Earth Simulator, G5 Cluster Drop In 'Top 500' List

daveschroeder writes "The November Top 500 supercomputer list has been published at SC2004. Topping the charts is IBM and the US Department of Energy's 'BlueGene/L DD2' beta system, at 70.72 TFlops, followed by NASA's 'Columbia' at 51.87.TFlops. For the first time in several publications of this list, Japan's Earth Simulator is no longer in the number one slot, falling to third. Virginia Tech's 'System X' Xserve G5 cluster, while 20% faster than the original cluster that debuted at number 3 last November, has fallen to number 7 due to the new entries, but remains the fastest supercomputer at an academic institution. Here's an excellent cost comparison (Google cache) of the top machines ('System X' is significantly cheaper than anything else in the top 20, not to mention cheaper than many things far below it in performance)."

Comments

Hmm

[foxalopex]

Err, I'm not sure if the costs can be accurately compared in this way. One needs to remember that a cluster of separate computers acting as a supercomputer compared to a custom designed hardwired system isn't exactly the same thing! Otherwise you can start comparing stuff like SETI which I'm sure is the world's cheapest supercomputer because it technically didn't cost anything to SETI themselves.

Re:Hmm

[Anonymous Coward]

Viruses are the new supercomputers.

Re:Hmm

[Anonymous Coward]

I'd guess the servers that send and recieve data from SETI clients aren't exactly cheap. Cheaper than building a more traditional cluster, but it certainly wasn't free.

Re:Hmm

[MidnightBrewer]

If you put conditions on the comparison, such as the computers are all housed at the same location and the money came from the same place, then the comparison works fine. Donations don't have to be counted, because that's up to each institution's discretion and individual efforts to collect.

Re:Hmm

[Wesley Felter]

All the systems on the Top 500 list are benchmarked running Linpack. If you can run Linpack on SETI@home, you're welcome to count it.

Re:Hmm

[Ibanez]

Well then, why not compare the earth? What other supercomputer has gotten closer to "42" than it?

Re:Hmm

[JamieF]

Agreed! I only trust benchmarkers who are physically attractive.

Yes, Ms. Booth Babe, I can see your points... uh, point. Flops? What are you talking about, they're quite firm in fact. OH! Oh right, FLOPS. Gotcha. Right. Yes. What's that? Hey baby, you can swipe my badge anytime.

Re:Hmm

[dustpuppy_de]

Deep Thought, actually. This supercomputer calculated that very number, as the answor to the Ultimate Question of Life, the Universe and Everything. Earth, Deep Thought's successor, was pretty close to calculating the question, without which the answer obviously doesn't make very much sense - but as we all know, it was destroyed by the Vogons to make way for a new hyperspace bypass.

More Information here.

Re:Hmm

[Flaming Foobar]

Otherwise you can start comparing stuff like SETI which I'm sure is the world's cheapest supercomputer because it technically didn't cost anything to SETI themselves.

I don't think SETI is eligible for that list at all.

Firstly, its performance isn't guaranteed. It is entirely possible people get fed up with it and stop donating CPU cycles tomorrow, which would make the processing power go significantly down.

Secondly, it doesn't (and really can't) have a proper operating system. You couldn't, say, play Doom 3 on it. It is designed to do a very specific task. I suppose something like SETI could be used for running java tasks or such, but it would still require that each node gets a complete task to run, because any of them can drop out anytime.

Re:Hmm

[Anonymous Coward]

While informative in your correction, you are also hopelessly German.

Re:Hmm

[flonker]

Norton Antivirus: Fighting Back Against Skynet!

Re:Hmm

[Tim C]

it technically didn't cost anything to SETI themselves

Well, presumably it's a tiny cost compared to that of a true supercomputer, but SETI costs them in bandwidth - both client downloads and updates, and uploading packets of work and downloading processed packets. Even if SETI itself isn't paying for the bandwidth, *someone* is...

