Gregg Reynolds wrote:
On 2/18/08, Geoff Steckel <[EMAIL PROTECTED]> wrote:
This is my last posting on this, take heart.
Please enlighten me if there are any -other-
http://acmqueue.com/modules.php?name=Content&pa=list_pages_issues&issue_id=26
See especially "Software and the Concurrency Revolution". An
articulate and very readable discussion of why it's hard to do
threading /with current techniques/. But they stop short of banning
the practice.
Thanks for the reference. It is well worth noting how much space
the article "Software and the Concurrency Revolution" puts towards
discussing locking and other methods of data access sequencing and
control. Controlling data access sharing is hard, and uncontrolled
access sharing is predictably bad.
I will admit skimming the issue in order to reply quickly.
I did not see any discussion (though it should be and probably is
in the text, just not in what I looked at) of the impact of
complex memory architectures on efficiency and concurrency.
That architecture can drastically change the optimum access paradigm
for best speed.
Also, something I also didn't see (though it probably is mentioned)
hardware vendors are seriously considering asymmetrical multicore processors
where there are significant differences in capabilities among the CPUs
on a chip. This also can drive program architecture towards using as little
memory space sharing as possible for highest speed in some cases.
geoff steckel
