Michael Ingre ([email protected]) asked about the speed of
-gllamm-- on various hardware platforms, in particular, on Apple's
dual G5 platform. His original email is attached below my name.
We have recently run some tests which indicate that the fastest
platform you can buy right now on which to run Stata is AMD's Opteron.
We have tested a 2 GHz Opteron using -gllamm- and found that the
Opteron outperformed both an Apple G5 and a 2.8 GHz Intel P4.
Apple's dual G5, for an application like Stata which only makes use
of a single processor at a time, will not run gllamm (or any other
Stata code) significantly faster than the same computer with only a
single processor would. For any application which does not take
advantage of multiple processors (most applications do not), the
only advantage a multi-processor machine gives you is that you can
run multiple instances of applications more quickly. It will not
provide any noticeably better timings for a single application.
--Alan
([email protected])
Michael Ingre's original email:
> -gllamm- is a fantastic piece of software however, it does not make my
> computer look too good .... so we are thinking about investing in new
> hardware. I'm wondering if there are any benchmarks available to aid in our
> decision.
>
> To give you some background, a typical dataset might be 9 individuals
> measured once every hour awake for 14 days (2176 observations). A random
> coefficient (2 random effects) model with an ologit link (9 categories) on
> that dataset took nine (9) hours on my PowerBook G4 (800Mhz).
>
> Nine hours is a long time but within reasonable limits (at least for the
> final analyses). However, I would like to analyze larger datasets and
> include one or two extra random effects.
>
> My questions are:
>
> Which platform is the fastest for analyses with -gllamm- or Stata?
>
> Are there any benchmarks available for -gllamm- (Stata) performance across
> platforms?
>
> I'm particularly interested in comparisons with Apples dual G5 computer.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