Re:Hmm

[faragon]

I can't understand why you've got a "Funny" score, while could be more adequate "5, Interesting".

Re:Hmm

[Anonymous Coward]

Off topic and thread:

Mods, please browse at "0" for the common good. Some gems always hidden. (This post isn't one of them...)

Thanks for your pains in advance, and sorry for this interruption.

Re:Hmm

[Anonymous Coward]

Exactly. The new #2 on the list is a cluster of 512-processor machines. The G5 clusters are at best clusters of 2-processor machines.

There is, believe it or not, a difference.

I love how the "excellent comparison" asks "Why did Virginia Tech choose Apple G5 over Opteron or Itanium?"

I mean surely you could ask, "Why did NASA choose Itanium over Apple G5 or Opteron?" NASA are on record as having made their choice of system a full 7 months after VT's machine was announced....

Re:Hmm

[Anonymous Coward]

I guess you are referring to capacity computing vs. capability computing. Just for those who are new to the terms, capacity computing loosely refers to a bunch of computers connected by a fast LAN. Capability computing refers to a bunch of CPUS connected by a high performance memory subsystem.

To further illustrate, consider the SETI@Home application. You can run multiple instances of SETI@Home in parallel, and the individual instances don't need to communicate with each other. The notion of parallel jobs that don't communicate with each other is known as "embarrassingly parallel". This type of application is well suited for capacity computing, since the individual processes don't need to communicate over the fast LAN. Even though it is a fast LAN, it's doesn't mean much, even the fastest LAN has latencies that are orders of magnitude slower then a high-performance memory architecture.

If you have an application where the individual processes (or threads) do communicate then the performance of the application is heavily influenced by the degree of communication and the speed of the network (or memory architecture).

So I guess the moral of the story is, the supercomputer you choose depends on the how much the individual threads of your application communicate with each other. Just because a cluster achieves many Gflops, if it is connected by a gig-Ethernet interconnect and your application needs to communicate a lot, you will get crappy performance.

Re:Hmm

[mjeppsen]

Wow. Parent is modded "Insightful"? Incredible. I guess "+5 Funny" was disabled...

The Dept. of Energy

[aztektum]

I hear they're using it to convert heat into electricity for the rest of the government. Hence their name.

Re:The Dept. of Energy

[thegoogler]

No, there using it to keep the earth warm.. you think that molten core stays molten all by itself?

Pizza arguments

[daveschroeder]

A few more thoughts...

Before anyone says "Of course System X is cheaper! Virginia Tech had free student labor to put it together! They paid them in pizza!"

The only thing anywhere close to System X is NCSA's Tungsten, a 2500 processor Pentium IV Xeon Dell Linux cluster. It cost $12 million, just for the asset (comparable to System X's $5.8 million overall price, including the upgrade to Xserve G5s). That's twice the cost, and over 2Tflops less performance. 2Tflops is a top 100 supercomputer...so it's a whole top 100 supercomputer poorer in performance, for an extra $6.2 million.

Another example is PNNL's 1936 processor Itanium2 cluster: 3.5Tflops less performance than System X, for $25 million.

Any way you slice it - no pun intended - System X is still a LOT cheaper, even if you allot, say $2M for professional installation and systems integration - an EXTREMELY liberal estimate, probably by an order of magnitude.

System X also has the highest Rmax per CPU of any system on the list, except for specialty non-commodity systems like Earth Simulator.

And on top of it all, last November, they hit #3 in the world, #2 in the US, and #1 academic, as well as the first academic site to ever exceed 10Tflops, all for less than $7 million in total - including all improvements to buildings, physical plant, and other infrastructure.

