Serverworks LE versus VIA Chipset

Robert G. Brown rgb at phy.duke.edu
Fri Dec 22 11:37:30 EST 2000


  Warning and disclaimer -- a genuine and inimitable rgb polemic
follows!  Hit "d" now if they bore you to tears!

On Fri, 22 Dec 2000, Dave Leimbach wrote:

> Actually there was an article an /. not too long ago about how PS2's have
> export restrictions due to the massive computing power stored within.
>
> I haven't seen the specs for a PS2 so I won't say its true or not.  I just
> remember the old article saying that it was feared they might be used for
> weapons research which is exactly what /. claims Iraq is doing.

Oh goodness.  One doesn't need PS2's to build a beowulf capable of doing
nuclear weapons research.  A rack of Celerons or Durons would do just
fine.  We're long since into the regime where the only issue is not
whether it can be done but how long one must wait for answers, and
Moore's law continues to push that time down exponentially with a
halving time at constant cost of somewhere between 9 and 12 months
(which has eaten, in turn, each and every line drawn in the sand by the
num-nums who try to control "supercomputer" exports on the grounds that
they can be used to design nuclear weapons over the last 25 years).  I'm
sure that Iraq could afford racks of alpha/myrinet boxen if it came to
that and if they could smuggle them in through the embargo (which I'm
fairly certain they could do if they so desired -- smuggling ANYTHING is
straightforward to somebody willing to spend megabucks on doing so, and
several hundred nodes would fit into a single good sized truck that
could cross anywhere on hundreds of miles of border after the right
palms had been greased).

If Iraq is seriously trying to build a beowulfish nuclear bomb research
cluster (or more believably, a V1-like "pilotless aircraft") out of
PS2's they're crazier (or a whole lot better equipped with high end
computer science/programming talent) than I would ever have believed.

Look at it this way:  This list exists in part to help CPU-starved US
researchers build supercomputer-class resources for cheap.  These
researchers include top quality computer scientists, theoretical
physicists, quantum chemists, aerospace engineers, and even a few
nuclear bomb dudes from Los Alamos I'll wager.  All in all, some fairly
high end talent.  Even so, for many of these folks building a working
beowulf (at least the first time) is a fairly major and significant
enterprise.  Programming it once it is built is another fairly major
enterprise, especially for code that is medium to fine grained or that
wasn't previously parallelized.  PS2's are cheap.  Anybody on this list
building PS2 based beowulfs?  Perhaps one day -- with a MIPS core and
various kinds of operations optimized in hardware it isn't entirely
unattractive but for the moment...

...probably not, and the reason is easy to see.  What is missing?  An
open source and hence extensible operating system?  Optimized, onboard
compilers (I suspect one has to hand code it at this point to get decent
speed, especially on general purpose code)?  Communications channels and
drivers (firewire maybe, PCMCIA maybe, but USB?)  A real networking
stack?  PVM and/or MPI?  Then, the speed issue isn't just raw node "cpu"
speed, it is also communications latency and bandwidth, memory latency
and bandwidth, and so forth, all of which contribute to the >>real<<
sustained speed of the system doing a general purpose calculation.  With
ordinary SDRAM and RDRAM on an ordinary memory bus, they're going to be
memory-bandwidth choked just like the lowliest Celeron or Duron as soon
as they start cranking on >>large<< linear algebra calculations, making
figures like "1000x as fast" meaningless in application.

A nice technical article that seems to be missing all the marketing
hyperbole can be found at

  http://www.arstechnica.com/reviews/1q00/playstation2/ee-1.html

After reading this, 2x or 4x I might believe, even as much as 10x for
certain classes of operation.  They do have 10 FMAC's on board and I'm
sure that they can crank out the linear transforms, and they have a very
clever DMA controller that allows lots of parallel I/O to occur while
cranking, and they have lots of clever little scratchpad caches
interfaced with the DMA controller, but all in all this looks like it
mostly optimizes internally parallelized smallish linear transforms, not
necessarily memory-bus-bound large scale linear transforms or general
purpose numerical code.  >>Even in the realm of its forte<<, cranking
out smoothly rendered images, it isn't clear that a general purpose PC
equipped with a high end graphics card won't be just as fast.  For
general purpose code (especially code that contains annoyances like
division, transcendentals, or basically anything but lots and lots of
DSP multiply/accumulates) they'll almost certainly be relatively slow.

