SGI Rolls Out "Personal Supercomputers"

Man... by muckracer · 2009-09-23 01:39 · Score: 2, Funny

Can you imagine a Beowulf cluster of those? :-)

Re:Man... by Anonymous Coward · 2009-09-23 01:49 · Score: 2, Funny

Can you imagine a Beowulf cluster of those? :-)
Yes yes, but does it run Crysis?
Re:Man... by cashman73 · 2009-09-23 01:52 · Score: 2, Funny

One of these ought to be just enough to be able to run Windows Vista! ;-)
Re:Man... by EvilBudMan · 2009-09-23 01:52 · Score: 2, Interesting

We'll I think there may very well be a downside to that. As this stuff gets cheaper, the ability for just anybody to figure out problems increases, and that problem could be how to make bad stuff like nukes, or worse a virus writers dream. Hey, five more years and this will be under $2,000 in the sweet spot possibly. Anyhow, I want one, but maybe they need to only let people run them that have passed a basic test on driving a computer.
Re:Man... by sajuuk · 2009-09-23 01:57 · Score: 1

One of these running XP ought to be able to run Crysis at maximum settings.
Re:Man... by Trahloc · 2009-09-23 02:18 · Score: 1

Naw no worries, anyone who buys one gets a free entry into the NSA/FBI/DHS/CIA/INS/ABC/CNN/CBS/PBS/NBC and even Fox databases.

--
The Goal: A long simple life filled with many complex toys.
Re:Man... by chrysalis · 2009-09-23 02:18 · Score: 1

Unless, of course, it runs NetBSD.

--
{{.sig}}
Re:Man... by Anonymous Coward · 2009-09-23 02:21 · Score: 0

Came for the Beowulf reference - left satisfied.
Re:Man... by Verdatum · 2009-09-23 02:25 · Score: 1

Can you imagine a Beowulf cluster of those? :-)
Thank you Internet. You are predictable. And I love it.
Re:Man... by Amiralul · 2009-09-23 02:30 · Score: 2, Funny

Yes, but... does it runs Linux? Because it's official now, Linux is bloated, as Linus stated.
Re:Man... by Anonymous Coward · 2009-09-23 02:34 · Score: 0

But was his statement confirmed by NetCraft? If it was, I must have missed it...
Re:Man... by Anonymous Coward · 2009-09-23 02:39 · Score: 2, Insightful

Come on, you can't be serious.
Your average desktop pc is a Super-computer compared to a desktop of say 10 years ago.
Take 10 more years and every pc will be a HPC by today's standards.
Surely having access to a HPC is not the biggest problem in creating your own nuke, or figuring out any problem.
It's not like these fast computers automagicly program themselves to solve difficult problems.
Re:Man... by EvilBudMan · 2009-09-23 02:47 · Score: 1

Fox? Now that's really bad.
Re:Man... by jbezorg · 2009-09-23 02:48 · Score: 1

Why use this when a bot net is safer? Sure, a hell of a lot slower but with parts of the computations occurring on different machines, it would be harder to trace.

--
I've lost all my marbles except one & It's fun to test angular & centripetal acceleration in my skull
Re:Man... by djnforce9 · 2009-09-23 02:53 · Score: 2, Insightful

I used to wonder the same thing about personal super-computers to be honest, but I think you'd end up frustrated and disappointed when trying to run games on these things.
Notice how it stated "80 core system". Most games are only designed to use up to two cores while maybe some use four (same thing goes for folding @ home). That leaves at least 95% of the super computer's total CPU capacity completely idle (and even if it could technically use all 80 cores, Crysis (or any other modern game) is not THAT demanding or else nobody could run it). Not only that, but you'd still have to factor in the GPU. It was never mentioned in the description of this machine about having some untold super GPU with the power of dozens of high end graphics cards that work together as one unit. If you had that, then Crysis would probably run like never before.
The only game (or rather games) I can think of which might benefit from one of these machines are anything that runs on M.A.M.E. This arcade emulator seems to be almost purely CPU based and multi-threads like crazy due to all the separate components from the original hardware.
Re:Man... by thepotoo · 2009-09-23 02:53 · Score: 1

That downside isn't just restricted to this computer; it's a symptom of technology advancing faster than human nature.
As has been said before, both on this site and elsewhere, for the first few thousand years of human existence, the extermination of humanity was well out of reach of everyone. As technology advanced it became possible for a group of people, working together, to develop a technology for mass destruction (the specific tech often referenced is nukes). Eventually, the group of people became smaller and smaller (theoretically, larger groups of people won't let each other actually use such weapons. Today, a handful of dedicated malicious individuals could develop a lethal retrovirus, a new biotoxin, well, heck, pretty much anything that we don't already have antibodies for would be devastating. But even this requires the cooperation of a few people, hopefully at least one of which would be sane enough to pull the plug before the plan was finished.
The thing that scares me is that within the next decade, projects like personal supercomputers are going to make the development of WMDs a one-man-one-weekend project. I'm not worried about nukes per-se, because a supercomputer won't magically grant you access to 99.99% pure uranium, but there are a lot of companies that would happily sell you the RNA sequence you request from them, no questions asked. Couple that with a supercomputer to calculate the most deadly strains of RNA, the best mechanisms to prevent mutation towards non-lethality, etc., and you're looking at a single person being able to wipe out humanity.
There's no easy way out of this without sacrificing personal freedom and anonymity (and without those there's really no point to having civilization at all). Obviously regulation will only take us so far (scientists debated - actually the US Government tried to step in and prevent Science magazine from publishing - access to early sequences of virus genomes, but information wants to be free, and someone else would figure it out eventually, so there was deemed to be no real point in hiding the information).
I wonder if one of the terms of the Drake equation should be the odds that a single wacko will take it upon himself to wipe out the species, multiplied by the probability of him succeeding, which is of course proportional to the advancement of the technology.

--
Obligatory Soundbite Catchphrase
Re:Man... by acsinc · 2009-09-23 02:54 · Score: 2, Funny

Only on slashdot are computer viruses worse than nukes.
Re:Man... by MrNaz · 2009-09-23 03:00 · Score: 2, Insightful

Because only on Slashdot is it commonly understood that computer viruses can give access to more nukes.

--
I hate printers.
Re:Man... by thepooh81 · 2009-09-23 03:02 · Score: 1

I'm so happy I wasn't the only one who noticed this.
Re:Man... by EvilBudMan · 2009-09-23 03:06 · Score: 1

If someone writes a program, cannot another use it? There is already plenty of software to make both good and bad things. It's all up to the users as usual.
And...I didn't say that was the biggest problems, but to design a working nuke without raising suspicion by testing it physically is a problem computing power can correct.
Like you know some counties may have them and not have tested them like possibly Israel. They probably are just using a design that someone else tested, possibly France that built their nuclear reactor.
Think about China, most electronics comes from there these days. They might sell some of this tech to anyone that doesn't like us.
It's a lot of mights and maybes, but I would want one for CAD/3D work mainly.
Re:Man... by WEqR0lDRR6I · 2009-09-23 03:07 · Score: 1

It's impossible to make an arcade/console game system emulator with any semblance of accuracy that can use SMP, or OS-provided and scheduled threads, for multi-chip emulation, unless the OS and CPU are specifically designed for *extremely* low overhead inter-chip synchronization, and even with that the performance benefit may not be worth it and you might as well just program an FPGA. :b Now, if you don't care so much about accuracy because games on newer systems are generally more tolerant of timing inaccuracies, SMP will help to a degree. And of course you could use queues and buffers to defer CPU-intensive processing whose results are unreadable(in any way, shape, or form) to the emulated program to another core, but that would result in significantly increased complexity and significantly hurt performance on non-SMP systems.
Re:Man... by Anonymous Coward · 2009-09-23 03:09 · Score: 1, Insightful

and that problem could be how to make bad stuff like nukes, or worse a virus writers dream.
First of all, writing a virus doesn't take any computer power to speak of.
Perhaps you meant a medical, biological virus? The term you are looking for is "sequencing" and it takes a hell of a lot more than knowing the info to actually make one. Besides, a few PS3's in a cluster would do just fine. It's the lab to try and make the stuff that's hard to get.
Nukes? You can buy a textbook on physics and know how to make a nuke. Knowing how to make one is no big deal. Actually getting the nuke material is going to be a very large problem right off the bat. Getting access to machining equipment that can actually manufacture the detonator to the very precise tolerances needed to go critical is even more difficult.
You can set the hat down for a while, let that tinfoil rest and cool off a little.
Re:Man... by gnick · 2009-09-23 03:10 · Score: 1

Actually that seems perfectly rational in the context given (i.e. some lone wacko developing one at home.)
Threat of a nuke: (Potential damage) * (Ability of wacko to obtain special nuclear material) * (Ability of wacko to use material in a bad way) * (Likelihood of a wacko going through the trouble to jump the hurdles, create, and deploy the bomb) = Pretty low
Threat of a computer virus: (Potential damage) * (Ability of script-kiddie to assemble a nasty virus) * (Likelihood that some script-kiddie might actually do that) = Pretty high
A rogue nuke detonation is far worse than even the worst computer virus we've seen to date, but computer viruses are a more credible threat. That said, neither threat should be ignored as both present very real and potentially disastrous dangers to life as we know it and there are certainly baddies out there pursuing both.

--
He's getting rather old, but he's a good mouse.
Re:Man... by thePowerOfGrayskull · 2009-09-23 03:12 · Score: 4, Funny

how to make bad stuff like nukes, or worse a virus writers dream
We geeks sure do have our priorities straight.
Re:Man... by EvilBudMan · 2009-09-23 03:15 · Score: 1

I think you are on to something. That DOES sound like one of the uses that I was thinking about but didn't want to state specifically here but since you already have:
Yeah, lab work with it is probably the greatest threat. The software is already out there.
Can't we have freedom and safety? We can do that but we all have to do it and make sure it's done. Even if those odds are very low, my intuition is this: People are Unpredictable. So I predict something will happen, but what it will be exactly, is unknown and probably a combination of things will have to happen, but it is one scary thought you provoked.
Re:Man... by EvilBudMan · 2009-09-23 03:27 · Score: 1

--Actually getting the nuke material is going to be a very large problem right off the bat.--
Nah, just build a breeder reactor and you get Plutonium. That right mix of explosives to make and implosion device and thermonuclear warhead design you wont find in a textbook, at least no specifics.
Enriching Uranium can be made easier with better centrifuge design. The computer controlled machining equipment that could machine to 0.001 inch or 0.001mm is everywhere. A lot of this is being auctioned off to the highest bidder from failed companies. How do you know the machining of the explosives around the outside is critical in that way. I would think dealing with the radio active core would be harder and who is to say it has to be spherical like the fat man bomb. There may be a better shape. Computer modeling can solve a lot of problems.
By I agree, the biological virus concerns are greater, and you have made the way to that even easier with your mention of the PS3, because I don't think the new models will run any other OS. I know the old ones ran Linux.
Re:Man... by lgw · 2009-09-23 03:33 · Score: 2, Interesting

As has been said before, both on this site and elsewhere, for the first few thousand years of human existence, the extermination of humanity was well out of reach of everyone. As technology advanced it became possible for a group of people, working together, to develop a technology for mass destruction (the specific tech often referenced is nukes). Eventually, the group of people became smaller and smaller (theoretically, larger groups of people won't let each other actually use such weapons.
The first European explorers to come to the Americas in the late 15th and early 16th centuries killed 90% of all humans in Central Amercia, and 95% of all humans in North America, without even trying. Modern technologies for mass destruction can't compete with the wooden boat.

--
Socialism: a lie told by totalitarians and believed by fools.
Re:Man... by badran · 2009-09-23 03:34 · Score: 0

The GP meant an actual Virus and not a computer Virus. Computer viruses only need a basic PC.
Re:Man... by Mister+Whirly · 2009-09-23 03:37 · Score: 1

Yes, the same day that BSD was officially declared dead.

--
"But this one goes to 11!"
Re:Man... by Anonymous Coward · 2009-09-23 03:38 · Score: 0

Yes at 640x480 and 4 fps
Re:Man... by MiniMike · 2009-09-23 03:45 · Score: 1

Because it's official now, Linux is bloated, as Linus stated.
When did Microsoft come out with a Linux distro?
Re:Man... by gnick · 2009-09-23 03:58 · Score: 1

Only on slashdot are computer viruses worse than nukes.
I think it must have been that phrase that confused me... Or the parent post that said "virus writer " rather than virus engineer or wicked micro-biologist...
Still, I think the at-home-terrorist-kits typically include ready-made poisons or diseases like anthrax or small-pox. Do we really have wackos trying to create custom diseases to hit us with? Seems like a lot of trouble.

--
He's getting rather old, but he's a good mouse.
Re:Man... by Knara · 2009-09-23 04:02 · Score: 1

Anyone with the smarts (and motivation) can "make bad stuff" that requires additional computing power just by creating a cluster (or arranging to use a cluster) out of COTS parts, frankly.
It isn't a new idea. Making processing clusters out of easily obtainable videogame consoles is a fairly old idea.
Plus, I'd imagine its easier to pay for / steal WMD creation documentation/instructions than it is to make them from scratch at this point.
Re:Man... by MarkvW · 2009-09-23 04:03 · Score: 1

I don't think anybody can fairly doubt that someday a suicidal, sociopathic person is going to create a biological weapon with the potential to wipe out humanity. It is also quite plain that there are people out there quite willing to use such a weapon.
You can regulate the tools to make such an awful weapon, but then there is always some dictator somewhere who needs money to fund his sybaritic lifestyle. He'll sell the evil tools to somebody else. Alternatively, the same tools used to genetically attack the species will also be used to genetically aid the species--more than we can even now imagine. When genetic treatment gets so helpful, the double-edged tools will be widely distributed.
And then you have power-seekers who derive satisfaction from peddling fear and projecting power. That crew can never get anything done (excepting the horror of war).
Seems to me that species survival requires either global communal unity or a diaspora from Earth. Neither of those is likely to happen. When Israel grants equal rights to resident Arabs, and Iranian women obtain rights of self-determination, then I'll reconsider my pessimism.
Re:Man... by Anonymous Coward · 2009-09-23 04:10 · Score: 0

As this stuff gets cheaper, the ability for just anybody to figure out problems increases, and that problem could be how to make bad stuff like nukes
Cheap computational power is superfluous to anyone capable of scrounging or precipitating the materials and equipment needed to create a nuke, let alone create the weapon. It's going on 65 years since the first and last offensive uses of nuclear weapons. The physics involved to create those weapons is well understood today, and it certianly didn't require billions of fancy electronic logic gates to get there back in the 40's.
Heavy computing might be helpful to someone hoping to scrounge every last iota of energy from an exotic fusion detonation process, but that kind of activity is relegated, simply by virtue of the required materials and industrial complex, to superpower nations who are capable of investing billions of dollars into weapons programs, who have no need for inexpensive computing... Furthermore, most of the groundbreaking research in these areas was already conducted in the 50's-70's and was likely done without so much as the computational power of a modern CPU running for a week.
Re:Man... by thepotoo · 2009-09-23 04:28 · Score: 1

A number is hard to get, 90% is really, really high though. I've heard everywhere from 35% to 70% before.
But the big difference is that they did that unintentionally - imagine the impact the disease could have had if the intent had been malevolent; 95% casualty rates followed up by systematic elimination of survivors. Moreover, imagine how quickly society could collapse if any single individual had the capability to create a disease like smallpox and decided to use it.
I'm not saying that smallpox wasn't bad, I'm saying that it was a technology that wasn't understood, and was certainly never used as a weapon.

--
Obligatory Soundbite Catchphrase
Re:Man... by Anonymous Coward · 2009-09-23 04:28 · Score: 0

Linux can only handle up to 4.096 cores, so handling 80 would be out of reach.
Not sure how you get 0.096 cores anyways, but thats Linux for you (was it designed on a Pentium? That could account for the rounding error)
Re:Man... by fprintf · 2009-09-23 04:49 · Score: 1

It seems a long time ago that I built my first SMP computer back in the days of dual celerons (thought I used dual PIIs). The thing that was frustrating back then was that my copy of Windows (98 as I recall) couldn't take advantage of the extra processor, so I was stuck at whatever speed my #1 CPU was going for my games. What a waste of money.
At that time, however, I discovered Linux and had a SuSE system dual booting on it. And yet about the only thing ready for SMP around that time was compiling the kernel. Oh how I used to laugh with glee at compiling a brand new kernel, after just a tweak here or there, and have it ready in under a minute. I'd tweak just one thing and recompile just because I could do so quickly, check it out and then revert the changes. I could sit there imagining the bozos with their single processor computers waiting for multiple minutes to do that same thing. I was so cool!
*insert obligatory Monty Python quote: Oh how I'd lay awake at night and dream of being spit at in the face.*

--
This post brought to you by your friendly neighborhood MBA.
Re:Man... by EvilBudMan · 2009-09-23 04:50 · Score: 1

--Plus, I'd imagine its easier to pay for / steal WMD creation documentation/instructions than it is to make them from scratch at this point.--
Maybe? Like you said it's all old idea, but seems to be getting less and less expensive and there are other unmentioned uses. I just used the hardest one as an example. Bio weapon manufacturing would be much easier and with the possibility of targeting you genetically, it's seems scary out there.
Re:Man... by EvilBudMan · 2009-09-23 04:59 · Score: 1

It's not about making something. It's about everything becoming a commodity to make AND there was electronics in the 40's AND why just superpower nations?
What about Pakistan and India for instance? It's also about not having the need to physically test it. So a small country may stockpile something in this manner without anyone knowing what they have until they decide to do a little ethnic cleansing or something similar. I used the hardest example, for there are other things even worse than that possibility, but anyhow....just read on, there is more.
Re:Man... by kimvette · 2009-09-23 05:02 · Score: 3, Funny

It not only can run Crysis, but it can run Crysis-on-Vista pretty well. with this supercomputer, maybe now 2009 can be the year Windows Vista will be ready for the desktop!