That first system might not have had ECC, but what it did do is break into the top 5, following all the rules of the Top 500 organization, for relative pocket change - for a price that was absolutely unheard of, sharing the spotlight with systems that cost $100 million or more - and also catapulted Virginia Tech to a supercomputing center of national prominence overnight, able to attract additional attention, funding, grants, and publicity. Not to mention testing and proving the suitability of a completely new OS, platform, processor, and interconnect for high-performance computing, increasing choice for all (and resulting in new clusters based on the same technology, such as the US Army/COLSA cluster). And even as new systems enter the top ten in the tens and hundreds of millions of dollars, System X retains the title of #1 at any academic institution, and shares the top 10 with the best of the best.

Seems to me that Virginia Tech pulled a real coup here, and a full year later, is still considerably cheaper that anything else. And now, it's being used for real scientific work. To bring a whole new platform onto the scene in essentially under a year and break into the ranks of the supercomputing elite virtually overnight, and to do it significantly, and sometimes ridiculously, cheaper than everyone else, is a feat that can't be ignored.

Re:Pizza arguments

[thpr]

Before anyone says "Of course System X is cheaper! Virginia Tech had free student labor to put it together! They paid them in pizza!"

No, my real question would be: What is the ongoing operating expenses of System X? After all, I'm interested in total cost of ownership, not in acquisition cost.

Re:Pizza arguments

[daveschroeder]

I'm sure it would be similar to any Linux cluster in the top 10; there's no reason to believe it should be any different or require significantly different levels of system administration and maintenance.

Re:Pizza arguments

[CODiNE]

I remember the power usage and heat generation was lower than the competition... they were able to save a lot of money on cooling systems by going with the G5's. That has to be saving them a lot of money day to day. I don't have the wattage numbers on hand right now, but I do remember G5's beating P4's, Opterons and especially Xeons, so there's no way Virgina Tech is paying what those x86 top ten people are in energy bills. The XServe's keep track of their fans and internal temperature, automatically letting the admin know when a machine is GOING TO FAIL and preemptively swap out faulty components before they cause serious damage to the systems. I don't know what sort of Apple Care that kind of system comes with, but I'm sure it's competitive with their consumer systems, I'd imagine any hardware failures are completely covered absolutely free for at least the first year, probably three or more if they pay Apple for any support.

-Don.

Re:Pizza arguments

[natet]

You are comparing apples to oranges when all you look at is the total cost of the system. First of all, the VT computer doesn't have the same level of support as other computer installations. I can only speak directly of the PNNL computer, as until recently I was on the team of administrators on that computer, but we had a team of hardware engineers on site to handle any of the hardware problems that occur in the operation of the system. Second. the PNNL supercomputer has nearly a half a pedabyte of disk. Yes, that's nearly 500 TB of disk space. That is where the bulk of the $25 million was spent. We purchased such a large amount of disk because the computer was being built to solve a specific class of problems.

Price per flop is a very poor mesure of the value of a supercomputer. All it amounts to are some good sound bites.

Re:Pizza arguments

[hernick]

A major cost is power and cooling requirements. According to Apple, a single drive, dual processor 2.0GHz XServe will use about 250W peak. Virginia Tech has 2.3GHz machines and Infiniband PCI-X cards. There's also networking gear and other support equipment to consider. So, we'll use a high figure of 350W per node, giving us a 40% overhead. As for heat production, same overhead, 1200BTU/h per node.

We're going to consider the worst-case scenario, under which we have a 100% load, year round, on all 1100 nodes. That gives us a power consumption of 385kW and 1320kBTU/h of heat generation.

Now, we need to get rid of that heat, and that's going to require a lot of power. My research indicates up to 300kW may be required, but that's a high number and actual requirements may be lower.

So, here we are, with 685kW required for power and cooling. That means a 6000MW/h a year.

Now, the cost of power is high, since you need to amortize and maintain the UPS equipment and the generators. We'll use a figure of 0.15$/kW/h, or 150$/MW/h. Very generous.

So here we are. The absolute worst case for power and cooling. Full load, year round, expensive cooling, overpriced power and amortized UPS and generators.

900 000$/yr. Below a million. It's not that bad, is it ? The real cost is likely below a half-million.

