[concurrency-interest] concurrency puzzle

Dhanji R. Prasanna dhanji at gmail.com
Wed Sep 13 18:19:03 EDT 2006


On 9/13/06, Joe Bowbeer <joe.bowbeer at gmail.com> wrote:
> In my mental model, I cast the compiler in the role of the Nixon
> administration, maintaining cautious optimism (optimization in this
> case) in its single-minded pursuit, while preserving plausible
> deniability about alleged illegal activities (prohibited actions) that
> may have contributed to the result.  This is much more interesting
> than reasoning about caches, I say, and it encourages the proper level
> of respect for the compiler (well, and loathing, too).

haha, I love it! Does this also mean the compiler will some day be
written in (go to) china? ;)

I think you could extend the analogy somewhat and say that Nixon (Java
VM) took us (developers everywhere) off the gold standard
(platform-dependency) allowing the American economy unprecedented
growth (bigger paychecks for those of us with the letters J2EE on our
resumes) for years to come (hopefully)!

On a more serious note, is the JVM capable of deciding where threads
are run in an SMP environment? For instance, could it say that two
threads passing through the same critical section of code several
times (hotspot) can run on the same processor and thus its volatile
members work faster (i.e. cached in L2 before going to main memory or
something like that)? Or is this a impl/platform dependent issue?

> _______________________________________________
> Concurrency-interest mailing list
> Concurrency-interest at altair.cs.oswego.edu
> http://altair.cs.oswego.edu/mailman/listinfo/concurrency-interest
>


More information about the Concurrency-interest mailing list