Re: (OT) The desktop supercomputer has arrived!

On Fri, 29 Jun 2007 14:20:37 +1200, in comp.lang.cobol you wrote:

<QUOTE>
Suppose you hire one person to clean your home, and it takes five
hours, or 300 minutes, for the person to perform each task, one after
the other," Vishkin said. "That's analogous to the current serial
processing method. Now imagine that you have 100 cleaning people who
can work on your home at the same time! That's the parallel processing
method.
</QUOTE>

The common response is:

"If a girl takes 9 months to have a baby, why can't 9 girls have it in one
month?"

But the reality is somewhere in-between. In the house cleaning, you
can have one person sweeping each room, but you can't have 10 people
washing the sink without getting in each other's way. We can all
get up from the table together, bring our dishes to the sink - but
then we have to back aside and let one person start washing - and
someone else can start drying. This drying works because we are
doing a well-known task and we know that each dish can be handed off
to the dryer before the next dish is washed. If we are washing
enough dishes for the next place setting, we keep a count - and wash
one more dish if the dryer drops one on the floor.

So how do we turn this into programming?

Read a transaction - hand it to the check the balance routine - hand
that to the calculate interest routine - hand that to the create bill
routine... Periodically check to make sure those routines are
fairly caught up and no (dishes/bills) are dropped - and we need to
know what to do if one is. Restarts aren't trivial.

But we do have a kid of parallel processing. The scheduler verifies
that bills and inventory can be run together and runs these two jobs
at the same time. All that we need to know is that one won't effect
the results of the other.
.