🚀 go-pugleaf

Jack Mason wrote:

> The last time I saw the term SMP, about 12 years ago, it was
> used for symmetric multiprocessor.  Is this still what it references?

	Yes.
 
> If so, I am missing the distinction between dual core and SMP since
> it used to be that dual core was just a special case of multiple
> processors accessing the same memory.  

	Dual core means two physical CPUs on the same die/casing. In both dual
core and non-dual core (separate cases for the individual CPUs)
applications the processors use the same memory.

> What I don't understand is how Intel/AMD can build a PC board that
> has more processors, but runs an application 20 to 30 times slower
> than a single processor board.  

	It varies with applications, of course, but it is not my experience
that this happens. Some old software, such as InterBase 6.x, does run
more slowly on a SMP system, but you can work around this by affining
it to a single processor and then there's no big penalty. I have seen
cases of cache poisining by low-priority processes on hyper-threaded
systems. But for the most part I find that SMP gives a big speed boost,
especially with concurrent or multi-threaded processes.

	In particular, people who have more than one heavy user doing a
CPU-intensive task concurrently in IB 7+ will almost certainly see an
advangage from SMP.

> Oh, well...
> Microsoft has never written a decent scheduler for Windows products,
> why should Intel have to learn how to correctly build processors....

	I beg to differ. The scheduler in Windows 2003 was significantly
improved from its predecessors and actually works pretty well.

-- 
Craig Stuntz [TeamB] � Vertex Systems Corp. � Columbus, OH
  Delphi/InterBase Weblog : http://blogs.teamb.com/craigstuntz
IB 6 versions prior to 6.0.1.6 are pre-release and may corrupt 
  your DBs!  Open Edition users, get 6.0.1.6 from http://mers.com