On Tue, 7 Dec 1999, Weiqi Gao wrote:
> Scott Murray wrote:
> >
> > On Tue, 7 Dec 1999, Matt Welsh wrote:
> > >
> > > [EMAIL PROTECTED] (Nelson Minar) writes:
> > > >
> > > > However, I disagree that native threads are required for serious
> > > > applications. Green threads work surprisingly well for many
> > > > applications. In some, they work better. I recently wrote a spider
> > > > program that was invoking another program in a subprocess and blocking
> > > > on IO to it. When I had 20 threads (and 20 subprocesses) the green
> > > > threads did much better than native threads. That's not entirely
> > > > surprising, either.
> > >
> > > This is all fine and good until you need to exploit an SMP system!
> >
> > To clarify my original statement, this is mostly what I meant when I
> > said "serious" java applications. From the customer support issues
> > we receive at my day job for the product I work on, it seems that a
> > reasonable percentage of the people running Java servers out in the
> > commercial world are using SMP boxes.
>
> And how are they exploiting the SMP system with green threads?
On Solaris, they use the native thread support in either the Reference or
Production VM. On Linux, they use either the Blackdown or IBM version of
the JDK. That was kind of my point, when I started this thread by saying:
Without native threads support, this release is IMO useless for running
any kind of serious Java applications on Linux.
Of course, it is true that my definition of "serious" does not necessarily
coincide with everyone else's, but I did kind of mean SMP scalability for
running application servers when I wrote that sentence this morning.
Scott
--
=============================================================================
Scott Murray email: [EMAIL PROTECTED]
http://www.interlog.com/~scottm ICQ: 10602428
-----------------------------------------------------------------------------
"Good, bad ... I'm the guy with the gun." - Ash, "Army of Darkness"
----------------------------------------------------------------------
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]