NeXT Pyro Dimension Cube Dual head setup with 64mb/32mb etc......UK

[David Finton]

If you take the today-eBay-price for a dimension cube of about $600 and
you know that the seller want to get about $1700 you will see he rated
the Pyro with $1100.
Sorry, but thats too high since a TurboCube board offers a better
performance (at least due to the better memoryinterface and 128MB).

Hi Andreas,

You could be right about the prices, but I was remembering that
someone told me of a Pyro Dimension that went for quite a bit more
than that recently. I thought he said $1200, but I could be mistaken.
I don't claim to know what it's worth, but I know I wouldn't part with
mine for $600, no way! :-) I get too much out of the machine.
I also put more than that *into* the machine, since I bought one
of the two last Pyros (two years ago?) for $400 (and I seem to
remember that being a competitive price).

I have to take issue with your comment about the TurboCube giving
better performance. I really don't think 64 MB vs 128 is as big an
issue as CPU and disk speed for NS 3.3 on black hardware. The Turbo
does have better disk access; but if your apps are CPU-bound, the
Pyro will not only beat the Turbo, it will also beat the Nitro!

Daniel Miles Kehoe was hired as an independent consultant to test
the Pyro and compare its performance to that of the Turbo and Nitro.
(See http://www.channelu.com/NeXT/Black/Pyro/ ) Kehoe concluded
that while disk-intensive operations were slightly slower on the
Pyro than the Turbo, the Pyro showed improvements averaging 23% in
processor-intensive operations, relative to the Turbo. Although
the Nitro also outperformed the Pyro on disk-intensive operations,
Kehoe determined that on average, the Pyro "closely matched the
legendary Nitro performance." The Pyro beat the Nitro on standard
benchmarks, as well as on a few real-world tests such as Mandelbrot,
WetPaint, Virtuoso, and JPEG conversion tests. So it would seem that
*for the most part*, the Pyro offers better performance than the Turbo.
Of course, it depends on what kinds of apps you're running. For my
neural network simulations, the Pyro was the fastest NeXT processor
made; that was my own personal benchmark.

--David Finton

[Andreas Berger]

Hi David,

Post by David Finton

If you take the today-eBay-price for a dimension cube of about $600
and you know that the seller want to get about $1700 you will see
he rated the Pyro with $1100.
Sorry, but thats too high since a TurboCube board offers a better
performance (at least due to the better memoryinterface and 128MB).

I have to take issue with your comment about the TurboCube giving
better performance. I really don't think 64 MB vs 128 is as big an
issue

Ever installed OpenStep on a Cube ;-)?

Post by David Finton
as CPU and disk speed for NS 3.3 on black hardware. The Turbo
does have better disk access;

and even a better memory-performance, not only 100ns vs 60ns access,
also optimiced timings to the cpu.

Post by David Finton
but if your apps are CPU-bound, the
Pyro will not only beat the Turbo, it will also beat the Nitro!

Shure, for example synthetic benches like bytebenchmark run in the CPU
cache and the highest MHz will win.

Post by David Finton
Daniel Miles Kehoe was hired as an independent consultant to test
the Pyro and compare its performance to that of the Turbo and Nitro.
(See http://www.channelu.com/NeXT/Black/Pyro/ ) Kehoe concluded
that while disk-intensive operations were slightly slower on the
Pyro than the Turbo, the Pyro showed improvements averaging 23% in
processor-intensive operations, relative to the Turbo.

Yes, but does this clearify a gap of ~$1000 to a Turbocube?

Post by David Finton
legendary Nitro performance." The Pyro beat the Nitro on standard
benchmarks, as well as on a few real-world tests such as Mandelbrot,

Hu? Mandelbrot as a Real-World test ;-).

Post by David Finton
WetPaint,

No, not "WetPaint", a CPU-RAW-Power-like test called: "Matrix-Smooth"

Post by David Finton
Virtuoso, and JPEG conversion tests.

Also special schoosen tests.
I remember the same thing on Apples PowerMacs. Everytime there was
testet a Blur-funktion in Photoshop to show, that a PowerMac is a lot
faster than a PC. Unfortunatly other filters show the oposite result.

Post by David Finton
So it would seem that *for the most part*, the Pyro offers better
performance than the Turbo.

No, even with these tests i read (Pyro vs. Turbo):
Average Gain Real World = 17
Average Gain Compiling = -14

...gives a performance advantage of 3% for the Pyro.
(no, i am joking, you can't add these results)
From interest are the compiler-result, wich fill mostly the memory.