As for the rest, well, how much pizza is really required to entice graduate students and professors to work on that machine ?

Re:Pizza arguments

[adiposity]

Well, I'm not going to disagree with anything you've posted here, but are we sure Apple charged these people anywhere near full price for the hardware? Just a question, I don't actually know the answer.

-Dan

Re:Pizza arguments

[jshriverWVU]

That doesn't sound bad at all. I went to a seminar at the Pittsburgh Supercomputing center a couple years ago. (First time I ever saw a Cray *sigh*) hehe..

Anyway one of the guys there told me they had an older cray model just sitting in storage because it cost to much to use Over $1million a month for power and the cost of the coolant Cray's used. My memory might be off, but that makes the VT cluster even cheaper with respect to power consumption. Also no specially needed coolant other than an AC.

Re:Pizza arguments

[Anonymous Coward]

"And on top of it all, last November, they hit #3 in the world, #2 in the US, and #1 academic, as well as the first academic site to ever exceed 10Tflops, all for less than $7 million in total - including all improvements to buildings, physical plant, and other infrastructure."

I'm not trying to take any thing away from the VT cluster, but the prices quoted on the earth sim always include the total cost of the specifically designed, earthquake proof building and everything associated with the project, whereas prices for the VT cluster negate this and only count the hardware VT had to pay for - none of the freebies.

How is it one can get a dell for less than an apple but mysteriously as soon as it becomes a cluster (using the same interconnect) the G5 is cheaper? apples to apples....

Re:Pizza arguments

[SorcererX]

$0.15? here most house owners pay $0.07 per kWh, and companies and schools get lower prices as they buy high volume, but assuming $0.07/kWh, that'd bring the price down to $420,000/year.

Re:Pizza arguments

[Anonymous Coward]

Thank you, but you already posted this same analysis a year ago, MR. STEVE BALLMER.

(Yes, we know that you invested in Apple to maintain an appearance of real competition in the marketplace. We simply are too clever for you here at Slashdot. Nyahahaha! Can't fool us!)

Re:Pizza arguments

[Anonymous Coward]

The XServe's keep track of their fans and internal temperature, automatically letting the admin know when a machine is GOING TO FAIL,

Huh? Very few of all the possible faults in a computer have any kind of temperature change associated with them. When a memory stick, a HD, a CPU, a system chip, a connector fails, you don't get any overheating. (You may get a slight temp decrease after the component has failed and stopped consuming power.)

Your scenario only applies to failing fans and heatsinks prying themselves loose... not very common occurrences.

Sorry, but you have bought into some not very tech-savvy information somewhere.

BTW, it has been establisehd in every discussion here and in specialised hardwarwe sites that the power draw of a 2 GHz G5 is on par with a 2 GHz K8. This comparison is made harder by IBM not readily giving out the max power but a "typical" power, but valid estimations have been achieved nevertheless.

There is a reason why the 2.5 GHz G5 couldn't be air-cooled quietly. (It could have been air-cooled, but not quietly. I have a 2.4 GHz Athlon 64 system that is whisper quiet with a Zalman CNPS7000-AlCU at 1300 rpm and two 120mm case fans at 1600 rpm. Go figure.)

IBM is a fabulous foundry, but so has AMD been recently (avoiding most of Intel's 90 nm problems, for instance). The 970FX is just a really good processor based on a damn good architecture, not an engineering miracle.

Re:Pizza arguments

[Tim C]

Minor nitpick, but I have a Physics background, so accurate units are important to me - do you really mean MW/h, or do you mean MWh? That's a *huge* difference in energy consumption, depending on which you choose and how long the thing is running...

(I'm assuming you meant MWh, as in "mega-watt-hours", of course)

Re:Pizza arguments

[Anonymous Coward]

Does anybody know what OS they are running on System X?
Are they actually running OSX?

Apple always promotes the hardware side of System X but I've yet to hear them say anything about the OS....

Anybody know?