--
The Christian Right is Neither (Christian nor right). See: Matthew 23, Matthew 25, Ezekiel 16:48-50
Re:Man... by I+cant+believe+its+n · 2009-09-23 06:13 · Score: 1

1. Buy powerful computer
2. ???
3. Rule the world

--
She made the willows dance
Re:Man... by lgw · 2009-09-23 06:21 · Score: 1

My point was: wiping out a very large % of everyone on a continent through disease, then systematically wiping out the survivors to make room for your friends: we've done it before. The difference between now and 500 years ago is that now many world leaders would be reluctant to use such tactics: these days, it's really the crazies with nothing to lose that concern us. That's actually a huge step forward from such tactics being used routinely by governments (have you read the Art of War - there's a significant focus how to eliminate most of the opposing army by disease before metting them in battle).
Weapons of mass distruction in the hands of a few crazies who often fail in their plans to go to the grocery store is less frightening than weapons of mass destruction in the hands of a government who considers them just another tactic.

--
Socialism: a lie told by totalitarians and believed by fools.
Re:Man... by Anonymous Coward · 2009-09-23 08:15 · Score: 0

There were electronics in the 40's alright. There were also 'computers'--people who did routine formulaic computations, and that was the state of the art for much of the war. So, your point was???
A small country would simply be better off using nerve gas and bullets for their ethnic cleansing. It's so much more cost effective than wasting tons of money on developing really big, terrifying fireworks... And the commoners will have about as much opportunity to defend against it, as they would a nuke.
why just superpower nations?
I told you, you daft idiot. Superpower nations are the only nations which have the resources to build highly efficient devices--the kind of which you would like the help of a supercomputer to design. India (by all rights a third world country) has tested only one very low yield fusion device. The very highly immobile device was something in the 30kt range. Compare that with the W87 MIRV warheads that are in the US arsenal: They're man sized, weigh about a thousand pounds and deliver a 300kt payload.
At a practical level, Uranium and Plutonium fission devices just aren't all that hard to build. You only need sound fission theory, basic machinery, and in the case of Plutonium implosion bombs (the most sophisticated kind of fusion devices which Pakistan and India are capable of developing), very precise ignition timing, with explosives that detonate in a very precise, highly uniform manner. Simple fission devices are not incredibly difficult to build, but you need a pretty sophisticated industrial complex to do it. Separating out fissionable uranium (about 0.7% of that found in the Earth) requires very fast centrifuges, which are not easy to build, and don't go unnoticed (Iran, I'm looking at you). You've also got to get the uranium into a gaseous state, and be able to precipitate it later, all of this requires a big complex, lots of workers, educated people, and a shit ton of money.
Very efficient, high yield fusion nukes need to be built very precisely. They must have a complex arrangement of photon reflectors, tampers, etc. Otherwise the X-Ray and gamma ray pressure won't be sufficient to ignite the fusion stage. Developing the correct arrangement and shapes of all of this **may** be aided by a computer. However, it just doesn't matter if you can cram all of the TOP500 supercomputers onto a $10 processor at Walmart, because having it doesn't help. Maybe you can simulate a protein which will kill everyone, but it doesn't do shit to help you make a nuke.
Re:Man... by TheTurtlesMoves · 2009-09-23 09:19 · Score: 1

It will only help make a nuke if they provide a nuclear UPS.

--
The Grey Goo disaster happened 3 billion years ago. This rock is covered in self replicating machines!
Re:Man... by TheRaven64 · 2009-09-23 11:26 · Score: 1

You joke, but it's probably correct. If you want to make a nuke, you also need to invest a few million dollars in uranium separation (getting pure enough U235 is difficult). Someone writing a virus doesn't need anything other than the computer. Although I'm not sure hour much help a supercomputer is to a virus writer, given that they need to make their code run on consumer machines.

--
I am TheRaven on Soylent News
Re:Man... by Enter+the+Shoggoth · 2009-09-23 13:34 · Score: 1

It's impossible to make an arcade/console game system emulator with any semblance of accuracy that can use SMP, or OS-provided and scheduled threads, for multi-chip emulation, unless the OS and CPU are specifically designed for *extremely* low overhead inter-chip synchronization, and even with that the performance benefit may not be worth it and you might as well just program an FPGA. :b
Now, if you don't care so much about accuracy because games on newer systems are generally more tolerant of timing inaccuracies, SMP will help to a degree. And of course you could use queues and buffers to defer CPU-intensive processing whose results are unreadable(in any way, shape, or form) to the emulated program to another core, but that would result in significantly increased complexity and significantly hurt performance on non-SMP systems.
Impossible you say.
I would agree that most OS threading models would get in the way but having SMP is an orthogonal issue: all that is required is a good real-time scheduler to take advantage of the extra processing power. That said I suspect most programmers could do with the assistance of a decent run-time as well, something supporting the actor model would make constructing such a beast easier.

--
Andy Warhol got it right / Everybody gets the limelight
Andy Warhol got it wrong / Fifteen minutes is too long.
Re:Man... by GaryOlson · 2009-09-23 15:12 · Score: 1

I suggest you reread history. European settlers purposely gave blankets to natives which had previously been used by infected (recently deceased) people. This killed off the natives rather effectively. I would call this a biological weapon which was understood.

--
Every mans' island needs an ocean; choose your ocean carefully.
Re:Man... by Anenome · 2009-09-23 20:19 · Score: 1

Perhaps he meant, you know, smallpox. That -is- worse.

--
"I Don't Have Enough Faith to be an Atheist"
Re:Man... by selven · 2009-09-23 23:28 · Score: 1

I think you were reading a European site and mistranslated the decimal
Re:Man... by thepotoo · 2009-09-23 23:48 · Score: 1

If you believe that, I suggest you reread history (start with some peer reviewed stuff, not just Wikipedia).
There is only evidence of a single incident of this, and perhaps a couple of letters suggesting that it might be possible to spread smallpox (doesn't mean it actually happened). So, yes, towards the end of the 1700's a couple of people realized that it could be used as a weapon, but it never actually was. While you can't trust everything you read on the internet, Google might be a good place to start.

--
Obligatory Soundbite Catchphrase
Re:Man... by Anonymous Coward · 2009-09-24 01:15 · Score: 0

Yeah, a computer virus may kill people (disrupt a hospital etc). Normal viruses usually don't do that but it's not inconceivable that some wacko won't specifically create and deploy such a virus.
Also, love that sig :)
Re:Man... by EvilBudMan · 2009-09-24 01:33 · Score: 1

Damn that really is possible, and not that hard.
Radio Active Decay = Heat Energy
Heat Energy & Radio Active Decay = Electricity (ca ching)
Well I may have a detail or two off, but you get the idea, right? Where is my damn flying car? Gotta go.
Re:Man... by EvilBudMan · 2009-10-01 08:20 · Score: 1

--Maybe you can simulate a protein which will kill everyone, but it doesn't do shit to help you make a nuke.--
You can simulate whether it will work or not without doing an actual test, that can be detected by other nations like the US. Look, you're right, Iran IS the largest threat. If they have just one working bomb, what's to stop them them from testing in New York harbor about a container ship if they can design a working one without performing a test? I know that they could probably make a gun device and U-238 without computers as well.
I'm only saying that what used to cost a lot of money AND was a big deal to make, is not so much of a big deal any more.
The math for this is easy. It's just grunt work that computers can do very well. It has to do with, you know the amount of calculations per second that a human with a slide rule can't do.
So if Iran doesn't test a bomb and one just happens to go off in one of our port cities, how do we know who to retaliate against?
The other options of us doing a first strike to prevent Iran from having a bomb or letting Israel do it, are bad choices.
So for your answer as to my point, is that too much technology without moral restraint is bad. I also am now coming to believe that if a device can be made that has a really good use, that also a really bad use will eventually be found it as well.
--I told you, you daft idiot. Superpower nations are the only nations which have the resources to build highly efficient devices--the kind of which you would like the help of a supercomputer to design. India (by all rights a third world country) has tested only one very low yield fusion device.--
Tested is the term here. Who knows what they have that they haven't tested? BTW what is a daft idiot? If that is something good and bad at the same time then you get my point.

PS3s by Anonymous Coward · 2009-09-23 01:40 · Score: 1, Insightful

Or you can save a good chunk of change and buy PS3s at $300 for 8 cores. That's 200 cores for under $8K. And open source too.

Re:PS3s by jgardia · 2009-09-23 01:47 · Score: 1

You have to buy the old one, the new one doesn't run Linux (or any other 3rd party OS).
Re:PS3s by Sique · 2009-09-23 01:49 · Score: 4, Funny

You could even go and buy Z80 compatible cores for US$ 0,95 each. That would get you more than 8000 cores for under 8K.

--
.sig: Sique *sigh*
Re:PS3s by Anonymous Coward · 2009-09-23 01:50 · Score: 0

Could, new slimline no longer supports a GNU/Linux install.
Re:PS3s by Entrope · 2009-09-23 01:50 · Score: 2, Informative

Good luck with that -- as much as I like my PS3, the new ("slim") PS3 models come without support for Other OS installation. Sony's official statements on the subject indicate that it isn't coming back, either.
Re:PS3s by GerardAtJob · 2009-09-23 01:53 · Score: 3, Informative

Can't use it for graphics... because all Linux versions running on the PS3 have no access to RSX, the Nvidia-sourced GPU.

--
I can't call that English ;-)
Re:PS3s by GerardAtJob · 2009-09-23 01:56 · Score: 2, Interesting

forgot the source : http://www.theinquirer.net/inquirer/news/1024278/linux-petitioners-sony-ps3-graphical-innards/ sry

--
I can't call that English ;-)
Re:PS3s by kurt555gs · 2009-09-23 02:00 · Score: 1

Morrow Designs had an S-100 running some sort of mini-unix in the early 80's on a Z-80.

--
* Carthago Delenda Est *
Re:PS3s by fuzzyfuzzyfungus · 2009-09-23 02:03 · Score: 5, Informative

If you only need a single dual-socket board, that is obviously a superior choice. Most of the 8k price on the base model is paying for the hardware you have the option to add, not the hardware you are getting.

Assuming you actually need one of the higher end configurations, though, the mac pro isn't going to cut it. A mac pro supports 2 quad core xeons. This SGI box supports 20 quad core xeons in a box of roughly equivalent size. Not to mention that each node on the SGI box supports 3 times as much RAM as the mac pro. Not playing the same game.

That said, the two other configurations they offer (see here) seem much less useful. The "intel 2-way" configuration gives you up to 20 xeons and 960GB of RAM. That is pretty impressive power for a box of the size. The "Intel 1-way" is based on dual-core Atoms. 2GB max of RAM per node and the extremely feeble Atom seems like a very odd choice. 19 Atoms in a box of that size is pretty blah density, and for most applications you'd probably have a faster, cheaper, and easier time with a basic quad-socket board running processors that weren't designed for netbooks. The "Graphics workstation" configuration is a single dual socket workstation board. Lots of PCIe slots; but probably not worth SGI's price for a basic workstation level performance.
Re:PS3s by robthebloke · 2009-09-23 02:07 · Score: 5, Interesting

nah. What put the boot into SGI systems was their premature jump to Intel Itanium processors. We (the CG industry) had been quite happy spending lots of cash for these pretty blue machines with Mips processors, and then one day Sgi declared they were dropping mips for Intel Itanium CPU's. The Itanium then had problems, and so SGi hastily crapped out a new mips CPU on their Fuel workstations. We didn't buy them, because we were waiting for the Itanium ones. So they switched to Intel Xeon CPU's running NT, and we didn't buy them, because as we know, the Itanium hit problems, and a dell workstation running linux was a cheaper option. Over the course of a couple of years Sgi machines literally vanished from the Cg industry.

Then to make matters worse, most of the engineers from the graphics dept of Sgi jumped ship, and all went to join Nvidia (Mark Kilgard et al). The comsumer grade Geforce cards had better OpenGL support + features than an Sgi unit at a fraction of the cost.

This is probably the only realistic comparison you can make between SGI and Apple. Apple (having seen a computer company crash and burn due to a switch to Intel) must have studied what went wrong with Sgi, and made damn sure they didn't repeat the same mistakes.... If Sgi had managed the transition as well as Apple, it would still be a powerhouse in the industry.
Re:PS3s by je+ne+sais+quoi · 2009-09-23 02:13 · Score: 4, Informative

You make some good points, the point I was trying to make is that if you're looking for a just a few nodes, that $8k price tag for the initial node is pretty steep. I noticed in TFA though that:

An Octane III with a 10 dual socket, four cores, Xeon L5520 processors, for 80 cores, 240GB of memory and integrated Gigabit Ethernet networking is priced at about $53,000.
This is actually a decent price for an 80 core system that's preconfigured. You wouldn't want to make a 10 node cluster of mac pros, you could do it easily, in fact my older system is essentially that, a bunch of independent nodes strung together over ethernet and sharing the home directory. You really don't get good scaling over the gigabit ethernet though, as least for what we're doing, so it's pretty pointless to go to more than a few nodes that way. I also noticed this as well:

Silicon Graphics was an independent company until May of last year, when it was acquired for $42.5 million by Rackable Systems Inc. Rackable subsequently changed the name of the combined companies to Silicon Graphics International Corp.
So my suspicion was right, this isn't SGI, it's a server company banking on SGI's name.

--
Gentlemen! You can't fight in here, this is the war room!
Re:PS3s by Anonymous Coward · 2009-09-23 02:24 · Score: 0

Sure, 200 cores for under $8k, with what, 256mb of onboard ram on each machine? Don't forget the '8 cores' aren't necessarily good for every type of solver, and you can't just drop a Linux for PPC binary on there and get direct access to the benefits of those Cell procs, there is an actual SDK from IBM that you use to develop software for those particular CPUs.
It's not a good general purpose super computer, but if you're got the time to port, and the right code, it could be worth it.
Re:PS3s by Anonymous Coward · 2009-09-23 02:34 · Score: 0

That reminds me of my trip down to IBM a while ago.
IBM guy: And we are selling these boards, that can be assembled into a supercomputer for about $1000.
Me: why would I pay $1000, when I can just rip that out of a PS3 that I buy for $600?
IBM: ... we don't talk about that.
Re:PS3s by yogibaer · 2009-09-23 02:36 · Score: 3, Interesting

Or an IBM Bladecenter , where you have a choice between Intel, P6 and Cell (PS3's processor) Processor Blades (http://www-03.ibm.com/systems/bladecenter/hardware/servers/index.html) Not a lot cheaper than the SGI solution but more value for money in my opinion.
Re:PS3s by Gilmoure · 2009-09-23 02:48 · Score: 1

I had Minux running on a Mac Plus a few years ago. No HD, booted from 800k floppy and then had a web server running on it. It wasn't apache, though. Can't really remember what it was. I'm not even sure if I still have this machine. Still, was a fun exercise in minimal computing. Sorta' like getting my Newton 2100 to serve web pages.

--
I drank what? -- Socrates
Re:PS3s by DrWho520 · 2009-09-23 02:52 · Score: 1

While clusters are cool for cranking away at data, any sort of real time application would get killed by the network latency pushing and pulling data between boxes. One of these SGIs can go to 80 processors, 960 GB memory and 726 GFLOPS but how fast can you get the data you need onto the processor it needs to be on? I find more and more latency acrros network, from disk and to the processor dominates performance. How many of those 726 GFLOPS are spent waiting for data to crunch? I would guess less than in a network cluster of machines.

--
The cancel button is your friend. Do not hesitate to use it.
Re:PS3s by RCL · 2009-09-23 02:56 · Score: 3, Interesting

Just curious (I'm honestly not trolling!): you mention that using Linux workstations was cheaper than SGIs. But were they capable of replacing the SGIs? As far as I remember, it was hard to find and configure a decent 2D card for use in Linux back then (1998-1999), and OpenGL support wasn't mature enough for professional use (at least in my opinion). And software support... well, don't know much about SGI's software, but given the current state of Linux software support, I doubt that Linux had anything comparable to (presumably) mature SGI offerings. Could you please elaborate on those topics?