Offhand I have no idea how one would go about sticking e.g. Myrinet (or
any other high speed network) into a PS2, but unless Sony has cracked up
I very seriously doubt that they've engineered what is, after all, a toy
in a way that would make doing so (AND integrating it into the operating
system and network stack!) "easy".  It's not "easy" on a PC or Alpha
with a well defined bus and the kernel sources in hand, and a bad job
yields poor performance.  If Iraq plan on actually using the PS2's in
parallel, they'll once again end up against the same IPC bandwidth and
latency barriers inherent in the beowulf design (except worse, as the
PS2 reportedly has relatively poor communications latencies), and their
presumably superfast CPUs will be sitting there twiddling their
metaphorical thumbs waiting on the network for anything but coarse
grained calculations.

>From my point of view, trying to engineer a PS2 beowulf would almost
certainly be an exercise in extreme pain, culminating in bashing the
damn things with a sledgehammer and throwing them into a ditch somewhere
(or giving them to a children's hospital, which is a much better thing
to do).  However, it is entirely possible that Saddam is desperate
enough (given the embargo that likely makes it moderately difficult for
him to get even cheap Duron boxes) to be willing to throw his available
computer hackers at the task.  Perhaps he just wants them to train his
pilots using flight simulators, since he's a bit short on planes...;-)

The possibility of his building an "unmanned aerial vehicle" on top of
the PS2 is not quite as crazy (and indeed is almost certainly feasible),
but again I fail to see the motivation for using a PS2 instead of, say,
an ordinary laptop computer (of which hundreds would fit into a single
truck, once again).  Hitler managed to build his "buzz bombs" with
1940's technology, so UAV's are clearly not particularly difficult to
make.  I'll bet that I could engineer one from scratch inside a month or
two (right after I learned to fly, of course:-) that would have a 90%
chance or better of working.  These days, with satellites kindly
providing us with electronically readable absolute locations (to within
10 meters or so) I find it excruciatingly difficult to believe that an
ordinary laptop with a handful of servos couldn't be glued into a piper
cub to build a UAV capable of taking off, flying to some predefined
location along some predefined trajectory remaining (say) 15 meters
above the ground (to clear ground trash but remain below most radar) and
then crash itself into any given 100 meter circle within its fuel range.

After all, the timescale of the steering and piloting corrections
required to fly any non-jet plane is long, long, long compared to the
speed of a laptop, and we're talking solving some simple negative
feedback coarse grained differential equations here.  As in "drifting to
the left, move the stick to the right" while integrating the trajectory
with realtime corrections from the GPS. Over seconds.  Pretty much the
same code used to write flight SIMULATORS, which have been around since
the original IBM PC (which was ALSO almost certainly capable of flying a
small plane).  Maybe flying a jet through rugged mountains would be a
bit trickier, but still -- a rack of playstation 2's?  Why not dedicated
moslem kamakazi's?  They are (to be cynical) preprogrammed and probably
more readily available, judging from the number of suicide attacks that
occur in the middle east.  Better at evasive maneuvers and dealing with
accidents of the road, too.

> could be true... screweyer stuff has happened..

This is certainly correct.  And I'm certain that (given enough effort)
PS2's COULD be used to build UAV's or even to do nuclear weapons
research (we'll see if they are featured in the next major bid to Los
Alamos;-).  It would be a most peculiar statement of the distorting
effects of the embargo if indeed it has driven Saddam to spend much
money importing PS2's and expecting his computer folks to deal with all
the manifold headaches of using them as general purpose computers (or as
de facto programmable device controllers).  OTOH, they >>do<< have a lot
of onboard "smarts" for doing e.g. "enemy AI" and simple physics
calculations, as any game system must (after all, what are they running
but moderately sophisticated classical physics simulations?).