Re:Pizza arguments

[grey1]

Hmm, I don't buy this "System X is better than anything (oh apart from a few systems that are specialty non-commodity systems)". I think NASA just bought a very large system from SGI, had it delivered and running in 4 months, and it uses standard SGI Altix systems as pieces of the cluster. No doubt many of the others are put together that way too.

The Earth Simulator was a wierd example - don't let that example blind you to what others are doing.

Re:Pizza arguments

[Anonymous Coward]

Dave,

I have a question. Can you point us to any scientific work actually performed on System X? I haven't been able to find anything concrete, and I know for a fact that for some time there can't have been any science done after the first round of benchmarks, since they quite literally took the whole thing apart and replaced it with completely different machines. This doesn't sound like the sort of thing anyone who cares about "cost of ownership" would do.

I love the "and now" it's being used for real scientific work. A full year after it was installed. What the hell took them so long? Can't have been because it was a first of a kind machine - so is the current #2, Columbia. The Altix 3000 BX line was announced only last week, and the new Itanium 2/9M CPUs that power it were formally launched by Intel earlier today.

Re:Pizza arguments

[reso]

virtually nothing! its run on lab rat and monkey feces which is converted to fuel to power generators. ah, the circle of life.

Re:Pizza arguments

[Anonymous Coward]

The apple cluster, last I heard, could hardly be called a supercomoputer. More of a stunt really. Meant to run linpak, and not much else. Then they will break it up into small clusters for real work.

The top500 list really needs to be revamped into a list of allocatable and usable for real science top 500 list.

Re:Pizza arguments

[Anonymous Coward]

Real supercomputers include things like:
Major storage systems
Real networks (both messaging and otherwise)
Major amounts of memory

All these things cost money.

Re:Pizza arguments

[Doctor Memory]

Yepper, I can vouch for that. My folks used to live close to Sandia Labs, and they had a Cray-I listed at one of their "garage sales" for $1000, my dad talked them down to $500 (Dad: "Would you take $500?" Sandia: "Yeah, sure"), then found out he'd need 208V 400Hz three-phase power (they weren't selling the MG set the Cray came with) to the tune of $900K/month. "That's why we're selling it." the Sandia guys said. Someone else eventually picked it up, don't know how much they paid, but it had been crossed off the list by the time we left.

Re:Pizza arguments

[justins]

To bring a whole new platform onto the scene in essentially under a year and break into the ranks of the supercomputing elite virtually overnight, and to do it significantly, and sometimes ridiculously, cheaper than everyone else, is a feat that can't be ignored.

Judging from the Top 100 list itself, it has been pretty thoroughly ignored by the "supercomputing elite" of which you spoke. Most likely because the Linpack benchmark isn't often indicitive of a supercomputer's usefulness, clustered Linux machines are more popular on the low end of supercomputing, and proprietary systems are more popular (and necessary) on the high end. Not to create a false dichotomy, there: Linux is running on the high-end systems now, too.

I also wonder if you can actually buy the software VT developed to manage their cluster. Or establish the special relationship VT had with Apple, which allowed them to work around Apple's supply problems.

Re:Pizza arguments

[frankie]

Virginia Tech paid Apple's standard educational price, which is about a 10% discount on retail (and tax exempt). The only special break from Apple was that VTech got the first 1000 G5 towers off the assembly line, before other customers and stores.

Re:Pizza arguments

[neuroklinik]

I visited the System X installation about a month after it was first installed (when it was still G5 towers). They run Mac OS X on those machines. At the time, it was Jaguar. Tiger will make a big difference in the usability of these clusters, with better support for the 64-bitness of the PPC 970.

Re:Pizza arguments

[ThosLives]

Absolutely: Xgrid

Re:Pizza arguments

[Anonymous Coward]

Mod parent up for using "a lot" and "allot" *correctly* in the same sentence!