--
Coding etudes
Re:PS3s by OrangeTide · 2009-09-23 03:05 · Score: 2, Interesting

An Nvidia Tesla has 240 cores at 1.3GHz for about $1300. You can put four of these in a system, lets say the base system would cost $1000. You could get eight systems loaded with 4 Tesla cards each for $49600 and possibly have enough left over to get a 10gbE switch. It would total 7680 cores, pretty close to your 8000 core idea. Except this one would actually work, where as there would be a tremendous amount of additional components necessary to get 8000 Z80s to communicate with one another, cheapest being a very long and slow token ring.
(yes, I get your point, number of cores is a poor metric)

--
“Common sense is not so common.” — Voltaire
Re:PS3s by Anonymous Coward · 2009-09-23 03:19 · Score: 1, Informative

Some rendering software lets you set up a cluster of rendering servers. They don't need a video card. They just accept data to render, and render it. The server sends the client back the data it needs. This is done in CPU. Using the GPU to do arbitrary floating point calculations is a relatively new phenomenon in industry.
Re:PS3s by drsmithy · 2009-09-23 03:28 · Score: 1

Or you could just buy this [apple.com]. For $3k plus a couple of thousand depending on how you want to configure it (it's cheaper if you add the memory and hard drive yourself from newegg), you can get a dual quad core with two ethernet ports and pretty decent graphics.
Alternatively, grab yourself a Dell Precision 5500 for the same 8 cores at 2/3 the price. Or, if you don't mind slumming it a little, an XPS 9000 only has 4 cores in a box, but works out at around twice the cores/$ as the Mac Pro.
Re:PS3s by dbIII · 2009-09-23 03:52 · Score: 1

A system with Tyan boards (main plus riser) giving you 8 SOCKETS with the cheapest 6 core opterons you can get is not going to be much more than twice the price of that Dell and give you 48 CPUs (so long as you have the same amount of memory on both).
Re:PS3s by Anonymous Coward · 2009-09-23 04:09 · Score: 0

Some rendering software lets you set up a cluster of rendering servers.
I'd hope ALL software lets you do that by now. You could that with Lightwave on the Amiga, nearly 20 years ago...
Re:PS3s by ColdWetDog · 2009-09-23 04:35 · Score: 1

Morrow Designs had an S-100 running some sort of mini-unix in the early 80's on a Z-80.

Ah yes. The old floppy killer. I've even seen one of those. Hitting that Dual Density 720K floppy like it was your next door neighbor's 16 year old sister. I don't think it actually did anything other that run the OS, but it was interesting....

--
Faster! Faster! Faster would be better!
Re:PS3s by rinoid · 2009-09-23 04:39 · Score: 1

FWIW -- I can't really get a Dell with the same components as the entry level Mac Pro (which uses the Xeon 3500 series in a single chip configuration) to be 2/3 less.
I can configure this machine to be the same basic (with a fairly superior graphic card I might add): Dell Precision T3500 64bit ... but the price is right up there around 3700 bucks which is more than the entry Mac Pro with same processor, etc...
Now we could go around and around on variances or qualities of this Intel processor line or that but just to be fair the components Apple uses, if configured in a Dell seem to make the Dell cost about the same. You just have a choice of going to the i7 line which arguably buys you about identical speed. Right?
Re:PS3s by mmkkbb · 2009-09-23 04:56 · Score: 1

Yeah, I noticed that the 'products' section on their website includes nothing resembling Graphics.

--
-mkb
Re:PS3s by dcw3 · 2009-09-23 05:27 · Score: 1

So my suspicion was right, this isn't SGI, it's a server company banking on SGI's name.
Well, Rackable certainly didn't replace the entire staff. I can personally vouch for that, having just spent the last month getting trained on SGI equipment, by "SGI" personnel...folks who were bought out. So, this isn't just another company using the SGI name.
Having been through a buy out before, it took quite a while to feel the impact from our new parent company. I would suspect that would be the norm, not the exception. Certainly, these systems were under design for quite some time prior to the buyout.

--
Just another day in Paradise
Re:PS3s by LifesABeach · 2009-09-23 05:59 · Score: 1

If I had a huge set of data for modeling analysis, how much disk space can this machine handle? Actually, if the requirements for the data is around 130 Petabytes, how would I hook it up to this 80cpu configuration?
Re:PS3s by robthebloke · 2009-09-23 06:25 · Score: 1

Yeah sure. I can't say this is representative of the whole industry, just my experiences... You've got to remember that we didn't just switch over night ;)

We'd been using various SGI units, O2's, octanes et al for a number of years - replacing them every year. Around 1998, SGI announced the end of their Mips line. This pretty much killed our Sgi upgrades, since we were waiting for the new Itanium machines. And we waited. And waited. And waited. And nothing appeared. By the start of 2000 or so, we had to start making other plans

The NT based Sgi workstations had arrived, but we never bought them, nor we were interested in them. We had used Irix for years, and had a fairly large custom toolchain developed for that platform - porting that to NT wasn't something we were really all that interested in doing tbh.

The new workstations we did get weren't SGIs (who'd buy an underpowered Mips workstation that was slower than a P2 for 3 times the cost?) but bog standard Xeon PC's. For a time we swapped the older Sgi units over to NT boxes, so that the artists were split roughly down the middle - half NT, half Irix. That situation only lasted for about a year i think, until Maya4.0 eventually got released for linux (IIRC start of 2002?) then we all switched over.

To be brutally honest, the switch to linux wasn't quite as smooth as we'd hoped! We had god knows how many problems with things like sound drivers, strange maya bugs and loads of other quirks. About the only thing that was fairly good was nvidia's drivers (at least they were good enough for our needs - we don't push the graphics cards as much as people thing we do! If it can draw triangles, and lots of them, that's enough for 80% of the departments... ). I'd say it took probably another couple of years before the problems had all been resolved.

At about the same time we fully switched over, Sgi released the Mips based fuels, but by then it was a little bit too late!

IIRC, the renderfarm had been gradually switching over for a year or 2 by the time we fully switched the workstations over. By all accounts that transition was seamless.....
Re:PS3s by Anonymous Coward · 2009-09-23 08:12 · Score: 0

Out of curiosity, what computer graphics firm did this take place at? No problem if you don't feel like sharing. Just curious.
Re:PS3s by CarpetShark · 2009-09-23 08:14 · Score: 1

You could even go and buy Z80 compatible cores for US$ 0,95 each. That would get you more than 8000 cores for under 8K.
Pfft. If you shop around, you can get many palettes full of old Apples for that. Each one of them has a core, AND you can keep you horse happy for a month.
Re:PS3s by BikeHelmet · 2009-09-23 08:28 · Score: 1

Lots of cores, but very little memory. For stuff like climate simulations, the 960GB is going to help more.
Re:PS3s by drsmithy · 2009-09-23 08:29 · Score: 1

Firstly, I was wrong initially. I accidentally selected the wrong starting config for the 500 and thus only got the single-CPU version not the dual. However, the price difference is still significant.
FWIW -- I can't really get a Dell with the same components as the entry level Mac Pro (which uses the Xeon 3500 series in a single chip configuration) to be 2/3 less.
I didn't say 2/3 less, I said 2/3 as much (so 1/3 less). I also said a Precision *5500*, not 3500.
As per the post I originally replied to, an 8-core Mac Pro is $3299 base. Adding Applecare to match the standard Dell warranty, and you get $3,548.
A Dell Precision 5500, with the same 2.26Ghz CPUs, 6GB RAM and an 80GB drive (this is a processing node, remember) is $2,731. Bumping the drive up to 500GB gives $2,977.
So not 30% cheaper, but still 15% - not insignificant (and unlike Apple, Dell will actually give you discounts if you're buying a few of them).
I can configure this machine to be the same basic (with a fairly superior graphic card I might add): Dell Precision T3500 64bit ... but the price is right up there around 3700 bucks which is more than the entry Mac Pro with same processor, etc...
Wow. I can't even begin to think what you've added to a *3500* to get the price up to $3,700 (for $3,700, I can spec a 3500 with 12GB of RAM, four drives in a RAID10, two video cards and two 20" screens). However, even assuming you meant a *5500*, for the $3,568 cost of a Mac Pro you get a 5500 with the same CPUs, twice as much RAM, a better video card, a slightly smaller hard disk, and enough change to buy dinner.
Re:PS3s by drsmithy · 2009-09-23 08:38 · Score: 1

A system with Tyan boards (main plus riser) giving you 8 SOCKETS with the cheapest 6 core opterons you can get is not going to be much more than twice the price of that Dell and give you 48 CPUs (so long as you have the same amount of memory on both).
Just the CPUs for that setup cost over $2,500 each (Opteron 8435s). Somehow I don't think you're going to be price-competitive.
There are very, very few situations that need that much processing power in a single box.
Re:PS3s by dbIII · 2009-09-23 10:26 · Score: 1

More like $635 Australian each for a cheaper 6 core opteron which still has decent speed. Your economy hasn't tanked so much that adds up to $2,500 US each.
There are many, many situations where you want that much power in a single box and would even like a cluster of them if you can afford it - just not in your garage or a web server. Ask nearly any engineer, geophysicist (or scientist in nearly any field), industrial designer, movie special effects people etc etc. Since it's getting to around the $5000 mark for such a 48 core system it suddenly gets to the point where you can find a lot of uses for them. Admittedly most developers do not have a clue how to write applications that can actually use more than a single core in an era where kids have handheld dual core video games (Nintendo DS), but the hardware is there.
The SGI thing described in the article is a step above that again for a price that is higher but within reach of situations where it would have been ignored before.
Re:PS3s by Anonymous Coward · 2009-09-23 11:06 · Score: 0

Or you could just buy them pre-configured...
Re:PS3s by Anthony · 2009-09-23 12:48 · Score: 1

The old SGI/Silicon Graphics hadn't been in the Graphics game seriously for years. HPC was their main grame.

--
Slashdot: Where nerds gather to pool their ignorance
Re:PS3s by Anonymous Coward · 2009-09-23 14:21 · Score: 0

OpenGL support wasn't mature enough for professional use (at least in my opinion).
You are aware that OpenGL was WRITTEN for professional use? By SGI, no less? And it was doing professional work for over a decade before Direct3D was invented? And that Direct3D is STILL not used for professional work?
For one thing, Direct3D was not designed with network transparency in mind - you can't get a cluster of Windows boxes to work together on rendering a scene with D3D. I'm going to handwave other notable features because I don't want to be here all night, so I'll direct you to Wikipedia.

And software support... well, don't know much about SGI's software, but given the current state of Linux software support, I doubt that Linux had anything comparable to (presumably) mature SGI offerings.
UNIX is UNIX - the point of having the source code is you can recompile.
Re:PS3s by Glonoinha · 2009-09-23 15:48 · Score: 1

Same way you hook 130PB of data up to any server. Network Attached Storage (NAS) array.
That said, what kind of modeling analysis are you doing that comes pre-loaded with 130PB of data?

--
Glonoinha the MebiByte Slayer
Re:PS3s by Glonoinha · 2009-09-23 15:52 · Score: 1

Persistence of Vision is a ray tracing application that has a cluster version - that was used for a while for demoing and benchmarking the performance and scalability of clusters. It's a damn fine way to demo a Beowulf cluster, and if you actually need a bunch of images rendered it is pretty nice once you have it configured and running.
I did exactly that for a project in school a few years ago. It was pretty sweet demo, and it felt good to finally have built a Beowulf cluster of my own.

--
Glonoinha the MebiByte Slayer
Re:PS3s by RCL · 2009-09-23 19:22 · Score: 1

Thank you for the answer :) It was an interesting reading. As early as 1998 I heard the rumours that Linux was used for CGI in Titanic movie, but I never read a first-person account on actual Linux usage in CG industry.

--
Coding etudes
Re:PS3s by Anonymous Coward · 2009-09-23 23:43 · Score: 0

Or get a used bladecenter with 28 cores.
Re:PS3s by drsmithy · 2009-09-24 00:40 · Score: 1

More like $635 Australian each for a cheaper 6 core opteron which still has decent speed. Your economy hasn't tanked so much that adds up to $2,500 US each.
You are looking at the 2000-series CPUs that only work in dual-processor configurations. You need 8000-series processors for your 8-socket board, and on NewEgg the cheapest ones are priced at 2,139.99 (missed them the first time and only saw the 8435s at $2,649.99 each).
There are many, many situations where you want that much power in a single box and would even like a cluster of them if you can afford it - just not in your garage or a web server.
No, there's aren't. You only need such a large amount of processing power in a single box if your applications is highly reliant on extremely low-latency and high bandwidth between processing units for decent performance. Those are niche situations.
Ask nearly any engineer, geophysicist (or scientist in nearly any field), industrial designer, movie special effects people etc etc.
Most of these applications are handled more than adequately with Ethernet, Infiniband and/or Fibre Channel interconnects. Rnedering special effects, in particular, is one of those "embarassingly parallel" problems where a large number of relatively low-powered machines work exceptionally well. This is especially true when you start looking at the relative prices (see below).
Since it's getting to around the $5000 mark for such a 48 core system it suddenly gets to the point where you can find a lot of uses for them.
It's not. Assuming you're referring to the Tyan S4985-E, your 48-core system is going to have a starting cost somewhere in the ballpark of US$20,000, using the 2.4Ghz 8431s @ ~$2,100ea (assuming about $3,000 for the rest of the machine).
Now, a quad-core 5500-series Xeon has about the same performance as a 6-core Opteron at the same clock speed. So you'd need 8 CPUs in a cluster to match the "48-core" machine you're talking about in terms of processing power. Using Dell's off the shelf hardware, you could do that for:
$14,000 (four dual-CPU Precision 5500s @ ~$,3500ea)
$11,000 (four dual-CPU R410s @ ~$2,700ea)
$9,000 (eight single-CPU XPS 9000s @ ~$1,100ea - though you sacrifice ECC to do it).
Admittedly most developers do not have a clue how to write applications that can actually use more than a single core in an era where kids have handheld dual core video games (Nintendo DS), but the hardware is there.
The biggest problem is that the vast majority of problems simply aren't parellisable, which is then exacerbated by the fact that CPU power is typically not the main bottleneck in a system.
Re:PS3s by Anonymous Coward · 2009-09-24 02:30 · Score: 0

he never said they used a desktop linux distro, and I'm assuming they didn't. You say you don't know about SGI's software, but it looks like you don't know about CG industry software either. I don't know more than a breath, but they *have* been using linux render nodes for years now. IIRC renderman runs on linux (not mentioned in their faq what os platform it uses). Not something you're going to find in your distro, though.
The CG industry couldn't give a fig about linux, or linux rendering software. What they do care about is the price, performance and quality of rendering solutions. If they write it themselves or pay for a solution another CG company has created doesn't matter. If it runs on windows, osx, a custom os, they don't really care. And this is why SGI should have cared very much about ensuring their offering stayed up to snuff.
Re:PS3s by dbIII · 2009-09-24 03:15 · Score: 1

First, look at those opterons again and look at the 4 socket plus 4 socket riser Tyan stuff again, you information is somewhat out of date which is why I'm talking about new possibilities while you are putting forward what has only recently become an incorrect argument.
Then there is this bit:

Most of these applications are handled more than adequately with Ethernet
You end up with annoyingly long periods of time where one node is still going and other nodes are waiting for it. When the price point shifts radically why take "adequate" when "good" isn't much more expensive. Also you have entirely missed the point that there are many jobs that are entirely CPU bound and that is what I was writing about, I'm saying there are many cases where you have jobs that are purely limited by how many instructions you can do and that is what you throw systems or clusters with a lot of CPUs at.
IMHO your example has a premise that experience with both Xeon and opteron systems shows has little merit about the speed difference, your misunderstanding of my statements and your mixing up of terminology of CPU to the stupid salesfolk form instead of anything of technical use is too much for me to handle this late at night. Most of my nodes have Xeons due to performance/price in the past but it is changing back again and for some stuff a single large box is far more useful. Some closed source applications handle clustering incredibly badly with things like long polling times to find an idle node so if you can avoid a few dozen ten minute waits per node per job a large machine at less than Sun and the old SGI prices starts to look very good.
Re:PS3s by mapesdhs · 2009-09-24 05:00 · Score: 1

robthebloke writes:
> nah. What put the boot into SGI systems was their premature jump to
> Intel Itanium processors. ...
Alas, not SGI's fault. Itanium was very late, and the first iteration
was not that good. This is why the original Origin2000 rack could
only hold 16 MIPS CPUs, because it was designed with the thermal
characteristics of Itanium in mind (much hotter). Years later SGI
released Origin3900 with up to 128 1GHz MIPS CPU in one rack, showing
just what was possible given the lower cooling requirements. But by
that time it was far too late of course.
Sadly this was a transition they could not prevent. Many of the SGI
CPU design team left in the winter of 1996/7 (for Intel or
elsewhere), drawn away by huge salaries. Intel played a skillful game
of bragging about how good Itanium was going to be and thus obtaining
but by bit the very design staff they needed (from SGI, DEC/Alpha,
etc.) to design Itanium. I knew someone on the R10K design team back
then; he said even floor sweepers in the CPU labs were paid enormous
amounts. As a design though, the current Itanium2 is quite good; John
Mashey (STREAM author) told me that many aspects of what SGI had been
planning for SN1/2 (Beast/Alien) are incorporated into Itanium2, eg.
the way it performs so well despite its low clock, though the chip
doesn't have the additional vector/media extensions which SGI was
working on (MDMX, etc.) SGI wanted a chip that offered vector
processing in a manner that allowed each register in a vector to also
be used for media processing - this would have given enormous
processing speed for video, etc. But without the staff, moving on with
these ideas was going to be difficult at best.
I was told that if SGI had continued with its own plans, and at the
time assuming Intel was able to release Itanium as planned, SGI's CPU
would still have been faster (by about a third or more) but too
expensive to compete. Thus, at the time, the switch to Itanium. I very
much doubt SGI would have switched if they'd known Itanium was going
to be so late. For a start, Origin would have had a much greater CPU
density, greatly increasing its price/performance ratio.
That wasn't really what ruined their business though. I've dealt with
a wide range of different companies over the years, helped sell $M
worth of SGI kit. IMO the 3rd-party sales model was very damaging,
tech info was not available easily enough about key aspects of
certain products, the marketing/PR was often woeful (which led to
poor targeting of customers for products, eg. selling O2 as an
upgrade over Indigo2 which it was never meant to be - this happened
at ILM), and numerous times, despite feedback/comment, they simply
didn't produce what customers wanted, eg. a speed-gfx solution
without advanced features such as 48bit (aimed at CAD/animation)
instead of the high-quality gfx designs like VPro. Likewise, lots of
daft design choices like Fuel not having audio or Gbit as standard,
Octane2's mbd not having U160 or Gbit, the later PCI cage for Octane
not using 66MHz PCI and a better XIO/PCI conversion ASIC, not using
the Cobalt core for O2+, not changing O2's PCI to 66MHz in the O2+,
not changing the Ethernet to Gbit or the SCSI to U160 in the O2, not
filling the other 4 GE sockets in the IRx GE board to double its GE
speed, etc. I could write a huge list, but you get the idea.
Point is, it was a combination of things. From secretive pricing to
bad marketing. They had excellent products which were poorly sold.
Once the rot set in, key staff left, including most of the two dozen
or so people I got to know over the years. NVIDIA did its best to
poach the gfx division, which worked very well for them, the original
GF256 being based on IR gfx. Irony is, ATI nabbed some of the same
people from NVIDIA later on. But even then SGI had opportunities to
get back in the game, eg. NVIDIA wrote

--
SGI Guru. Email: mapesdhs@yahoo.com | Tel: +44 (0)131 476 0796
Re:PS3s by drsmithy · 2009-09-24 20:02 · Score: 1

First, look at those opterons again and look at the 4 socket plus 4 socket riser Tyan stuff again, you information is somewhat out of date which is why I'm talking about new possibilities while you are putting forward what has only recently become an incorrect argument.
From Tyan's page:

Supported CPU Series: AMD 45nm Quad-Core Opteron 8300 Series Processors [...]
AMD's FAQ also notes that 2000-series CPUs only support two socket boards.
If you have information about 2000-series CPUs working in 8-socket boards, then by all means share it, but it certainly appears that would be news to both Tyan and AMD.
You end up with annoyingly long periods of time where one node is still going and other nodes are waiting for it. When the price point shifts radically why take "adequate" when "good" isn't much more expensive.
Because at this point it appears "good" is still much more expensive.
Also you have entirely missed the point that there are many jobs that are entirely CPU bound and that is what I was writing about, I'm saying there are many cases where you have jobs that are purely limited by how many instructions you can do and that is what you throw systems or clusters with a lot of CPUs at.
I haven't missed the point at all, I'm simply highlighting the fact that such jobs - where a cluster of machines interconnected with Ethernet, Infiniband, or similar won't do it just as well, and a lot cheaper - are few and far between.
IMHO your example has a premise that experience with both Xeon and opteron systems shows has little merit about the speed difference, your misunderstanding of my statements and your mixing up of terminology of CPU to the stupid salesfolk form instead of anything of technical use is too much for me to handle this late at night.
Uh huh. Perhaps you can enlighten me about how I have misused any terminology rather than wave your hands about.

one swallow does not make a summer... by Anonymous Coward · 2009-09-23 01:40 · Score: 1, Insightful

and one Xeon does not make a super-computer.

Re:one swallow does not make a summer... by Anonymous Coward · 2009-09-23 01:46 · Score: 1, Funny

You need to get out more; one swallow sure makes a weekend.
Re:one swallow does not make a summer... by jgardia · 2009-09-23 01:49 · Score: 1

you can put up to 20...
Re:one swallow does not make a summer... by Anonymous Coward · 2009-09-23 01:52 · Score: 0

I'm a loser though. One swallow would make my decade.
Re:one swallow does not make a summer... by oldspewey · 2009-09-23 02:32 · Score: 3, Funny

African or European?

--
If libertarians are so opposed to effective government, why don't they all move to Somalia?
Re:one swallow does not make a summer... by Anonymous Coward · 2009-09-23 03:21 · Score: 0

I'm not picky.
And, since you are posting on /., likely you aren't either. ;-)
Re:one swallow does not make a summer... by Anonymous Coward · 2009-09-23 04:26 · Score: 0

Twenty quad-core Xeon 5500s @ 3.33GHz does. The near-Terabyte of RAM doesn't hurt either.
Turbo Boost Intel procs at 3.3GHz regularly hit 3.8-4GHz each, so 60k MIPS per processor should be a rather conservative guess, putting the system into the >1 TIPS range for a fully loaded unit.
So, yes, I imagine most of the /. crowd would consider it a supercomputer, even by today's standards.

Windows? by Anonymous Coward · 2009-09-23 01:41 · Score: 0

Soooo, can you put Windows on it? *ducks*

Re:Windows? by CarpetShark · 2009-09-23 01:48 · Score: 2, Funny

Soooo, can you put Windows on it? *ducks*
What do you think it's for?
Re:Windows? by afaik_ianal · 2009-09-23 01:50 · Score: 2, Informative

I know you're joking, but from the article:

It can be preconfigured with Windows Server or its HPC Server 2008, as well as Red Hat and SUSE Linux servers.
Re:Windows? by alta · 2009-09-23 04:00 · Score: 1

Actually, windows is not on the supported list of OSes for SOME MODELS... RHEL, Suse, etc are.
The marketing makes it sound like you get a single computer with 80 cores, which is not the case. Basicly you get one machine that's the controller. In the case there are room for a number of sleds, each it's own diskless server. They are intended to boot over bootp or PXE or some other network boot. By themselves, they do nothing. Only the basic $8k box is a personal computer. Once you add the first sled it becomes a 'cluster in a box.'
Because of the clustering it's completely useless as a windows workstation. Vista/Win7 is not going to cluster crap for you. Windows server CAN cluster, but 'common people' (those not on /.) don't go that far for their DESKTOP computers.
So this is only practical if you have the knowledge and software to set this up as a cluster. First thing that comes to mind are the guys at Pixar. Map it out on the workstation, save it to a network share, then the blades pick it up and render. Actually, they'd probably have a distributed rendering system, so it would be even more seamless.
I wouldn't mind using it for web development. Program on the workstation, one sled runs apache, the other runs mysql on a ramdisk :) Another running memcached. Yeah, I know. Overkill.

--
Do not meddle in the affairs of sysadmins, for they are subtle, and quick to anger.
Re:Windows? by rinoid · 2009-09-23 04:41 · Score: 1

Can you make a Hackintosh out of it is the better question!

It's no Cray CX1 by Anonymous Coward · 2009-09-23 01:46 · Score: 0

A key feature instead is the system's ease of use.

But does it provide EASE OF EVERYTHING(TM) like the Cray CX1?

Oblig by Anonymous Coward · 2009-09-23 01:47 · Score: 0

960GB out to be enough for anyone...

BTW, compared to the number in the original (640K, or 655360 bytes), 960GB (close to 1030792151040) is more than SQUARE that number (doesn't quite reach a square if measured in bits instead of bytes -- 640KB = 5242880 bits, the square of which is 27487790694400 bits or 3200GB). Nevertheless, the proportion is truly awe-inspiring.

Re:Oblig by Red+Flayer · 2009-09-23 03:32 · Score: 1

BTW, compared to the number in the original (640K, or 655360 bytes), 960GB (close to 1030792151040) is more than SQUARE that number
Big deal. I have a 1 GHz processor that is almost a decade old, and a 1.2 GHz processor that is six months younger than the other one. The new one has a number more than the CUBE of the old one! And they're only six months apart!!1one!

--
"Trolls they were, but filled with the evil will of their master: a fell race..." -- J.R.R. Tolkien on Olog-hai
Re:Oblig by Anonymous Coward · 2009-09-23 03:47 · Score: 0

The square of a number doesn't change when you change it's units. Check your maths, idiot.

look for the killer ap by MickyTheIdiot · 2009-09-23 01:47 · Score: 0

You need a killer ap in order to start getting the sale of these up to the level of a breakthrough computer product.

I say that killer ap will probably have something to do with sensory suits and porn.

Re:look for the killer ap by Ant+P. · 2009-09-23 07:15 · Score: 1

Killer Access Point? Is that like a wifi Killer NIC?

If you *need* one, why not build one? by TechForensics · 2009-09-23 01:49 · Score: 2, Interesting

Wouldn't most people who would NEED a supercomputer be able to build one much more cheaply using a dozen workstations? It's hard to see how this SGI system might be sold (except perhaps as a replacement for an overburdened business-office server).

--
Those are my principles, and if you don't like them... well, I have others.

Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 01:56 · Score: 0

Or just get put on BOINC.
Re:If you *need* one, why not build one? by NoYob · 2009-09-23 02:01 · Score: 2, Interesting

Wouldn't most people who would NEED a supercomputer be able to build one much more cheaply using a dozen workstations?
Is there any networked or cabled solution that's as fast as a bus on a motherboard? Having those machines communicate with one another and syncing the computations is a lot of overhead that reduces speed and adds complexity.
I see computer animation uses for this. I also see math geeks (hobbyists) buying their own to run their current hobby project. Engineering departments using one to run simulations at a faster rate and cheaper.
It's cheaper than the Apple solution so I see movie editors using this.
You just know the gamers will jump on this!
This thing will sell like hotcakes.

--
It's NOT me! It's the meds! I'm on 1000mg of Fukitol.
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 02:25 · Score: 2, Informative

The answer to your question is Infiniband, which is actually what is used in the Octane III systems.
Re:If you *need* one, why not build one? by vlm · 2009-09-23 02:29 · Score: 2, Informative

Wouldn't most people who would NEED a supercomputer be able to build one much more cheaply using a dozen workstations.
This is a simplification, but is more or less correct:
Xeon FSB width 128 bits by 1.333 GHz equals 170 Gigabits/sec bandwidth between processors.
Commodity ethernet between commodity workstations, 1 Gigabit/sec bandwidth between processors.
If your application runs on 1/170th the interprocessor bandwidth, agreed, it would be cheaper. If not, then it's not a relevant comparison.

--
"Science flies us to the moon. Religion flies us into buildings." - Victor Stenger
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 02:33 · Score: 0

It's not really geared toward gaming.... how many games out there are capable of really taking advantage of an 80 core system? Just because you run something on a system with 80 cores doesn't mean it goes 80x faster. Rendering, math, engineering could certainly benefit from this.
Re:If you *need* one, why not build one? by ShadowRangerRIT · 2009-09-23 02:42 · Score: 3, Insightful

Adding to the PP: The overhead and redundant hardware involved in dozens of networked machines would also mean that, to achieve equivalent performance, you'd likely be using twice the power if not more (you might save a little if you rack them with a single PSU for the whole rack, but it's still going to use a substantially greater amount of power).
My home PC (a state of the art gaming PC as of January 2007), discounting the monitor, uses around 360 kilowatts at peak load (running one CPU and one GPU copy of Folding@Home while copying between the various disks to keep them spun up). Of that, only around 60-70 watts is the CPU, call it an even 80 once you add the memory. The GPU, motherboard, hard disks, and power supply losses eat up a lot of the rest.
If you need 80 cores worth of processing power with frequent interprocess communication, you'll need an 80 core machine, or 100-200 cores split across multiple machines. If we assume eight cores per machine, and 16 machines, if they have even half the power overhead of my machine that's going to run an additional 140 watts per box, or an additional cost of 2240 watts. Over the course of one month, that's roughly 1600 kilowatt/hours of overhead, or about $250-350 dollars of power. Every month. For the entire life of the machine (assume 10 years for a corporate or research box), that's around $36000 (remember, that's on top of the cost of the single box super computer). And that's before you factor in the cost of *cooling* the additional heat produced by the additional machines.
Don't get me wrong, there are advantages to the networked supercomputer design (redundancy and failover, the cheaper components mentioned, etc.). But there is also a place for the all-in-one super computer.

--
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
Re:If you *need* one, why not build one? by symbolset · 2009-09-23 02:43 · Score: 2, Interesting

Sure, you could do it with a cluster of workstations. You would need some insane interconnects. OR, you could just buy this pre-configured system from SuperMicro with dual quad-core Nehalems and 4 Nvidia Tesla C1060 GPU Cards. That's 960 thread processors @1.3 GHz if you don't overclock, 16GB of DDR3 @ 1.6 GHz on a 512 bit bus, 16 threads of system CPU with up to 96GB of system RAM. It pulls close to 4 TFLOPS, in a desktop machine. You probably could break into the top500 with ten of them with decent interconnects since the #500 spot is Rmax 17.09 TFLOPS and Rpeak 37.64 TFLOPS. If you prefer a top 3 OEM, you can get that in a Z800 workstation from HP.
To put that in a time scale for you, that one desktop available today by itself would have easily been one of the top 100 supercomputers in the world only five years ago and would still have been in the top500 3 and a half years ago.
A little spendy for a wordprocessing and light spreadsheets, but a sweet piece of gear nonetheless.

--
Help stamp out iliturcy.
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 02:50 · Score: 0

360 kilowatt? Sure about that one?
Re:If you *need* one, why not build one? by ShadowRangerRIT · 2009-09-23 02:52 · Score: 1

Sigh. Yes, that was a typo. 360 watts.

--
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
Re:If you *need* one, why not build one? by ShadowRangerRIT · 2009-09-23 02:53 · Score: 1

360 watts, not kilowatts. While incorrect, I would hope the three orders of magnitude difference would make the typo obvious and ignorable, particularly since it doesn't affect any of the other calculations, but apparently not.

--
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
Re:If you *need* one, why not build one? by mister_playboy · 2009-09-23 03:08 · Score: 0, Redundant

Your computer puts out 360 kW? It must get terrible gas mileage.

--
Do what thou wilt shall be the whole of the Law ::: Love is the law, love under will
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 03:17 · Score: 0

Just to get over the nitpicking - your main point is sensible, and I've been thinking about similar things myself. (I've been asked to spec up something much smaller at work - in the 16-core range, ideally for not much more than $10'000. It would be great to have a bit more budget and buy something neatly integrated, be it this or a bladecenter. As it stands, I'll probably be clustering two 2xQuadcore 1U servers...)
Re:If you *need* one, why not build one? by hot+soldering+iron · 2009-09-23 04:02 · Score: 0, Redundant

Sorry, just got a nit to pick. 1 Kilowatt = 1000 watts. Most PC power supply manufacturers consider a 1000 watt supply to be "high power", and charge like it. Where the hell did you get a 360 kilowatt supply? Did you have the power company drop you a 3-phase transformer at your house? Does Wolf Creek Nuclear Power Plant give you a price break?!?

--
When you want something built, come see me. If you want correct grammar and spelling, get a F*ing liberal arts student.
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 04:03 · Score: 0

But then you do not have a symmetric shared memory system. Many algorithms are much easier and faster with a symmetric shared memory system.
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 04:21 · Score: 0

uses around 360 kilowatts at peak load
I think you mean watts unless its a ~1000 node cluster
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 04:29 · Score: 0

You mean 360 watts at peak load.
If your computer burns 360 kilowatts, I think you're doing it wrong. Also, does your heatsink act as your and your neighbor's water heater?
Re:If you *need* one, why not build one? by RicktheBrick · 2009-09-23 05:21 · Score: 1

I think that it will be soon cheaper to generate your own results using your own computer than relying on the generosity of people to donate computer time. There is the cost of bandwidth to download data and upload results. There is the fact that all results must be accomplished by at least two computers to ensure accuracy. If they do not agree than more computers must be used until there is agreement on the results. There is the cost of electricity and wear and tear on the donor's computers and whether or not that money could be donated to the cause and generate much bigger results. If the general public is used to create the results than the results should be open to the public. BOINC is still a great way to get results but its time is definitely limited.
Re:If you *need* one, why not build one? by Decollete · 2009-09-23 05:37 · Score: 1

Won't be a gigaFLOP if it turns out later that your 'most people' can find a use for this.
Re:If you *need* one, why not build one? by symbolset · 2009-09-23 05:37 · Score: 1

Update: the 16GB 4-rank DIMMS are validated now, so some of these platforms support up to 192GB of system RAM - but only if price is no object. They're expensive.

