Running Statistics On Multiple Processors 

 You just bought a state-of-the-art PC with dual processors and yet your model still runs forever? Well, your statistical software is probably not multi-threading, meaning that despite the fact that your computer actually has two processors, the whole computation runs only on one of them. Don’t believe me? Well check your CPU usage, it's probably stuck at 50 percent (or less).  

 You might ask why statistical software doesn't use both processors simultaneously. The fact is that splitting up computations to two or even more processors is a non-trivial issue that many software packages do not accomplish yet. This may change in the near future, however, as the advent of dual processors for regular PCs exhibits increasing pressure on statistical software producers to allow for multi-threading. 

  

  

  
 


 In fact, Stata Corp. has recently released  Stata/MP , a new version of Stata/SE that runs on multiprocessor computers. Their website proclaims that: "Stata/MP provides the most extensive support for multiple-processor computers and dual-core computers of any statistics and data-management package." So this bodes well for Stata users.  

 What’s in it for Non-Stataists? People at S-PLUS told me yesterday that there is "currently an enhancement request to add functionality to S-PLUS that will allow it to use multiple processors. This request has been submitted to our developers for further review." Unfortunately no further information is available at this point.  

 In my favourite software R, there are  efforts  to get concurrency and potentially parallelism. Currently, the  SNOW package  allows for simple parallel computing. 

 It will be interesting to see how other statistical software producers like SAS, LIMDEP, etc. will react to this trend toward dual processing. Does anybody have more information about this issue?