Re:Pizza arguments

[whizzer1187]

You might want to consider that one is not likely to see anywhere near peak power consumption. That would require the memory controller and all of the processing units (multiple AltiVec units, multiple ld/st units, multiple integer units, and multiple fp units) to be active. The units that aren't used for awhile (order of a few seconds) are turned off and don't draw power. It's a nice feature.

Re:Pizza arguments

[Anonymous Coward]

Grandparent took into account the cost of infrastructure (new lines to the building, UPSes, etc) as part of the per kWh price.

Re:Pizza arguments

[LinuxHam]

The XServe's keep track of their fans and internal temperature, automatically letting the admin know when a machine is GOING TO FAIL and preemptively swap out faulty components before they cause serious damage to the systems

Just like the IBM xSeries servers I'm deploying right now. Our servers automatically order replacement components if/when components fail. This includes CPUs, memory, fans, hard drives.. just about anything. IBM big boxes have been doing this for decades, and distributed systems for a number of years now. Don't get too excited.

Re:Pizza arguments

[b-baggins]

---
How is it one can get a dell for less than an apple but mysteriously as soon as it becomes a cluster (using the same interconnect) the G5 is cheaper?
---
Because the Dell you get for less than a Mac is a piece of crap Dell. When you get the high-quality Dells required for this sort of cluster, the price difference shifts to Apple's favor.

Re:Pizza arguments

[Anonymous Coward]

And now, it's being used for real scientific work.

And that's the difference between System X and the Tungston - the Tungston went into production a year ago while the main thing System X has done is make a mess of our server room. What's a year's use of a supercomputer worth?

Don't get me wrong, I think the System X is nifty and all, but the fanboyism (not accusing you of fanboyism, just commenting on many comments that I've seen) on /. gets old when it has as of yet to do anything useful.

Re:Pizza arguments

[Gorbag]

Our servers automatically order replacement components if/when components fail. This includes CPUs, memory, fans, hard drives.. just about anything.

That explains all those empty boxes for processor upgrade cards and memory when I got home last Friday, the damn computer was upgrading itself again. I'm not looking forward to getting my Visa bill this month...

Re:Pizza arguments

[justins]

Not what I was talking about, but okay...

Re:Pizza arguments

[markhahn]

VT demanded a price of $300/node (INCLUDING switches!) for interconnect, and was laughed at by several vendors. they wound up with the set of vendors who were willing to subsidize as the cluster as a loss-leader. everyone in the community knows that the cluster was non-repeatable, and thus $5M is not the actual price.

Re:Pizza arguments

[SavoWood]

Any way you slice it - no pun intended - System X is still a LOT cheaper, even if you allot, say $2M for professional installation and systems integration - an EXTREMELY liberal estimate, probably by an order of magnitude.

I'd like to thank you for being the first person in /. history to correctly use the words/phrases "allot" and "a lot" in a post on the site. Twogether on an single runon sentense. Uhmayzeng. =-)

Tongue in cheek - instructions for the humor impaired.

Re:Pizza arguments

[evilviper]

Now, we need to get rid of that heat, and that's going to require a lot of power.
Not at all. Air conditioners use lots of power, but a cooling system that, instead, uses many large exhaust fans would be every bit as good, and use up a very tiny ammount of power.

Incidentally, I happen to be a fan of ducting the input and exhaust directly, to improve cooling/heat exhaust massively. Ducting is cheap, and does a hell of a job, esp. when combined with a few exhaust fans, and perhaps a chilled water system to cool the incomming air.

My research indicates up to 300kW may be required, but that's a high number

An incredibly high number, IMHO, though that may be the case with this particular half-assed setup.

Re:Pizza arguments

[Anonymous Coward]

LOL

I see the pe1750 for $3166 (no discounts) with dual 3.0-1M + 1G mem + 2x1000M ethernet + 73G HD which uses scsi and no ATA option. Not mentioning the redundant PSU option or the remote management option (no this is NOT remote desktop).

the apple cluster node with 1G is $3249...

as I said LOL!!!