--
Help stamp out iliturcy.
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 06:57 · Score: 0

If you read the specs carefully that's exactly what it is
Re:If you *need* one, why not build one? by ShadowRangerRIT · 2009-09-23 06:58 · Score: 1

Slashdot really needs to have some sort of comment editing system. I'd be happy with an "append-only" system (to avoid trolls making posts, drawing replies, then editing them to make the repliers look stupid). That way, when you make one single mistake, people who read the comment will see the correction.
Compare to the current system: a single, obvious mistype draws *four* nitpick remarks over 15 minutes after a correction reply was posted. Seriously, does anyone actually read replies to a post to see if the comment has already been made? I'd be fine with it if the replies were funny, but half were really weak jokes, and the other half were in earnest. Good lord people!

--
$_ = "wftedskaebjgdpjgidbsmnjgcdwatb"; tr/a-z/oh, turtleneck Phrase Jar!/; print
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 06:59 · Score: 0

My home PC (a state of the art gaming PC as of January 2007), discounting the monitor, uses around 360 kilowatts at peak load
How do you connect those 360 kW to the grid? Or do you have your own generator in the back yard?
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 08:25 · Score: 0

"My home PC (a state of the art gaming PC as of January 2007), discounting the monitor, uses around 360 kilowatts at peak load"
360 Kilowatts... really? 360,000 watts?
Do you get a thank you card from the power company each month?
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 08:33 · Score: 0

360 KW? Holy crap, what are you running?
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 09:16 · Score: 0

Mmh - in my defense, I didn't see any others at the time I wrote one of those comments. Still, apologies for derailing early...
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-23 12:19 · Score: 0

No, your gaming PC does not use 360 kilowatts. Misplaced decimal point?
Re:If you *need* one, why not build one? by Glonoinha · 2009-09-23 16:12 · Score: 1

From what I'm reading, the biggest issue I see is people wondering 'What can I do with it?'
THAT is what you can do with the cluster of workstations - proof of concept. Using the hardware you have laying around your test lab at work, you can build a prototype 'supercomputer' over regular GigE or 100BaseTX if that's all you've got. Demonstrate your proof of concept after you've used the same platform to code against, coming up with the same approach you would use on the real hardware since it's all running off the shelf OS's. See that when you run it on one machine it takes 9 hours, and when you add a second machine it only takes 7 hours (due in part to network latency and what have you.) Kick in a third and a fourth machine and it runs in only 5 hours - almost 50% the time when running on four machines vs. one machine.
With this, you will have already coded your prototype and been able to demonstrate the economy of scale and benefit of running the same application on a monster cluster from SGI - it all comes down to dollars vs. hours. If your application is modelling the impact of thousands of real life factors on the stock market and by giving yourself a 4 second advantage you can turn a $1,000 profit per day - the machine pays for itself in two months. If that same 4 second advantage is worth $10,000 a day - the purchase is a no-brainer.

--
Glonoinha the MebiByte Slayer
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-24 13:09 · Score: 0

360kilowatts! Do you have char marks on your desk yet?
Re:If you *need* one, why not build one? by symbolset · 2009-09-24 15:33 · Score: 1

Of course I agree with you. Any organization of a reasonable size has desktop PCs coming off the regular rotation. Right now that means systems like these DC7600s. They have 3.2GHz single core processor and a gig or two of RAM and gigabit networking. 4 year old Dells have the same Spec. I have a cluster of 8 of these I use for quick stuff down in the basement - they're set for wake on LAN and netboot. Add a cheap gigabit switch and some DRBL configs, and you've got a beowulf cluster. They're useful for checking stuff out, testing VDI solutions and whatnot.
If you're rotating them out they're costless and if your problem works with asymmetrical nodes and you have good power, you've not only solved the computation deficit you've deferred the recycling problem as well. If you have to buy them at $1000 for five, that's $200 per core and you're probably better off buying modern quad-core desktops. The Small Form Factor ones work best for ad-hoc clustering.
But this only works if your needs are smaller - like for a pilot - and if the problem you're solving is not very granular (it can be divided only into larger chunks). For fine grained problems best approachable with shared-memory solutions, or scaling to thousands of cores, you get more bang for your buck with the GPGPU solutions because they have 600 cores on one $1,500 card.
That said, I do agree with you - the problems solvable by a rack of surplus desktops are numerous and interesting.

--
Help stamp out iliturcy.
Re:If you *need* one, why not build one? by Anonymous Coward · 2009-09-26 11:46 · Score: 0

You have a computer that uses 360 kilowatts? I'd hate to see your electricity bill.

Mac Pro Cheaper? by metalcoat · 2009-09-23 01:49 · Score: 1

Hell, a Mac Pro is even cheaper with 8 cores. Who in their right mind is going to spend $8000 for a single core pc?

Re:Mac Pro Cheaper? by Anonymous Coward · 2009-09-23 01:53 · Score: 0

because the Mac Pro isn't expandable to 80 cores or 960gb of ram.
Re:Mac Pro Cheaper? by afidel · 2009-09-23 01:57 · Score: 1

It's not a single core, it's quad core (octal core with HT enabled), but it's still insanely expensive. I can get dual 5560's and 72GB of ram for about the same price from HP. The "loaded" system with 80 cores and 240GB of ram and low end 5520 CPU's is at least in the realm of reasonable pricing especially if it includes all the interconnect gear (I'm assuming they aren't using ethernet or else it really makes no sense to buy this).

--
There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Re:Mac Pro Cheaper? by the+linux+geek · 2009-09-23 02:10 · Score: 2, Interesting

This just in: Dual-socket workstations are cheaper than high-end desktop blade enclosures with up to 960GB RAM. Who knew?
Re:Mac Pro Cheaper? by tnk1 · 2009-09-23 02:55 · Score: 1

The presumption of this system is that you will actually put more processors in it than you could simply put in a PC. So comparing this to a desktop or even a small server is pointless. The price point they are giving is something of an illusion because you will never use this system with only one, two or even just four processors. If you did that, you would be wasting your money.
The basic unit is offering expandability to a number of cores and memory modules that your PC or Mac simply cannot provide. The catch is that whether or not you actually purchase more cores than a PC can take, you are paying for the onboard infrastructure for all 80 and its memory up front. That is why the unit cost is so high in the basic configuration.
So, this only becomes interesting for configurations that begin to justify the initial investment in the backplanes. The lower that cost is, the fewer CPUs/memory modules that you have to justify buying for it. The system will likely only hit its best price point at its maximum configuration, but if it provides intermediate configurations where there are no reasonable competitors, it can compete at that level as well. So, a lower basic cost is an important consideration, even if its not really what you will ever see in the wild.
Re:Mac Pro Cheaper? by cbiltcliffe · 2009-09-23 03:11 · Score: 1

Because according to a Mac fanboi, Apple can do no wrong, and anything possible in computerdom, on any hardware, no matter how high end, is completely possible, and in fact faster, on a COTS Apple desktop machine. :-/
Not to say that Apple doesn't make some good stuff, but they're not the be all, end all of computers.

--
"City hall" in German is "Rathaus" Kinda explains a few things......

Cue The Egg Roll: by Anonymous Coward · 2009-09-23 01:51 · Score: 0

I've ordered a dozen for my nuclear weapons design team.

Yours In Peace,
Kim Jong iL

Shifting Standards by seven+of+five · 2009-09-23 01:53 · Score: 2, Insightful

I've seen the term 'personal supercomputer' so many times over the past 20 years. It's just baloney marketing. What you have on your desktop RIGHT NOW is more capable than some of the original CDC machines. So what?

Re:Shifting Standards by OrangeMonkey11 · 2009-09-23 02:03 · Score: 1

Also i think someone who really need one of these for isn't too keen on going into Best Buy and purchasing one considering the idiot that works there.
Re:Shifting Standards by cbiltcliffe · 2009-09-23 03:13 · Score: 1

But if Best Buy did sell these, some sales monkey would undoubtedly be trying to sell one to a little old lady who wants it for web browsing and email, because "Email needs a pretty fast computer."

--
"City hall" in German is "Rathaus" Kinda explains a few things......
Re:Shifting Standards by lymond01 · 2009-09-23 03:49 · Score: 1

It's like this: just because you say your home PC is a "supercomputer" because it has all the performance of a "supercomputer" doesn't make it one. You need to have a little plastic bar glued to the front wherein is written, in dazzlingly Arial font, "SuperComputer". Otherwise, no one will believe you. Oh, just buy it already.
Regards,
SGI Marketing and Management.
Re:Shifting Standards by ascari · 2009-09-23 05:12 · Score: 1

You're spot on. And it's usually made by Silicon graphics, has a name that starts with an "O" and costs 8 grand. And always a huge flop. (No pun intended)
Re:Shifting Standards by OrangeMonkey11 · 2009-09-23 05:41 · Score: 1

LOL you know it
Re:Shifting Standards by Decollete · 2009-09-23 05:42 · Score: 1

And they'd still skimp on RAM like your average consumer non-super PC to keep grandma in pace
Re:Shifting Standards by Anonymous Coward · 2009-09-23 05:42 · Score: 0

Just wanted to say you don't know what you are saying, and that I thank SGI a bazillion if they commercialize this! I will be getting one when I finally have a job and they keep the price below $8,000. This is just... SO FUCKING AWESOME.
Re:Shifting Standards by ceoyoyo · 2009-09-23 15:17 · Score: 1

I think the notebook sitting on my lap would have been on the top 500 list when I started grad school.

$8000 for a single processor by Lord+Byron+II · 2009-09-23 01:53 · Score: 0

Ouch! I never understood the need for all of this specialized "server-class" hardware when cheap-o commodity hardware and a little elbow grease works just as well. Maybe most people don't want to put the work into it, considering the huge jump in price between retail consumer and server pricing, I've never been able to justify shelling out those kind of bucks.

Re:$8000 for a single processor by TerranFury · 2009-09-23 02:06 · Score: 2, Informative

Well... It's 8k, and the CPUS, according to a quick google, are ~2k apiece new (?! For one CPU?). So presumably you can get the full 80-core experience for 168k.
For comparison, a fast commodity rig might cost, I dunno, 1.5k? Times 80, and you're at 120k? So this thing, fully decked out, is possibly 40% more expensive than an equivalent commodity setup? If it's commensurately faster -- which is easy to believe as the processors are on the same mobo instead of strung across a network -- then it could be a net win to use this machine, maybe?
Anyway, it's hard to say. I'm using ballpark numbers and the results are the same order of magnitude, so it might go either way. The point is that the price doesn't seem completely absurd, at first glance at least...
Re:$8000 for a single processor by Anonymous Coward · 2009-09-23 02:09 · Score: 0

Ouch! I never understood the need for all of this specialized "server-class" hardware when cheap-o commodity hardware and a little elbow grease works just as well. Maybe most people don't want to put the work into it, considering the huge jump in price between retail consumer and server pricing, I've never been able to justify shelling out those kind of bucks.
If it's supposed to scale to 80 i7 chips on a single OS image then it's definitely not shit you'd buy at the store.
Second: Some of you think everything's easy because you've read a wikipedia article on it. Or that your average computer janitor has the skillset to setup and maintain a specialized system like this.
Re:$8000 for a single processor by walshy007 · 2009-09-23 02:11 · Score: 5, Interesting

As someone who has a whole Sun Enterprise 5500 rack in his room, There is indeed a great difference between server class hardware and commodity gear, where shall we start.
Multiple power supplies, varied in number depending on your load out but hot swappable and configured as such that 1-2 of them can die before your system goes down. Along with diagnostic interface and usually visible indicators going 'part failure, replace asap'.
Same with cpus, hot swappable cpu/memory boards are a must, so long as a single cpu remains functioning the system should still run albeit at a lower capacity.
While I've already mentioned psu redundancy, the AC power outlets it uses would usually have redundancy also, with two separate connections to different circuits or ups etc.
Anyway, no commodity hardware does this, only high end, high availability stuff has this, and you will pay through the nose for it. If this octane has these features, it is very cheap for what it is.
Re:$8000 for a single processor by Weedhopper · 2009-09-23 02:20 · Score: 2, Insightful

Ouch! I never understood the need for all of this specialized "server-class" hardware when cheap-o commodity hardware and a little elbow grease works just as well. Maybe most people don't want to put the work into it, considering the huge jump in price between retail consumer and server pricing, I've never been able to justify shelling out those kind of bucks.
I can't decide if you're a troll or an amateur. If you're the latter, I meant that in the nicest possible way.
If you don't understand that there's a difference between "cheap-o commodity hardware" and server hardware, you don't understand nearly as well as you think you do.
Re:$8000 for a single processor by Eddy+Luten · 2009-09-23 02:22 · Score: 0

The only benefit that I can think of when going over the specs and datasheet is the fact that it can support a shipload of memory. But I agree that it's definitely not worth the 8 Gs.
Re:$8000 for a single processor by LoRdTAW · 2009-09-23 02:38 · Score: 4, Insightful

You forget service and support! Sun will make sure that for the serviceable life of your machine they will have replacement parts on hand and technical support for your machine. Imagine a commodity system looses a motherboard, will you be able to get the exact one three years down the line? And with pretty much every board maker located in Taiwan will they give you proper tech support in a timely manor? Will they ensure you get matching memory and CPU's? That's the other strong point of server class hard ware that is thoroughly supported by the vendor.
Re:$8000 for a single processor by NNKK · 2009-09-23 02:56 · Score: 1

Hardware fails, and commodity hardware fails more. Using it on a large scale for systems with a high-availability requirement needs engineering/operations skills and resources that many companies simply do not have, and even at the companies that theoretically have the resources, they often fail.
Designing and building applications to withstand server failures can be extremely difficult, especially if you have to scale, and properly testing them can border on the impossible. Sometimes you're better off just getting hardware with a sufficiently low failure rate and dealing with the rare outages as they come up.
Re:$8000 for a single processor by jimbolauski · 2009-09-23 03:11 · Score: 1

When someone says server class think redundancy. If a something breaks on your cheapo server it will be down until it is repaired costing the company money in lost productivity, or even sales. With a "server class" system a singular point of failure will not kill it and the part can usually be hot swapped so NO down time. Trying to build redundancy into a chepo server would be a lot of work easily costing more to build then a "server-class". Lets say you did manage to build a server on the cheap with redundancy there will be no one to fix it or diagnose it if the builder is unavailable so unless you plan on never vacationing I would stick to the out of the box solutions.

--
Knowledge = Power
P= W/t
t=Money
Money = Work/Knowledge so the less you know the more you make
Re:$8000 for a single processor by jittles · 2009-09-23 03:17 · Score: 1

As someone who has a whole Sun Enterprise 5500 rack in his room
Do all the blinking lights impress the ladies? Oh wait, this is slashdot, no ladies have probably seen the inside of your room :P
Re:$8000 for a single processor by cbiltcliffe · 2009-09-23 03:24 · Score: 1

The 5550 is a quad core. So for 80 cores, you only need 20 procs.
That makes the 80 core experience $8000 + ($2000 * 19) = $46,000.
For commodity hardware, you'd have to be getting your computers for $575 for single core machines, or $1150 for dual core. And that doesn't count the discount you'd probably get on the CPUs for buying 19 at once.
Although you might get a similar discount on commodity machines, but it's certainly not guaranteed....
Then there's all the hardware necessary to connect up 40 or 80 machines. Have you priced out a 40+ port Gigabit switch lately? Plus the 40 Cat6 cables, plus all the screwing around trying to get the clustering working, since the target market for this will probably rarely, or never have done it before...
Then there's the extra wiring your house/business is going to need to plug in 40 computers.
Sure, this thing will probably need a 50 amp circuit if it's loaded with CPUs, but a single high current circuit is much easier and safer than a 40 outlet octopus that would be needed for a cluster.

--
"City hall" in German is "Rathaus" Kinda explains a few things......
Re:$8000 for a single processor by walshy007 · 2009-09-23 03:30 · Score: 1

Ironically, those that don't know what it is tend to be like "You have a towering monolith in your room, but where are the monkeys?" (2001: a space odyssey reference)
Re:$8000 for a single processor by KC7JHO · 2009-09-23 03:49 · Score: 1

The price for a fully decked out system is $58,000 (according to an earlier post), however the floor space / cooling system alone would justify the cost increase. This would make an excellent virtual server system provided your servers did not require a massive amount of data storage locally. I wonder if we could get one with a fiber channel connector so it would interface with a large disk array?
Re:$8000 for a single processor by wagnerrp · 2009-09-23 04:00 · Score: 1

I've had it described recently that computers are cheap, people are expensive. It's worth shelling out for high end equipment to prevent your employees from wasting their time due to hardware issues. Any given consumer PC is likely to be stable, but when stringing together several hundred for a simulation, you're going to have problems. When you lose a week's worth of work because one of your (165) nodes was causing silent data corruption because of a cheap, onboard gigabit ethernet controller, those expensive server grade NICs and infiniband/myranet cards start looking better and better.