Post by David Finton
Of course, it depends on what kinds of apps you're running.

Yes.

Post by David Finton
For my neural network simulations, the Pyro was the fastest NeXT
processor made; that was my own personal benchmark.

So you have to check, if you spend $1000 additional for that performance
gain.

[David Finton]

Post by Andreas Berger
Hi David,

Hi Andreas-- Sorry you feel that Pyros aren't worth as much as
Turbos. Your strongly held point of view is certainly legitimate,
although the issues you raise are highly debateable. A cynical
man might conclude that you're trying to sabatage the market for
Pyro machines. At any rate, your post (true or not) could certainly
have that effect.

I'm puzzled that you seem to be arguing with me that Pyros aren't
worth $1700--when I never made that claim! (Notice that I claimed
that I didn't know the market value of the Pyro).

What I hear you saying falls into two main points:

1) The independent consultant's conclusions are bogus because the
benchmarks are artificial; the test that counts is compilation
speed.

2) RAM speed and amount of RAM are the deciding factors--not CPU
speed.

You're welcome to your opinion, but I think both points are
highly contested.

On the first point, everyone agrees with you that artificial
benchmarks don't show how the system performs in your own
environment, with your own tasks. But everyone also agrees that
we need some objective comparison, which is why people use artificial
benchmarks. Maybe you also believe SPECmarks are artificial, and
therefore meaningless. If so, we'll just have to agree to disagree.
Note that the Pyro has the highest SPECmarks of any NeXT CPU.

I can say this: The Pyro clearly outperforms the Turbo on my tasks.
My custom neural network simulations run faster on the Pyro (with
64 MB) than on the TurboColor (with 128 MB, 70 ns RAM).

In addition, my day-to-day experience is that my Pyro machine
responds pretty much as my Turbo--perhaps better. I guess I could
sit down and see how long it takes to open apps, render Web pages,
transform figures, or latex my documents, and compare scores for
both machines. I have noticed that TeXing is faster on the Pyro.
But these measurements wouldn't matter to you, since you seem to
only care about compilation speed. If that's the case, stay with
your Turbo--but please don't try to tell us that our usage or
preference for the Pyro is bogus just because we use the machine
differently than you do.

On your second point, the consultant claimed that the Turbo and Nitro
had superior disk access, and that this is why they had better scores
on the compilation test. You claim that the real reason is because
they have more and faster memory. I suggest that we put it to the
experts in this group: is the faster bus and disk access more
important, or is more / faster RAM more important for fast compiles?

Happy NeXTing,

--David

[Andreas Berger]

Post by David Finton
Hi Andreas-- Sorry you feel that Pyros aren't worth as much as
Turbos.

No, maybe the reason is my poor english, i have never meant that. I see
a Turbo-board for about $200 and think about, if a Pyro-board *today*
(with the options to use a 10 times faster HP Gecko or a Intelbox) is
a difference of about $1000 worth. Not to think about it, that the
reserve of 1000ukp (1ukp is about $1.67 without currency exchange fee)
was the minimum and not the maximum to reach.

Your mentioned $400 for the sales price gives a completly other result
in this case.

Post by David Finton
Your strongly held point of view is certainly legitimate,
although the issues you raise are highly debateable. A cynical
man might conclude that you're trying to sabatage the market for
Pyro machines.

(Remember, comments about the auction begun *after* ending that and not
during that)

OK, thats not fine. But, its legitim to want $1700, to want $10.000 or
more and to try to find a buyer. But if no one will pay that amount of
money and i will weeping in the newsgroup about that,
/*
Message-ID: <***@eunomia.uk.clara.net>
*/
i have to calculate that someone will give his own impressions about the
wanted price.
I do the same thing, try to sell things in newsgroups for prices they
are worth for me but unfortunatly not worth for the rest oft the world,
but if no-one will buy, i have to alter the price or to put the thing
back into a chamber and waiting for better times.
Insulting the possible buyers isn't the way to become rich.

Post by David Finton
At any rate, your post (true or not) could certainly have that effect.

Sorry about that, but i think, a today-buyer of such piece of hardware
know exactly the goals and the entrapment of such a thing and will pay
what its today worth.

Post by David Finton
I'm puzzled that you seem to be arguing with me that Pyros aren't
worth $1700--when I never made that claim!