Re:Just imagine

[Spodie!]

Hell, I'd just love to have access for a weekend and load up Folding@Home on it. Love to see those stats!

And we thought Macs were expensive

[Anonymous Coward]

Guess not.

Re:And we thought Macs were expensive

[cbreaker]

You only have to buy 1100 of them to get a discount.

Power architecture does well

[blamanj]

5 of the 10 top machines use the Power archtecture, either the Power4 or PPC family.

Re:Power architecture does well

[ghutchis]

And 8 of the top 25 by my count. Not bad at all--and quite a chance from the last few rankings, where Intel really ruled.

Re:Power architecture does well

[Brett Johnson]

It is actually 6 of the top 10, and 13 of the top 25.

Re:Power architecture does well

[Brett Johnson]

I think it is interesting that 11% of the top 500 are Power architecture, and 64% of the top 500 are intel based systems. Yet 50% of the top 10 are Power architecture and only 20% of the top 10 are intel architecture. Also interesting is that the Power based systems seem to have twice the Mflop/dollar ratio over the intel systems.

Re:Power architecture does well

[Anonymous Coward]

I think it is interesting that 11% of the top 500 are Power architecture, and 64% of the top 500 are intel based systems. Yet 50% of the top 10 are Power architecture and only 20% of the top 10 are intel architecture. Also interesting is that the Power based systems seem to have twice the Mflop/dollar ratio over the intel systems.

That's because x86 is a horrible architecture. On top of that, x86 instructions are translated into microcode before they're executed, so you end up with an unknown (maybe you could ask the folks at Intergraph about it) architecture emulating a crappy architecture in hardware. Better architectures exist (ARM, MIPS, POWER, 68000, PA-RISC, toy architectures used in introductory computer architecture classes, everything else), but Intel won out in the marketplace. You can still get better chips, but you pay more and have less support.

That's why you'd be better off investing in AMD over Intel. AMD hit upon what Intel should've done years ago. The x86-64, for those who don't know, supports x86 binaries as well as its own new architecture. Think of it like an x86 chip with the underlying hardware exposed. If Intel had exposed the hardware that x86 instructions get translated to, they'd have had a clear upgrade path instead of having to dork off x86 out of the blue. AMD embraced and extended x86, and marginalized its future without doing any actual damage to it or x86 users. It's flat out genius.

In the meantime, almost anything performs better than x86, and with less power consumption. It makes those mini-ITX boards look like jokes, because instead of engineering a low powered MIPS board/processor, the VIA folks did another x86. It may have been good from a business point, but it's horrible from an engineering standpoint, and that sums up Intel and x86 fairly nicely.

Re:Power architecture does well

[Anonymous Coward]

You're either just misinformed, or karma-whoring with Captain Obvious statements.

The current x86 CPUs are really good. Come to think of it, RISC never fulfilled the main promise: simple cores that run at a high clock and of which you can pack many into one chip. (POWER4 and up are obvious exceptions -- the dual-core layout wasn't feasible for FX970, for instance.)

Think of x86 as an efficient code compression format. The decompressors (decoders) on current Athlons and Pentiums are really good, too. When you say "better architecture", bring into the picture the consumption of memory bandwidth and cache space, not just some partly academic "elegance".

AMD64 is well planned and executed, but it's not the miracle you make it out to be. Just new 64 bit instructions, and very importantly, more (32/64 bit)general purpose registers. It doesn't expose anything any more than the K7 Athlons did. (Where did you get that fantasy?)

There are tons of Mini-ITX boards in use, so the joke quite seems to be on you. The big thing which you completely sidestep about them is, of course, software support. The huge mountain of x86 software for all industrial and personal purposes that just installs and runs. Try MIPS or ARM and your system ends up more expensive due to rarer chipsets, limits your software choices (gimme a selection of open source media players, for instance), and gives you no benefits over a VIA C3 minibox.