On the other hand, graduate researchers are dirt cheap too, so universities like to go for the cheaper, less robust hardware.
Re:$8000 for a single processor by TheTurtlesMoves · 2009-09-23 09:32 · Score: 1

My old uni did just that. The down time with 128 PCs was pretty high. Sure you could use it we less "cores" but... then its not a 128 core machine then is it.... The real killer was paying sometime just to get the broken down ones fixed.

We went with blades next. Up front costs was more. After 2 years it was cheaper over all than the cheapo version.

--
The Grey Goo disaster happened 3 billion years ago. This rock is covered in self replicating machines!
Re:$8000 for a single processor by leromarinvit · 2009-09-23 10:56 · Score: 1

Ok, redundancy and high availability is an issue. But isn't it still a lot cheaper to just duplicate the entire box (or hell, 3 commodity boxes are probably still a lot cheaper than one of these...) and use software which can fail over gracefully?

--
Proud member of the Ferengi Socialist Party.
Re:$8000 for a single processor by Anonymous Coward · 2009-09-23 11:31 · Score: 0

cheap-o commodity hardware and a little elbow grease works just as well.
I'm guessing you've never worked in ... technology.
Re:$8000 for a single processor by DMUTPeregrine · 2009-09-23 18:08 · Score: 1

Except that a gigabit network would be too slow for the sorts of workloads you'd normally use a system like this for, so you want 40+ infiniband links. Each Infiniband HCA is about $650, plus the switches: 2x 24-port switches at $3500 each. That adds another $20,000. Of course, that's neglecting all the fiber cable you need for the actual connections. For reference, the quick google search price results: HCA (card) Switch

--
Not a sentence!
Re:$8000 for a single processor by dangitman · 2009-09-23 21:23 · Score: 1

will they give you proper tech support in a timely manor?

I don't know about a timely manor, but all my tech support comes from a stately manor, where we can spend a few hours hunting quail before retiring to sip brandy and admire the house's portraits.

--
... and then they built the supercollider.
Re:$8000 for a single processor by Anonymous Coward · 2009-09-23 22:19 · Score: 0

Actually, you can get like eight to twelve i7-class taiwanese systems for that money.
I wish our company used cheap taiwanese parts - we have old SunOS 5 boxes here, ultrasparc 2, but since replacing it is so costly, we're stuck with it.. until it dies. And there's no spare.
It's got roughly half the performance of a 586 MMX/166. Not bad for 300mhz risc chip.
Re:$8000 for a single processor by Anonymous Coward · 2009-09-23 23:35 · Score: 0

Uhh.. maybe you don't understand the meaning of the word commodity. The point is, if a commodity system loses a motherboard, you don't NEED to get the exact one to replace, any one of the same specs will do. Same for RAM and cpu. And yes, you can buy a part 15-20 years later. Pentium II's for example: http://www.pcprogress.com:80/product.asp?m1=pw&pid=INPII400SECC2
Not that you would want to... By using commodity parts, you can upgrade regularly without requiring a 6-figure maintenance contract (or paying 6-figures to have a tech on staff) in order to do it.
Re:$8000 for a single processor by Anonymous Coward · 2009-09-24 02:41 · Score: 0

Had exactly this problem with a research machine sourced from IBM. It was shipped with NT4 and ran without issue for years until... something went bye-bye. Either the motherboard or the CPU, couldn't tell which. Unfortunately it was a few years old by this time so both motherboard and CPU were problematic. The problem was the system had been installed with proprietary control software that had one of the more paranoid protection schemes I've ever seen. The only way to retrieve the 'serial number' was from a running, functional box. The license floppies were useless as they had had the information removed from them as part of the original installation process. The license key was apparently encrypted somehow/where on the drive and required the software to be running to "unlicense" it and restore the information to floppy. Which meant we had to have the box up and running to rescue the license key. And it was NT4 so just swapping motherboards doesn't work. You also couldn't just slave the drive to another system and run the software to unlicense it.
Some checking showed that IBM sold the same designation motherboard (*exact* same designation) with different chipsets. Which in NT4 will keep it from starting. The software vendor showed no love whatsoever (spend $$ for new license was the only tune they'd sing). IBM wouldn't provide motherboards of that designation (too old, rolled off their inventory) which left ebay and the like. I managed to escape the inevitable disaster, but money was spent on motherboard/cpu and nothing was able to boot the drive (which was still good, was able to image it, etc.).
The point being continued availability can be important outside of just servers. Research equipment is notorious for "rolling off" support with "buy our latest" being the provided solution.
Re:$8000 for a single processor by cbiltcliffe · 2009-09-24 02:54 · Score: 1

I realize Gb network would be too slow.
But would commodity hardware have an expansion slot quick enough to take advantage of Infiniband?
I don't know much about it, but I assume you'd need at least a PCIe 16x slot, and even that would probably be too slow.
Considering a lot of the cheaper commodity hardware (I'm looking at you, Compaq!!) doesn't have anything other than 32-bit PCI expansion slots, or maybe a PCIe 1x on their newer stuff, it's pretty much impossible to use commodity hardware to come in cheap enough for this, even if all the network hardware was free.

--
"City hall" in German is "Rathaus" Kinda explains a few things......

Obligatory by arhhook · 2009-09-23 01:53 · Score: 1

Yeah, but will it run linux?

Re:Obligatory by uncle+slacky · 2009-09-23 02:03 · Score: 2, Informative

The correct question (for a supercomputer) is of course "Does it run Fortran?".

--
Windows is like the faint smell of piss in a subway: it's there, and there's nothing you can do about it.
Re:Obligatory by Bigbutt · 2009-09-23 02:35 · Score: 1

Yep. And Windows. Heck, maybe even OS X :)
[John]

--
Shit better not happen!

Or then you could invest in by Anonymous Coward · 2009-09-23 01:53 · Score: 0

my old Pentium 2. It either won't put you on the TOP500 supercomputer list but it also is simplier.

For $100 it's all yours.

Re:Or then you could invest in by Anonymous Coward · 2009-09-23 02:50 · Score: 0

I'll give you 20 bucks.

Picture by TechForensics · 2009-09-23 01:54 · Score: 4, Informative

Picture here: http://www.ubergizmo.com/tags/octane-3

--
Those are my principles, and if you don't like them... well, I have others.

Re:Picture by psergiu · 2009-09-23 02:05 · Score: 2, Insightful

Ugly :( Gone are the beautiful SGIs we knew. :(

--
1% APY, No fees, Online Bank https://captl1.co/2uIErYq Don't let your $$$ sit in a no-interest acct.
Re:Picture by Anonymous Coward · 2009-09-23 03:14 · Score: 0

Beautiful SGIs left when they changed their logo to the current ugly form.
Re:Picture by theManInTheYellowHat · 2009-09-23 03:24 · Score: 1

That is the ugliest SGI ever made!! I spent the late 90's on Indys, O2s, Octanes (the real ones), Onxys, and Origins and they all were beautiful. Maybe they should outsource design to Apple and get some mojo back.
Re:Picture by frank_adrian314159 · 2009-09-23 03:41 · Score: 1

Wow! It's a box! With an ugly logo. I'm impressed...

--
That is all.

Holy Bad Marketing Batman by damn_registrars · 2009-09-23 01:55 · Score: 4, Informative

Who was the idiot who thought that it would be a good idea to call this the "Octane III"? This has almost no resemblance to the SGI Octane systems of that past, which were graphics workstations running Irix with MIPS processors. I think the only thing that makes them similar is the price range.

This goes right up there with Honda constantly recycling their product names; passport, odyssey, pilot, and more recently insight.

--
Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.

Re:Holy Bad Marketing Batman by psergiu · 2009-09-23 01:57 · Score: 3, Insightful

It's UGLY !!! And GRAY !!!
That's no SGI.

--
1% APY, No fees, Online Bank https://captl1.co/2uIErYq Don't let your $$$ sit in a no-interest acct.
Re:Holy Bad Marketing Batman by Scott+Kevill · 2009-09-23 03:29 · Score: 1

That's no SGI.
It's a space station.

--
GameRanger - multiplayer gaming service for PC and Mac games
Re:Holy Bad Marketing Batman by stg · 2009-09-23 06:09 · Score: 1

Yeah, I was pretty disappointed when I saw that! It actually looks worse than the Octane I. Even my US$70 generic case looks a lot better...

A similar personal computer, the Cray CX-1, looks awesome.
Re:Holy Bad Marketing Batman by barzok · 2009-09-23 06:30 · Score: 1

At least both Honda Insights have been hybrid cars.
Re:Holy Bad Marketing Batman by damn_registrars · 2009-09-23 06:45 · Score: 1

At least both Honda Insights have been hybrid cars.
True, it is not as monumetally idiotic as the reuse of the other names that I mentioned. However, the new "Insight" is so much larger than the former that I suspect you could almost park the original inside the new one.

--
Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.
Re:Holy Bad Marketing Batman by Anonymous Coward · 2009-09-23 09:47 · Score: 0

Correct, it isn't SGI. Just another company by the same name. It is Rackable. SGI assets were purchased (after SGI's 2nd bankruptcy) by Rackable. Rackable then changed its name to SGI, with a mostly Rackable product mix.
Re:Holy Bad Marketing Batman by turtleAJ · 2009-09-23 23:03 · Score: 0

It's like if Honda is in a odyssey of recycling everything. It would gleam an insight if they did a pilot for these names. Don't know what to do with the passport. =)

Remember Apple's Supercomputer Ad? by Anonymous Coward · 2009-09-23 01:59 · Score: 0

Remember Apple? In the final days of the PPC, when the thing was as slow as a turd, competitively, they ran ads proclaiming themselves a "Supercomputer".

Now, they've switched to Intel *and* reverted back to 32-bits (from the 64-bits they used to brag about, too, during the last PPC days)....and they're faster than they ever were!

Re:Remember Apple's Supercomputer Ad? by mikael_j · 2009-09-23 02:26 · Score: 1

I'm assuming you're a troll but I'll bite anyway, modern Macs all run on 64-bit Intel CPUs, early Intel Macs did have 32-bit CPUs though.
/Mikael

--
Greylisting is to SMTP as NAT is to IPv4
Re:Remember Apple's Supercomputer Ad? by Anonymous Coward · 2009-09-23 02:47 · Score: 0

The first PowerBook G4 was considered a supercomputer legally in the US because of their retarded classification at the time (the first PB G4 hit over a GFLOPS, the State Department comes a-knockin "sorry guys, you can't sell that thing there, there, there and there", Apple uses that for marketing)
Also, considering Power/PowerPC have been 64-bit since v5 - IBM is at Power 7, the problem was not just one of lack of juice, but the inability to make laptop-ready versions of the G5; the G5 macs were pretty sweet gear - to make it short, the laptops were never 64 bit, until Core2 got out.

Yes they are by stokessd · 2009-09-23 02:00 · Score: 1

"They aren't selling personal supercomputers at Best Buy just yet."

Sure they are, it just depends on what era supercomputer you are comparing that commodity computer to. A modern desktop machine is insanely fast with inconceivable amounts of ram and disk storage if you think back a couple (several) decades. Best-buy will never sell super-anything, it's not their game. But the computers we take for granted are insanely capable machines based on the problems tacked in the past by supercomputers.

Now get off my lawn,
Sheldon
(who did his master's thesis on a 16MHz machine)

Re:Yes they are by Anonymous Coward · 2009-09-23 02:12 · Score: 0

Indeed. And in another few decades, the Octane III will be an ancient mainframe that nobody wants to work with, and less powerful than a cell phone.*
*If Moore's Law keeps rolling for a while.
Re:Yes they are by Anonymous Coward · 2009-09-23 02:36 · Score: 0

More to the point, Best Buy actually does sell modern-day supercomputers, at least by SOME definitions. The Commerce Department's Bureau of Industry and Security is responsible for setting export controls on items like weapons systems, supercomputers, and other stuff that they think will be used by nefarious foreigners for Bad Things (tm).
If you look at the current definition of "supercomputer" that the BIS uses, you'll find that it's pretty easily reached: in short, they think that a supercomputer is any computer capable of achieving 0.75 teraflops. That sounds like a lot, until you realize that a modern graphics card is capable of performing well over that. Unfortunately, the definition used by the BIS doesn't really account for the fact that graphics card floating-point calculations are single-precision and aren't completely standards-compliant (limited NaN and exception support, for instance)-- it just considers how many calculations are done per second.
So, by US Government standards, Best Buy has actually been selling supercomputers for a while. Crazy, eh?

Guess lightning can strike twice by sunking2 · 2009-09-23 02:01 · Score: 3, Insightful

Isn't this basically the failed business model that put them under the first time?

Re:Guess lightning can strike twice by TheGratefulNet · 2009-09-23 02:18 · Score: 3, Interesting

that was known, internally as the WBT (wintel box thing).
I kid you not.

--

--
"It is now safe to switch off your computer."
Re:Guess lightning can strike twice by dangitman · 2009-09-23 21:13 · Score: 1

As an SGI customer at the time, I was invited to a lavish product demo of their new Wintel machines. Boy, was that a "jump the shark" moment. The audience was just shocked and bemused. It was interesting, because it kind of set the scene for the following years, where many other businesses seemed to follow the same model of MBA-driven marketing of vanilla products as the new hot sauce. Ultimately culminating in the dot-com bubble, I believe.

--
... and then they built the supercollider.

Super Computer by Demonantis · 2009-09-23 02:08 · Score: 1

This is a super computer not some thing you would use for normal every day computer activities. I think it would cater more to people that are doing protein bending or other extremely processor intense activities. Most of people wanting to use one probably have access to a University or Government Super Computer so that only leaves the self-employed or small business research market. I would not be surprised if even more of these appear from other companies because that market is one of the few that is growing.

Why? by L4t3r4lu5 · 2009-09-23 02:11 · Score: 1

What will a home user do with an 80 core, 1TB RAM sysetm? Ray tracing? Protein folding? Local weather prediction? All things really high on the list for personal computers.

Still, you'd never need to heat your house again.

--
Finally had enough. Come see us over at https://soylentnews.org/

Re:Why? by the+linux+geek · 2009-09-23 02:30 · Score: 1

Its not a PC, its a workstation. How many home users are going to shell out $8k for a base configuration?
Re:Why? by walshy007 · 2009-09-23 02:33 · Score: 1

What will a home user do with an 80 core, 1TB RAM sysetm? Ray tracing? Protein folding? Local weather prediction? All things really high on the list for personal computers.
I'm not sure about you, but my most immediate thought would be to simulate an extremely complex neural network, likely easily using over half the ram on that task. Combine that with coding an infrastructure for it to learn patterns well from a source of input like a webcam, and hilarity ensues.
If it could get to be of about the intellect of a small bird that would be awesome, but even if it is a spectacular failure and nothing valuable is produced, it would be awesome :)
Re:Why? by malevolentjelly · 2009-09-23 02:41 · Score: 4, Funny

What will a home user do with an 80 core, 1TB RAM sysetm? Ray tracing?
Sometimes I need a giant mirrored ball as a pick me up when I'm down, or a photo-realistic digital recreation of a bowl of fruit. What's wrong with that?

Protein folding?
They're not going to fold themselves.

Local weather prediction?
I don't trust the NWS, though. I generally try to run my own weather models at home every morning before leaving for work. I have to do something with these petabytes of NASA satellite data.
Re:Why? by agnosticnixie · 2009-09-23 03:25 · Score: 1

Neural networks are a misnomer.

they can keep up by Tim4444 · 2009-09-23 02:13 · Score: 2, Funny

Aw crepe, if these become commonplace M$ might rewrite Windows using dot net, and of course Sun would write a knockoff in Java. By then Linux will have 8 different windowing toolkits necessary for the basic apps and 29 sound systems. Oh well, I guess it's back to 0x7C00...

Re:they can keep up by Silverstrike · 2009-09-23 03:48 · Score: 1

I wonder if you're being modded funny for implying that Java is a knockoff of .NET, because that is pretty funny. Or sad...
Re:they can keep up by DahGhostfacedFiddlah · 2009-09-23 05:56 · Score: 1

He might be referring to the fact that C# has been winning the language-feature war recently, and Java has been adding not-quite-as-good knockoff features (see generics).

--
Last post!
Re:they can keep up by FlyingGuy · 2009-09-23 08:58 · Score: 1

There is much speculation that it was a project originally dreamed up at Borland. If you look closely at O Pascal ( aka Delphi, aka Object Pascal ) they are remarkably similar.
Given that they were both architected and written by the same guy, I would say that .Net is a knock off of Object Pascal.