I wonder if he's running the market on PDA's? or any of the myriad
programmable device controllers available to the world today?  Or if
he's set up an ASIC shop in (say) Taiwan?  In a world where devices as
evil as singing christmas cards with "infinite" battery capacity exist
(hammer time again:-) custom silicon is obviously WAY too cheap for our
own good...;-)

> > Actually, if you beleive NBC news, we should all be using Play Station
> > 2's.  They actually had someone on saying that the PS2's are "1000 times
> > more powerfull than any desk top computer".  I may have the quote a
> > little wrong, but that was pretty much what he said.

The /. article suggests that it is a 300 MHz, 128 bit CPU, "every bit as
powerful as the processors found in most desktop and laptop computers"
(no pun intended, I'm sure -- a bit >>more<< powerful would have me
doubt my own understanding of binary logic:-).  I'd suggest that this is
a much fairer statement, although I'm also sure that the CPU is (as
advertised) very good at doing graphics transformations.  The system is
still constrained by the current VLSI scale and silicon available to all
systems designers, and still must talk to memory and peripherals over a
bottlenecked bus.  When a number like 1000x more powerful appears, I
think back to the old days, when various "special purpose" CPU's touting
incredibly fast (for their day) floating point speeds were knocking
around the PC market.  In many cases the claim was technically true from
(as Obi Wan would say) a certain point of view -- if one loaded the
CPU's registers in a particular way and was doing just the right mix of
instructions and didn't have to go even to an L2 cache, one could get 10
or 100x the nominal throughput of a general purpose CPU plodding along
on operations out of main memory.

DSP's, for a while, were often touted for this sort of purpose, as was
the (IIRC) i860, and when it comes right down to it, the PS2 is YADSP
being offered up as a general purpose computer.  The incredible pain of
programming a DSP for a general purpose calculation without anything
like an operating system or real programming language was conveniently
ignored.  The i860 faired somewhat better, but never had supercomputer
vendors exactly quaking in their boots.  The PS2 is in better shape than
most of the DSP-based efforts -- it at least has been engineered from
the beginning for >>something<< like general purpose use and does have a
general purpose CPU integrated in the engine, but still...

This long ramble, at the end, has a bit of a beowulf-relevant point.
The beowulf concept (that differentiates it from the manifold virtual
parallel machine efforts that preceded it) has from the beginning been
one of COTS technology and mainstream, proper OS and programming
support.  It is quite difficult enough to engineer a beowulf with a high
quality, open source operating system, over the counter devices
(especially network devices) living on a standardized bus, excellent
compilers, and venerable (and "mostly" debugged:-) communications
libraries.  Building "one-of-a-kind" cluster supercomputers out of
components being used far from their intended purpose and missing most
of the above is the sort of thing undertaken by companies like Thinking
Machines -- it takes years and top-level expertise and a major
investment in engineering and when you're finally done, you find that
Moore's Law ate all of your efforts alive and you're lucky if you can
recover development costs on sales before even the dimmest customer
begins to realize that maintenance alone is costing more then the
hardware that would replace the whole shebang.  Or (in the case of the
CM5) possibly even the costs of electricity and cooling. [Which also
happens even with beowulfs -- ten 200 MHz pentium class nodes cost
approximately $600/year to feed and are likely slower, for most
purposes, than a single Duron node that costs $600.]

I wonder if Iraq executes its computer scientists who fail?  I wonder if
all those PS2's are really Christmas gifts for Saddam's many relatives?
I wonder when we'll see the first UAV terrorist attack?  I wonder when
people will finally wake up and realize that killing each other to
achieve a better way of life is as oxymoronic as it gets?

   rgb

-- 
Robert G. Brown	                       http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567  Fax: 919-660-2525     email:rgb at phy.duke.edu




_______________________________________________
Beowulf mailing list
Beowulf at beowulf.org
http://www.beowulf.org/mailman/listinfo/beowulf



More information about the Beowulf mailing list