Sorry about that, this figure came from the reserve price, that the
seller want. I had assumed, that you also know about these minimum of
1000 ukp to get that Pyro?

Post by David Finton
In addition, my day-to-day experience is that my Pyro machine
responds pretty much as my Turbo--perhaps better. I guess I could
sit down and see how long it takes to open apps, render Web pages,
transform figures, or latex my documents, and compare scores for
both machines.

Sounds good.

Post by David Finton
On your second point, the consultant claimed that the Turbo and Nitro
had superior disk access, and that this is why they had better scores
on the compilation test. You claim that the real reason is because
they have more and faster memory.

David, i had also some expirience with a few kinds of accelerator boards
for the Amiga. With the same clockspeed you get about twice the speed
with the best memoryinterface vs. the worst. (exept bytebench and other
L1-cache filling programs).

Post by David Finton
I suggest that we put it to the
experts in this group: is the faster bus and disk access more
important, or is more / faster RAM more important for fast compiles?

If someone will sell me a Pyro, i can test it for my own *<g>*.

[werther]

Guys,

You are discussing NeXTs in terms of their usefulness and speed, which
seems to me the completely wrong approach! NeXTs are beautiful, both
in terms of their looks, and their OS- and technical design. In fact,
one might be so enthralled by their beauty that he/she wants to use
NeXTs forever. To such a person, processing speed may be important, I
give you that.
However, *very* few people are so dedicated to NeXT systems,
especially when they realize they can run NeXTStep/OpenStep on Sun,
Intel, or HP harware at very low prices. I currently run NS 3.3 on a
Sun SparcStation 20, and it outperforms any NeXT that the world ever
saw (except maybe for the legendary NeXT RISC Workstation [BTW: Has
anybody ever seen one? Does it work? How many exist?]).
So, in order to take processing speed as an important issue in your
pricing considerations, you would have to be addicted to the actual
NeXT hardware, and unwilling to run NeXTStep on other boxes.

Why am I willing to spend a lot of money on NeXTs?
Because they are a fascinating part of computer history! I don't use
my NeXTs (not even the SS20 with NeXTStep) to do any work. I just own
them, and enjoy their existence! Speed is not important at all.
My mode of reasonig goes like this:
1) I discover a NeXT item for sale.
2) I want it.
3) If it's rare, I am willing to pay more for it, because there is a
lower chance I will find another offer at a better price. Also, I am
afraid that over time most NexTs items will have moved out of the
hands of people who just happen to own one, and are willing to sell at
a low price, into the hand of people like me, who love them, and would
only sell at a high price (that is, collectors, who don't see them in
terms of usefulness, but in terms of beauty and historical
significance).

That is, I get 'em while I can.

Regarding the Pyro vs. Turbo debate: Pyros are quite rare. They are
somewhat legendary (not on the same level as the NRW, or as Nitro, but
still). They are some of the ultimate (in its true latin meaning of
"last", as well as "highest") NeXT items there are.
Turbos are also nice, but they are more common. They haven't got quite
the same "specialty" feel to them that Pyros have. I personally
consider a Pyro quite a bit more desirable, and while I will not
disclose the price I'm willing to pay for a Pyro Dimension Cube, I can
tell you I would currently expect to pay far more than for a Turbo.

It is simply stupid to say "NeXT is too old"! Would you say a 1910
Rolls Royce Silver Ghost is "too old" because it has no airbags???

Please consider that NeXTs are collecor's items. They are not "useful"
in terms of modern computing needs.

Many thanks for enduring my rant!

What on earth is going on!
A Pyro.
A Dimension.
A Cube.....?
People were emailing me after the auction offering me £344, saying that was
a very good offer......to me, thats an insult! Looks like its time to put
the thing back into storage........

A wonderfully designed machine and OS... but... it's too old.
To be honest, with it's TCP packet size limit, handcuffing both Nextstep
and Openstep on high ping rate networks (i.e. - cable modem and dsl)
keeping transfers under 25kb a second, it's pretty useless for the
internet.

[David Finton]

Post by werther
It is simply stupid to say "NeXT is too old"! Would you say a 1910
Rolls Royce Silver Ghost is "too old" because it has no airbags???

As an advert that came out in Spring '93 put it:
"Your NeXT computer is now obselete. But so is the Batmobile.
That doesn't stop Batman--and neither should it stop you!"

It's *cool*, man. ;-)

David Finton

[Alex M]

I've still got the NeXT Pyro Dimension cube if anyone is interested.......