Date: Fri, 5 Jul 1996 18:46:21 -0400
In-Reply-To: <4revlt$8ba@news2.inlink.com>
Message-ID:
References: <833058917.18622.0@melech.demon.co.uk> <4qfgm3$vde@mule2.mindspring.com> <4qfpos$fek@news.netvoyage.net> <4qh9k7$m6c@lace.colorado.edu> <4qkj7j$9ha@digital.netvoyage.net> <4r3s3b$r2r@news2.inlink.com> <4r8p0s$jrv@mojo.europe.dg.com> <4revlt$8ba@news2.inlink.com>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Rex Ballard - Director of Electronic Distribution
Standard & Poor's/McGraw-Hill
Opinions expressed do not necessarily reflect
the Management of the McGraw-Hill Companies.
http://cnj.digex.net/~rballard
On 3 Jul 1996, Sangria wrote:
> In article <4r8p0s$jrv@mojo.europe.dg.com>, Colin Smith, colin@mellifluous.europe.dg.com says...
>
> >: Welll, you're the one claiming NT isn't upwardly scable, I think the
> >: burden of *that* proof is on you...
> >
> >What's the point of putting 250 CPUs in a box if the software cannot use
> >them effectively?
>
> Amazing, when I bring this point up about Fujitsu's 64 CPU Linux
> machine, I'm flamed and accused of "avoiding the issue". Now you
> bring up the same point?
It's an interesting concept. Does NT/SMP still do single threaded disk
queuing? In other words am I going to end up with two CPUs "waiting for
Godout at the speed of light" because the task has to wait for a vertical
refresh or for the desired sector to come under the 7200 RPM drive.
> Let me ask you this question wrt to the Fujitsu box:
>
> Have you ever seen one of these boxes running?
> Have you ever seen Linux running on it?
Actually, I would probably want to run Fujitsu's version of Unix on it.
I'm sure they've tweaked the kernel nice and tight. I was at IBM when
they tweaked their AIX kernel for the ES/9000. It's hard to tell the
difference, as a user, between the ES/9000 running AIX 3.x and Linux.
besides, I can compile most of the applications under AIX as well. If I
buy IBM's version of the standard application suite, I can also get NLS
extensions so that my application can speak Chinese, Japanese, Russian,
Polish, Italian, Latin, American ASCII, or even American EBCDIC (yuk).
Yes, the Linux operating system kernel is a bit primative comparied to
AIX, Solaris, Mach/OSF/1, or HP OSF/1. I might have to do a bit of work
to find the right commands to configure the printer. Some of the system
admin commands are more primitive on "Real Unix" platforms as well.
Standards are really great things. If Microsoft would take following them
more seriously, the Unix/Open Systems community might take NT more
seriously as a standard setter.
> blah, blah...
>
> This is the reason why I didn't want to bother with machines most
> of us would never get a chance to work on let alone see.
Like all those machines that only run under UNIX Look-alikes.
> BTW, how effective is Linux's scaling without kernel level threads
> and that annoying single lock around all the CPUs for all kernel
> space function calls (or did they get rid of that with the newer
> releases)?
Once upon a time, Mr. Gates promised a "Better Unix than Unix".
"Unix" is little more than a structure of agreements that have been
hammered out by over 1 million engineers over the last 20 years. It
includes an entire structure of standards and conventions, supported by
General Public Licence Reference code. Linux is actually a kernel,
adopted as the GPL kernel reference model (the previous BSD model had
proprietary code from AT&T that couldn't be bypassed), which has been
molded to conform to, and support the entire base of GPL software.
Of course, variants of this GPL software are optimised and tuned for
different types of hardware by their vendors. In addition, vendors
provide a nice package of supporting applications that are specific to
their machines, which I can run from my console using X11. To me as a
user, it all looks like one very big machine. When I get started with a
FreeWAIS searcher, It's mind boggling to realize I'm sometimes searching
over 100 hosts -- at the same time, including my own.
Rex Ballard
http://cnj.digex.net/~rballard
From rballard@cnj.digex.net Fri Jul 5 19:20:52 1996
Status: O
X-Status:
Newsgroups: comp.os.linux.advocacy