Airey wrote:

> R as is just great for me. I also use Stata 9 which is also a great  
> environment. I'm not a statistician. I will be in the market for a  
> new computer in January, and our data sets are getting very large-- 
> microarrays. Certainly, R has much more development via  
> Bioconductor in this area. But I think it will be interesting to  
> compare performance on something like the pending 8 cpu Mac Pro  
> with the multiple CPU 64 bit version of Stata (only a grant can  
> afford that), with R. So here's my question. Is R already capable  
> of taking advantage of more than one core?
>

Yes, the CRAN R binary build for Macs uses vecLib which uses multiple  
cores (since OS X 10.4.8), so for operations involving BLAS you get  
parallelization for free. R doesn't use any other means of  
parallelization internally. As of using R for parallel computing, as  
Elijah was pointing out you may want to use snow (snow+rpvm is  
available in binary form from CRAN).

Your question didn't involve 64-bit, but your subject suggests so -  
there is currently no 64-bit Intel binary of R available from CRAN,  
but an experimental build is likely to appear soon.

Cheers,
Simon