--
Hey KID! Yeah you, get the fuck off my lawn!
Re:they can keep up by Tim4444 · 2009-09-24 02:32 · Score: 1

open office...

the .dot reference is also about reimplementing native code in a slower language/environment, not about comparing .net to java per se... especially since java predates .net by a good number of years...

i was hoping ppl would take this as a comment on general trends of late and not a technology specific attack... a glass house kinda thing...

Better options by Anonymous Coward · 2009-09-23 02:15 · Score: 0

Are a Tesla with a zillion GPU cores, and some others if you don't mind some programming. In MFLOPS/$ they kick this thing's butt. I'm not familiar with other options than CUDA, there's only so many hours in my day, but gosh, that looks like a much better deal, and of course can be put on a killer Intel Mobo anyway -- and you have both.

Here, I run on solar power though, and the cost of running even one big NVIDIA card in a hot mobo is such that it doesn't run much, so my criteria would be flops/watt hour. The thing is noisy and you could probably cook eggs on the case or in the exhaust air stream. On the other hand, if you're say training neural nets or other real compute bound job, it beats having to keep a cluster up and running by a mile.

Despite the prevailing skepticism here, this is... by mmell · 2009-09-23 02:15 · Score: 2, Interesting

a supercomputer chassis. Not unlike getting an IBM BlueGene with ONE cell processor on ONE card in ONE unit on ONE rack. I suppose it's still a 'supercomputer' (since nobody's really defined what a supercomputer is). The architecture is there for true, multiprocessor multithreading in a highly scalable framework. Way cool!

Then again, I'm buying up Marvel SheevePlugs as fast as I can afford 'em. With built-in 1000TX networking and a Kingston SOC chip delivering approximately the same performance as a 1GHz Intel CPU, I figure I can network 'em together and have a scalable (Beowulf) supercomputer for a lot less money and only a modest investment in elbow-grease. The uBoot environment is already smart enough for TFTP boot and root over NFS (which is how IBM does the magic, IIRC). All I need is a monotasking kernel to serve to my nodes and I'm in business. For now, I'll settle for the standard Linux 2.6 kernel and take the modest performance hit.

nice pricetag! by nimbius · 2009-09-23 02:27 · Score: 0

did this company forget the economic recession entirely? no one can afford to buy this.

--
Good people go to bed earlier.

Erwin? by gzipped_tar · 2009-09-23 02:31 · Score: 1

So, SGI is finally putting Erwin into *real* SGI machines?

--
Colorless green Cthulhu waits dreaming furiously.

Less than 500. Compare with Xserve. by mattr · 2009-09-23 02:34 · Score: 1

Bottom of the Top500 as of June is about Max 17 Tflops, Peak 37 Tflops. So if this is really 0.6 Tflops well... maybe one of these would make one node of a many-noded supercomputer.

Interesting to compare to a rack of Apple Xserves. Each rack is 8 cores (same cpus as the Octane III it seems). Again about $50k for 80 cores. Looks like sgi is aiming at that segment.

Can anybody with Xserve experience say how these would compare? I see Apple has something called Xsan too.

Re:Less than 500. Compare with Xserve. by KC7JHO · 2009-09-23 04:10 · Score: 1

This would take up a lot less space than a full rack would, I wonder about the cooling capacity it would require as apposed to a full rack?

Please tell me it's not used for entertainment. by professorguy · 2009-09-23 02:38 · Score: 1

whole Sun Enterprise 5500 rack in his room

Does it run 24 hours/day (as the high availability options would suggest)? How much power does it consume? Why are you burning all that power all day?

I'm hoping for an answer like: "I'm modeling a quantum electrodynamic system at 100 hours of processing per microsecond." Or maybe "I'm trying to find a zero of the Zeta function off the critical line." Or even "I'm trying to factor a big number." Please don't tell me "It sits idle 23.9 hours a day and then I play Spacewar on it once in a while."

Although we all know the likelihood of each of these answers.

Re:Please tell me it's not used for entertainment. by falcon5768 · 2009-09-23 02:48 · Score: 1

oh how about he owns his own hosting service? I know a number of people who do this and make a nice amount of cash on the side doing it, including a co-worker of mine who also designs the sites for businesses too.

--
"Slashdot, where telling the truth is overrated but lying is insightful."
Re:Please tell me it's not used for entertainment. by walshy007 · 2009-09-23 03:23 · Score: 1

How much power does it consume?
A fully decked out system can use 5760 watts (24 amps at 240v) or so it says on the back power panel which has a couple 25 amp iec connectors on the back. the 5500 system by itself wouldn't use anything near that though, about 1600 watts. There is plenty of remaining space in the rack for other things however, the backplane only uses 1/4 the rack.

Does it run 24 hours/day (as the high availability options would suggest)?
Alas, it currently is not in running order, I am short a few parts which I have found on the internet for reasonable prices refurbished however shipping is a bitch. The items are quality but large and heavy and I don't reside in the US where most of them were made. When purchased I do intend on buying quite a few spares though :)
The restoration of it to working order is a side-project of mine, the 5500 system itself is just old enough that it is not cost-effective to run it for computational items power wise, but when in running order I will occasionally fire it up to assist in some tasks, but not often. Primarily the high availability and somewhat unique architecture is what I have it for. Any "mission critical" items I have will run on it, since if they mean that much to me it will justify the power expenses.
To be honest I kind of wanted an SGI Onyx 3800, but because this machine was in non-functioning order and sun no longer support them, I basically got it for free. I just had to organize transport to get it to my house. the rack was 2 inches wider than my door, took some effort to get in :)
I am however looking at mounting some 1u compute nodes in the remaining space.

Why are you burning all that power all day? I'm hoping for an answer like: "I'm modeling a quantum electrodynamic system at 100 hours of processing per microsecond."
I'd hate to disappoint, but the most geeky of my uses would have to be verilog simulation (for cpu design). Finite element analysis for the structural integrity of some... items.. i'm working on. Finally I do on occasion do 3d rendering.. it takes as much cpu time as you can feed it, always :)
Of course this will soon likely be run on the linux cluster I'm installing in another quarter of the space the rack has free. For the moment I just have a four node cluster of c2 duos for these purposes that double as desktops for the rest of the family.
Re:Please tell me it's not used for entertainment. by professorguy · 2009-09-23 05:25 · Score: 1

verilog simulation (for cpu design). Finite element analysis.... 3d rendering
Thank you! You've restored my faith in humanity. I was frightened, but these uses actually require massive compute power. Rock on!
Re:Please tell me it's not used for entertainment. by Anonymous Coward · 2009-09-23 05:52 · Score: 0

I don't know about you, but I'd classify trying to find a zero of the Zeta function off the critical line burning power for nothing in that it's highly unlikelyt that you will find one but not finding one proves nothing.
Re:Please tell me it's not used for entertainment. by CompMD · 2009-09-23 06:25 · Score: 1

Sounds like the rack I got. I have an E6500 (8 CPU, 5GB RAM) with an E4500 (4 CPU, 4GB RAM) in the bottom of the rack. Each machine has 12 18.6GB hard drives. It took a little work to get mine running. When I got it, it had been turned on once (when it was installed) and shut down once (when three power supplies in the 6500 failed). There's some nice folks at a place called Serverworlds in Minnesota that should have whatever parts you need at a reasonable price. I got some used power supplies from them for pretty cheap, and I needed a new cabinet exhaust fan set, which I also got from them at a good deal. I have the 6500 running on a 208V circuit when I need it, its not kept running all the time. However, when I do fire it up, it is fast to boot and works great. I use the systems for computational fluid dynamics simulations using NASA CFD codes, and advanced 3D CAD. Some advanced engineering tools simply just won't run on commodity hardware or operating systems. Not the fastest thing in the world, but with 12 CPUs and 9GB RAM, and the fact that it is Sun Enterprise gear, I can't complain about what I got for less than $500.

octane, we're going to rock-tane by Anonymous Coward · 2009-09-23 02:48 · Score: 0

that thing called octane,
it swings with performance...

Of course by Anonymous Coward · 2009-09-23 02:54 · Score: 0

this is when all the windows fan bois comes out and scream that you are not funny and that it does not require a great deal more power to run it.

What OS does it run? by Anonymous Coward · 2009-09-23 02:57 · Score: 1, Interesting

I read TFA, apologies if I missed it, but, what operating system does it run? Please, for the love of all that is good, let the answer NOT be IRIX.

Re:What OS does it run? by xrayspx · 2009-09-23 03:30 · Score: 1

According the datasheet on their site, helpfully not linked in either the summary or article, it ships with RHEL5, SLES 10 or 11, or Windows Server 2008 or HPC Server 2008.

--
I like music
Re:What OS does it run? by Bill+Dimm · 2009-09-23 03:48 · Score: 1

2nd to last paragraph from TFA:

...It can be preconfigured with Windows Server or its HPC Server 2008, as well as Red Hat and SUSE Linux servers...

Deja Vu by Anonymous Coward · 2009-09-23 03:05 · Score: 0

Not the first to market on this, Cray came out with one last year, http://tech.slashdot.org/article.pl?sid=08/10/21/1217252

is a 1 cpu Xeon machine news worthy? by Latinhypercube · 2009-09-23 03:06 · Score: 1

O how the mighty have fallen. On a brighter note, I'm looking forward to the eight core xeons coming out this year....

Re:Despite the prevailing skepticism here, this is by gedw99 · 2009-09-23 03:07 · Score: 1

EXACTLY. Its just a box. It woudl be very easy to build on of these yourself i my opinion. And a bunch of ARM chips on mini arm motherboards and you have yourself a large capable web cluster. etc.

The Computer to End Geek Fights? by Comatose51 · 2009-09-23 03:08 · Score: 1

With that much personal computing power, just put virtualization on there and run as many VMs as you want to your heart's delight! No point in arguing the superiority of various OSes. Just make a VM of one and run it at the same thing as VMs of other OSes. If you want, spin up a few VMs and say you have a Beowulf cluster, which puts the whole idea on its head. Have your own virtual datacenter inside a single computer! If you want to be all MBA about it, create your own "cloud" inside. This is like Legos but for grown nerds (or more like second set of Legos since grown nerds still play with Legos).

--
EvilCON - Made Famous by /.

Yeah right by mister_playboy · 2009-09-23 03:13 · Score: 1

You can't fool me, Kilgore Trout!

--
Do what thou wilt shall be the whole of the Law ::: Love is the law, love under will

Yet Another Personal Supercomputer by DynaSoar · 2009-09-23 03:16 · Score: 2, Informative

Here's just a brief search for personal supercomputers of days gone (not too far) by. Most if not all are cheaper than the SGI. Being older they may not stack up spec-wise, and the definition will always be changing anyway. More than one claim to be 'first', and to SGI's credit they only claim it's 'their' first.

http://tech.slashdot.org/article.pl?sid=08/11/23/068234

http://www.researchchannel.org/prog/displayevent.aspx?fID=569&rID=4263

http://aslab.com/products/workstations/marquisk942.html

http://www.reghardware.co.uk/2006/06/07/tyan_unveils_typhoon/

http://www.hpcwire.com/features/Cray_Unveils_Personal_Supercomputer.html

--
"I may be synthetic, but I'm not stupid." -- Bishop 341-B

Re:Yet Another Personal Supercomputer by Hercules+Peanut · 2009-09-23 06:18 · Score: 1

Nice list but I think you forgot this http://www.apple.com/ca/press/1999/08/Seybold.html

perverted definition of supercomputer by peter303 · 2009-09-23 03:22 · Score: 1

The definition of a super is a system at the top order of magnitude of speed and memory. Since the current record is two petas, a super would be one hundred teras. A one tera computer is a super of a decade ago.

atom? by Crispy+Critters · 2009-09-23 03:25 · Score: 1

I can't find a price for the configuration with a bunch of atom processors, so it is hard to say whether it is an economical way to create a certain amount of processing power. I also wouldn't have the numbers to put a price on the possible savings due to lower electricity and cooling requirements. None of this matters unless you are running something that can use that many processors efficiently.

Re:atom? by fuzzyfuzzyfungus · 2009-09-23 03:37 · Score: 1

The thing I find really weird about the atom configuration is that it is less dense than the Xeon one.

Somehow, they can get 20 quad core firebreathing server processors into this box(with 6 RAM slots apiece, and 10 HDDs); but only 19 netbook processors, with only 2GB max of RAM each and all but one of the nodes diskless. How did that happen?

Killer app... by lurking_giant · 2009-09-23 03:32 · Score: 1

The system code named DESCH (named for Joseph Desch, who led the secret WWII project which developed a decoder for the Nazi Enigma encrypted messages.) collects terabytes of raw data gathered by the USAF Gotcha's synthetic aperture radar equipped UAV's orbiting over an area of interest in a war zone. The system images a 5km dia "city sized" view and processes the result into 3D image maps while recording to disk for review. The 400 Megapixel per second streaming images allow zooms into areas of interest, observation of minute changes and the ability to track personel and vehicals in the urban battle field. http://www.wpafb.af.mil/news/story.asp?id=123165818/

Re:Despite the prevailing skepticism here, this is by evol262 · 2009-09-23 03:41 · Score: 1

Let me know when you assemble what is essentially a blade chassis (it's really more comparable to the Onyx/Origin, Superdome, or Sun's E$numK systems given that you have the option of using PCIe nodes instead) with Infiniband interconnects in a deskside form factor, plus redundant hot-swappable components and the ability to run the entire thing as a single OS image yourself.

--
"The more corrupt a society, the more numerous are its laws." -Tacticus

Why is it better than a Tesla? by VGVL · 2009-09-23 03:44 · Score: 1

Why would I buy the SGI at 726 Gigaflops instead of a single Nvidia Tesla card and get 1 Teraflop? I could buy four cards and get four teraflops for a lot less money than the SGI. There probably is something else going for the SGI, what is it?

Re:Why is it better than a Tesla? by mr_death · 2009-09-23 18:44 · Score: 1

Simple -- the Teraflop number for the Tesla is single precision. Tesla using double precision is only good for ~100GF. Given SGI's horrid financial history and ongoing issues, I'd go with another machine -- the Cray CX-1, or an IBM BladeCenter.

--
It's Linux, damnit! Pay no attention to renaming attempts by self-aggrandizing blowhards.
Re:Why is it better than a Tesla? by mapesdhs · 2009-09-24 03:39 · Score: 1

This isn't the old SGI anymore, it's Rackable.
You're right about the double precision fp though.
Another bonus is the lower power consumption of an
integrated system compared to a cluster, which
means lower TCO.
Ian.

--
SGI Guru. Email: mapesdhs@yahoo.com | Tel: +44 (0)131 476 0796
Re:Why is it better than a Tesla? by mr_death · 2009-09-24 15:41 · Score: 1

This isn't the old SGI anymore, it's Rackable.
Well, SGI's business didn't get magically better just because they were acquired for a song. Rackable really doesn't understand HPC, so I think it will be a difficult integration. I'd further expect some gyrations on their future direction, as Rackable management figures out how to play the game. In any case, we're all in interesting times.

--
It's Linux, damnit! Pay no attention to renaming attempts by self-aggrandizing blowhards.
Re:Why is it better than a Tesla? by mapesdhs · 2009-09-28 03:30 · Score: 1

mr_death writes:
> Well, SGI's business didn't get magically better just because they were acquired for a song.
Very true!
> Rackable really doesn't understand HPC, ...
What is your evidence/rationale for this? Just curious. The feedback I had from the SGI CEO was rather positive.
> ... so I think it will be a difficult integration. ...
More difficult than the Cray merger? I doubt it. :D
> ... I'd further expect some gyrations on their future direction, as Rackable management figures
> out how to play the game.
If UltraViolet does well, they will be able to stabilise.
> In any case, we're all in interesting times.
Yup! I won't be ditching my IRIX SGIs anytime soon though. :D
http://forums.nekochan.net/viewtopic.php?f=3&t=16720350
http://forums.nekochan.net/viewtopic.php?f=14&t=16719759
http://forums.nekochan.net/viewtopic.php?f=14&t=16721054
http://forums.nekochan.net/viewtopic.php?f=14&t=16721251
(and numerous others)
Ian.

--
SGI Guru. Email: mapesdhs@yahoo.com | Tel: +44 (0)131 476 0796

G4 PowerMac is MY personal supercomputer by nitroamos · 2009-09-23 03:45 · Score: 1

Since apparently the definition of a supercomputer is a machine capable of 1 gigaflop, SGI was scooped by Apple 10 years ago!

enlighten me by confused+one · 2009-09-23 03:50 · Score: 1

Would someone please enlighten me as to how this is better than a cluster of commodity servers? Or a cluster of workstation class machines? Or a cluster of commodity servers with a workstation class machine as the head node? I'm not seeing it; and, the SGI looks pricey.

Re:enlighten me by dwater · 2009-09-23 22:30 · Score: 1

I would start the list with :
1) smaller
2) easier to maintain
3) less mess
4) cheaper running costs (power, aircon)
5) support/professional services from the manufacturer and from people who know your business
6) just easier overall
7) I expect it's faster too
8) you can make a cluster of these too
some of which is debatable, of course
So long as you can pony up, I see many advantages.

