[erlang-questions] multicore performance fine grained concurrency

[Johan Montelius]

On Thu, 06 May 2010 23:52:57 +0200, Kenneth Lundin  
<kenneth.lundin@REDACTED> wrote:

> On the other hand there are many examples of real applications that
> really run well
>  in a single smp enabled VM on a multicore HW,

Agree that this is the case and I would accept a small performance hit to  
have smp enabled as default but if the execution time can double or  
quadruple just because there is not enough or to fined concurrency then I  
think one should have smp disabled as default.

Dual core is almost standard today and quadcore is soon in the laptops of  
my students. Many of them don't even think about multicore parallelism but  
still would compare runtimes with other systems or even try to analyze  
their run-times and see how they compare with their calculated figures  
 from a model (which will not work since they don't take scheduling  
thrashing into account).

How many even do experiments with smp disabled, their might be many  
examples of real application out there that could cut the execution time  
in two if they tried turning smp off.

I'm sure this will improve and I'll be willing to help.

  Johan

-- 
Associate Professor Johan Montelius
Royal Institute of Technology - KTH
School of Information and Communication Technology - ICT