--
Max.

Eh by Anonymous Coward · 2009-09-23 03:52 · Score: 0

Check out Infiniband.

I suck at hardware, so I have no idea if you're going to get CPU-to-Memory speeds, but on the (relative) cheap, you can push 20Gb/s bi-directionally.

On the more expensive, you can do 40Gb/s+ bidirectionally.

The only caveat is lack of support outside of high-end storage/supercomputing clusters. :P

80 cores, 1TB of memory, and you got modded up? by SuperBanana · 2009-09-23 04:04 · Score: 4, Insightful

I've seen the term 'personal supercomputer' so many times over the past 20 years. It's just baloney marketing. What you have on your desktop RIGHT NOW is more capable than some of the original CDC machines. So what?

What you have on your desktop RIGHT NOW is most likely more powerful than the Cray Y-MP by a factor of three, if you've got a quad-core Core2 Duo; those babies push +1Gflop.

It's also 1/50th to 1/100th as capable as this supercomputer (or more- I don't know the relative performance between a current desktop processor and current Xeon.) Yes, it's relative, and relatively speaking, this is most certainly a supercomputer. In terms of memory, the maximum amount of ram you can put into a consumer-available motherboard is around 64GB, maybe 128. This has a maximum of 10 times that.

80 xeon cores, 1TB of memory, and you call it a "marketing ploy"? And you got modded up "insightful"? May the hand of metamoderation come on down from high.

--
Please help metamoderate.

Re:80 cores, 1TB of memory, and you got modded up? by Anonymous Coward · 2009-09-23 05:00 · Score: 0

quad-core Core2 Duo
I think you mean Core2 Quad.

Personal Supercomputing by CopaceticOpus · 2009-09-23 04:09 · Score: 1

It may not be enough to simulate global weather patterns, but surely it's powerful enough to predict the weather in my parents' basement. Drafty today? Time to pull out my electric "7 of 9" blanket and nuke a few frozen cheeseburgers.

Seriously though, I doubt that there is much demand for personal supercomputing.

Interconnect is NOT at FSB speeds by PeterM+from+Berkeley · 2009-09-23 04:39 · Score: 1

You're not getting Xeon FSB communication between any arbitrary two of the processors.

The machine's organized into boards with 2 CPU sockets on each, which can support a quad-core XEON. I suppose the two Xeons on the processor board could communicate at FSB speeds.

However, between boards, for interconnect you get either dual gigabit ethernet (2Gb/s) or QDR Infiniband (10Gb/s), a far cry below your claimed 170Gb/s.

In other words, the comparison to a cheaper network of workstations is really quite a fair one.

--PeterM

Re:Interconnect is NOT at FSB speeds by flux · 2009-09-23 05:31 · Score: 1

However, Infiniband latency (according to Wikipedia) is in the range of 200 nanoseconds, which is around 100-1000 faster than a gigabit network. That can make a huge difference in certain operations. Also it seems to me the Infiniband protocol is more suitable for clustering than just shoveling data over local network, with operations such as direct memory access and transactions.
Also I wouldn't say the 8Gbit Infiniband latency is anything to sneeze at, after all, 10Gbit Ethernet gear is just coming.
Re:Interconnect is NOT at FSB speeds by ceoyoyo · 2009-09-23 15:16 · Score: 1

The dual gigabit interlink does seem like a bit of a ripoff. Unless you're really squeezed for space you'd probably be better off building yourself a little cluster.
Infiniband is a completely different ballgame though, much more so than the theoretical signalling rates might indicate. On the other hand, you can also buy Infiniband cards for a regular computer, so you might still be better off rolling your own.

Sure. Done. by mmell · 2009-09-23 04:54 · Score: 1

On second thought - I'll live without the ultra-highspeed interconnects since 1000TX is all I can afford. Hot swapping individual components (nodes) will be strictly a software function, a dead node can easily be unplugged and replaced without too much fuss, although any jobs running on that particular node will need to be failed and restarted. I'm not interested in sub components such as power supplies, memory modules, cpu's, etc. - they're no longer component-level items but are now subcomponents and thus beneath my attention.

Single OS instance? I don't suppose you would consider TFTP/NFS to be solutions to this (the standard for IBM's BG and for Beowulf clusters)? I sure do - although I don't think Redmond's products will work quite that way I know that *NIX operating systems do. Think of each node as a fat client (although I still use them as thin clients, not needing the local SSD disks for anything more than swap space).

Form factor doesn't intrigue me. If that's important, I recommend you build your own case with lots of pretty transparent covers and neon lights/LED's. I'm sure you could come up with something straight out of an Irwin Allen show; I'll settle for nailing a bunch of power-strips to a hunk of plywood, personally. Don't forget - 10W per active node. Okay, at 1,000 nodes I'll be drawing 10KW; but for now, I'm running a LOT less than the 200W your average desktop box draws and getting lots more FLOPS per watt.

When all computers are super... by nadando · 2009-09-23 04:57 · Score: 1

...none of them are.

Obviously not the best solution for all apps by joggle · 2009-09-23 05:18 · Score: 1

Seems like this still wouldn't be the best solution even for what SGI used to be commonly used for--CG. If you are running 3ds Max (or some other rendering software) it seems you would still be better off with a dozen cheap, independent rendering nodes than one of these things decked out. This is because the nodes do not need to communicate with each other (only between the slave and master nodes, and that's only at the start of rendering each frame and at the end of rendering a frame) so there is no need for a high-speed connection between all of the nodes or to run it under a single OS environment.

Re:Man... - Graphics? by Anonymous Coward · 2009-09-23 06:21 · Score: 0

Isn't the whole point of the CUDA standard to use all the graphics cores (like 800 for ATI) for floating point? If you run two cards, that's 1600 1024-bit cores. Why would anyone choose Intel procs for floating point?

Re:Sure. Done. by evol262 · 2009-09-23 06:31 · Score: 1

If 1000TX had anywhere near the latency of Infiniband, maybe. There's no doubt that clustering has its point, but you cannot possibly believe that individual nodes communicating across 1000TX (or 1000FX, for that matter) are going to compete with a chassis like this in this segment. No, I don't consider TFTP/NFS/PXE booting to be solutions in this manner, since nodes in a cluster are not the same thing as a single image, and they have no access to peripherals which may be in the chassis (if one goes with the PCIe expansions).

The form factor matters not because of aesthetics, but because fast interconnects and redundant components matter for some applications. It is not a competitor to desktops.

--
"The more corrupt a society, the more numerous are its laws." -Tacticus

What??? This is Super??? by Nom+du+Keyboard · 2009-09-23 06:35 · Score: 1

This new supercomputer's peak performance of about 726 GFLOPS

Excuse me, but...this doesn't compare to a single ATI Radeon or Nvidia GPU card. How is this super?

Even if you're talking DP FLOPS, I think GPU computing has got to be more cost effective than this.

--
"It's the height of ridiculousness to say for those 9 lines you get hundreds of millions."

Re:PS3s WTF by Informative · 2009-09-23 06:37 · Score: 1

Not sure why you would say that unless you aren't aware that PS3s are being used to make HPC clusters for scientific computing:
http://www.top500.org/blog/2007/10/27/rough_guide_scientific_computing_playstation_3

Not very much bang for your buck.... by maroberts · 2009-09-23 07:03 · Score: 1

Wouldn't an NVIDIA Tesla based system give you a lot more horsepower for a lot less money, and a lot lower power consumption?

--

Donte Alistair Anderson Roberts - hi son!
Karma: Chameleon

Botnet != Supercomputer by GameboyRMH · 2009-09-23 07:54 · Score: 1

Botnets are useless for number crunching. There may be many CPUs involved but the communication between processors is dead slow and unreliable. The ability of botnets to send lots of traffic from a huge number of Internet connections in different locations makes them ideal for spamming and DDoSing.

Supercomputers have few or no Internet connections and have no more potential than office PCs for spamming or DDoSing. They do have many powerful CPUs that have no problem communicating with each other as fast as they can, which makes them ideal for number crunching. The use of a supercomputer to a black hat would be for breaking encryption (or they could be useful to comic book villains for simulating nuclear explosions), but a rack of PS3s might be faster than a botnet, and definitely more difficult to trace.

--
"When information is power, privacy is freedom" - Jah-Wren Ryel

Re:Botnet != Supercomputer by Anonymous Coward · 2009-09-23 15:02 · Score: 0

You are an idiot

Party like it's 2002 by nanojath · 2009-09-23 08:26 · Score: 1

But... when I bought my iMac G4 Steve Jobs told me it was a supercomputer!

--

It Is the Nature of Information to Transgress Artificial Boundaries

Got that last one right. by mmell · 2009-09-23 08:43 · Score: 1

Not even remotely for the desktop. Computers are for the desktop. Supercomputers are special application tools. Most competent programmers have a hard time thinking in terms which lend themselves to widespread parallelization. Software for massively parallel environments is virtually non-existent, except for the few rare instances of software created to be distributed (e.g., setiathome, foldingathome, the prime-search application, etc.). Even these were written as distributed applications, not truly parallel ones.

For my budget, yes, I need to make quite a few concessions. Gig copper comes with the plugs and it's as fast as I'm likely to be able to afford; but I can still handily outrun SGI's offering in the 'bang for buck' category clear up until I hit around a hundred, maybe even a couple hundred nodes.

Re:Got that last one right. by evol262 · 2009-09-23 09:11 · Score: 1

Given the form factor, I'd say, yes, for the desktop (in the same way as an E3k, deskside Onyx, AlphaServer 2100, et al). The difference between distributed and parallel is essentially zero other than swapping client/server to parent/child threads. A system like this would be ideal for localized rendering jobs, virtualization, and a number of other tasks that may be undertaken. More than that, it's modular. You don't need to get ridiculous amounts of licenses for Vmotion, Windows, ESX, or whatever it may be, and it can be expanded as the business gets the money.

Frankly, this is not aimed at your budget or what you can afford. I'd question your ability to compete on price/performance for all but the most trivial tasks also (a fully loaded Octane III is damn close to list price on the components).

--
"The more corrupt a society, the more numerous are its laws." -Tacticus

Programmer's dream... by mx_mx_mx · 2009-09-23 10:34 · Score: 1

just imagine how fast compiles will be done on this system.....

make -j80 ...

--
Linux forever

Re:Despite the prevailing skepticism here, this is by Anonymous Coward · 2009-09-23 12:59 · Score: 0

SheevaPlugs are cool (I have one), but without an FPU they are pretty useless for supercomputing.

pricing? by OrangeTide · 2009-09-23 13:55 · Score: 1

I had difficulty finding pricing quickly enough to make a simple slashdot post about it. The pricing is non-obvious.

--
“Common sense is not so common.” — Voltaire

To Expensive for What you get by stoicio · 2009-09-23 15:43 · Score: 1

I just priced out hardware for a 64 core MPI parallel computer system.
The commodity hardware came in at ~ $7800.00.

The SGI base system only comes with one [*ONE!!!*] processor at
near the same price....

They must be offering one hell of a warranty to ask that price
for essentially a high end workstation. (A nice colorful SGI case maybe.)

As far as 'Easy' goes; Well, most of the apps that use supercomputing
these days require parallel processing. Maybe the parallel API they are
using is something special. If the software is provided for free
it could be worth it. Probably not....

Re:To Expensive for What you get by mapesdhs · 2009-09-28 03:40 · Score: 1

Exactly what did you 'price out' in a single box with proper redundant PSUs, etc.?
With a maximum CPU setup, Octane III offers 50% better price/performance than a bunch of Dell servers.
For inter-node coms, the interconnect is GigE or Infiniband, much the same as for any cluster. The
sw is no different (doesn't need to be).
If Rackable (now SGI) has carried on with the same traditions as the old SGI, the support will be very
good. I adminned an Onyx2 16-CPU 5-pipe IR2E, a dozen O2s, dozen 320s, CAVE and RealityCentre for nearly
4 years and the support from SGI was always excellent. I hope the new SGI keeps the same levels of quality.
Ian.

--
SGI Guru. Email: mapesdhs@yahoo.com | Tel: +44 (0)131 476 0796

Well someone has to say it. by ignavus · 2009-09-23 16:13 · Score: 1

960GB of RAM should be enough for anybody!

--
I am anarch of all I survey.

Seems an excellent offering by GauteL · 2009-09-23 20:16 · Score: 1

First of all, this is a shared memory system, meaning it isn't really comparable to a cluster-supercomputer using distributed memory. Building a big shared memory system isn't usually something done by hobbyists or regular sysadmins.

If you need a shared memory system, you need a shared memory system and no amount of cluster nodes can do the job for you.

Instead this is more like an entry, low end, version of some of SGI's other offerings such as the Altix 4700 which supports up to 1024 cores in a shared memory system.

That you can get an 80 core shared-memory system with 240GB of memory for $53k is news to me, but then I haven't been watching the prices of these systems lately. Also the price point is certainly attractive enough that it may make sense to buy one of these rather than convert your shared-memory code to MPI, even if such a transition is possible for your purpose. Given complex enough algorithms the transition could easily cost you much more than that in manpower.

Actually, I just realised my posts probably reads a bit like SGI astroturfing....

Re:Seems an excellent offering by mapesdhs · 2009-09-28 03:24 · Score: 1

Octane III is _not_ a shared memory system. You'll have to wait for UltraViolet for that.
Should be obvious it's not btw, as current i7s don't have the relevant NUMA support.
The default interconnect is GigE, with Infiniband optional.
In August I helped a company source three Dell R710s for rendering moderately complex
3DSMax models (24 cores total, all i7 XEON 2.93GHz, 96GB RAM total). The Octane III
would have offered nearly 50% better price/performance. Ah well...
Ian.

--
SGI Guru. Email: mapesdhs@yahoo.com | Tel: +44 (0)131 476 0796

I've got a personal Cray 1 supercomputer at home. by ciderVisor · 2009-09-23 21:50 · Score: 1

Well, not specifically a genuine Cray 1. However, the cheap-as-chips generic Windows PC that I bought at the local electrical warehouse pisses all over the performance of that archetypal 'Supercomputer'. All the figures trotted out these days for the Top 500 are just so much willy-waving. Ridiculous.

--
Squirrel!

Distributed computing system != botnet by GameboyRMH · 2009-09-24 02:19 · Score: 1

If a botnet relied on a consistent connection to any kind of single centralized server, it would run for just a few hours (maybe longer, depending on where exactly it's located) before the IP was blocked and a SWAT team is dispatched to the server's location. Folding@Home wouldn't work very well if ISPs and private firewalls were trying to block it, virus scanners were trying to remove it, and the server was only online for short periods from differing locations. It might sort-of work if some technical problems can be overcome (and depending on exactly what kind of calculations are being done), but it wouldn't work well.

--
"When information is power, privacy is freedom" - Jah-Wren Ryel

Re:Distributed computing system != botnet by Anonymous Coward · 2009-09-24 02:56 · Score: 0

http://www.informationweek.com/news/internet/showArticle.jhtml?articleID=201804528 Russian botnet operators lease out botnet time for computationally intense stuff all the time (usually breaking encryption).

Not exactly old-style SGI, but still a good spec. by mapesdhs · 2009-09-24 03:33 · Score: 1

Although Octane III is a far cry from the MIPS/IRIX systems which the
product name harks back to (and I should know, I have 60 of them), it
is a decent design nonetheless, even though I'm sure the aesthetics
will not appeal to many. Strangely for me, it looks rather like my PC
from the front (Centurion Plus 534 case).

I recently had to spec out a 24-core renderfarm for a small design
company in Spain, consisting of three Dell rack servers, each dual
quad-core i7 XEON 2.93GHz, 32GB ECC RAM, etc., which came to
16000 UKP + tax (list price). From the information I have at the moment,
an Octane III with a similar spec (ie. six CPUs, 96GB RAM) would provide
nearly 50% better price/performance, which is very respectable indeed.

I'll see if I can obtain some pricing, find out whether this system
does compete to a decent degree. It absolutely does with 80 cores, but
a more mid-range spec will be of greater interest to the companies I
typically deal with.

Ian.

--
SGI Guru. Email: mapesdhs@yahoo.com | Tel: +44 (0)131 476 0796

PR0N by Whorhay · 2009-09-24 05:10 · Score: 1

Set it up to constantly crawl the web and download all the latest pr0n, then sort and categorize it all before finally prioritizing and assembling it into a nice presentation mode.

It does not have fast interconnect?! WFT! by zvrk · 2009-09-24 05:55 · Score: 1

Specs suggest that it is just a cluster under your desk (and a hot one if I can notice). No fast interconnects - i.e. it's just another custom beowulf...
To slashdot - overload a? This comment will not be saved until you click the Submit button below. You must wait a little bit before using this resource; please try again later.

Slashdot Mirror

SGI Rolls Out "Personal Supercomputers"

303 comments