No, I wouldn't put x86 in a PDA just yet. Maybe if C3 gets good SSE2 or if Transmeta creates something worth suffiently big mass production. (Off topic: Where is the Crusoe that runs multiple different OS instances directly on the "Code Smurphing" hardware emulator, without any separate emulator or virtualiser software? I'd pay handsomely for such a system.)

Re:Power architecture does well

[Anonymous Coward]

Where is the Crusoe that runs multiple different OS instances directly on the "Code Smurphing" hardware emulator, without any separate emulator or virtualiser software?

Amen, bro! (Altho it would be Efficeon now)

Linus, go back there and make them geeky again.

Re:Power architecture does well

[Epistax]

Well the thing is any system could be at any position on the list. These aren't the computer makers competing, it's the computer buyers. You want the fastest computer in the world? Fine, pick a processor. To say any particular vendor has any particular spot is to say that vendor was chosen, and really nothing more. I'm sure anyone could beat the top spot by 20% if they received the order.

Re:Power architecture does well

[kai.chan]

5 of the 10 top machines use the Power archtecture, either the Power4 or PPC family.

Coincidentally, my top machine use Microsoft Windows and it has a record of AlwaysFlops.

Re:Power architecture does well

[ipjohnson]

Good post. On a side note I have a brother named brett johnson. Is that your real name?

Re:Power architecture does well

[1qa2ws3ed]

"[it] limits your software choices (gimme a selection of open source media players, for instance)"

mmm, is this a joke? nearly all the best media players out there are open source.

Re:Power architecture does well

[evilviper]

The same was true of the Alpha architecture, up until just a few short years ago.

Re:Power architecture does well

[Chaset]

Actually, I read some time ago that the AMD-K5 had a feature where you can do just that. In a certain mode, it will run binaries compiled for its RISC-like core directly without going through the x86->internal code translator.

"Nobody" used this feature, so it was dropped. I forgot where I read this, though. Probably the "great uP of the past and present" page. That's a good read for hardware geeks.

EARTH TO MAC ZEALOTS:

[Anonymous Coward]

System X is not a single system image machine. It will not work well for problems requiring low internode latency and high internode bandwidth in comparison to a real supercomputer.

Christ, clusters are not the end all and be all of high performance computing systems.

Re:EARTH TO MAC ZEALOTS:

[Anonymous Coward]

Don't discount infiniband... it has one of the lowest internode latencies available (quadrics has a lower latency, but lower max bandwidth). So if an OS that supported multi-machine spanning was used in a senario like VA tech, you'd be dead wrong. The hardware is there, just not the software, just not yet.

Re:EARTH TO MAC ZEALOTS:

[Anonymous Coward]

Infiniband is no replacement for the kind of performance you get out of a real internal bus.. i.e. what Cray uses. Those are your real supercomputers and your silly Apple clusters, while nice and very economical for many situations, will fail badly for many others. SO you're dead wrong man.

Re:EARTH TO MAC ZEALOTS:

[kc8apf]

While you are correct that clusters are not the ultimate solution for high performance computing, single-image computers are not a great solution either. They require specific optimizations to be done for the particular system and do not lend for easy system upgrades.

Re:EARTH TO MAC ZEALOTS:

[Anonymous Coward]

I don't know how you can say that They aren't easy to do upgrades. The 128 processor single image system I maintain is as easy as upgrading a workstation. And the applications generally don't require any optimization just to get running. Most cases it is as easy as running on a workstation, just alot bigger.

Re:EARTH TO MAC ZEALOTS:

[ceeam]

What machine does have the biggest flops per thread I wonder?

Re:EARTH TO MAC ZEALOTS:

[Anonymous Coward]

What? You mean benchmarks aren't universally applicable indicators of performance in every possible case?

MAIS NON!!!!

Re:EARTH TO MAC ZEALOTS:

[kevingolding2001]

Yeah, but imagine a beowulf clus... Ummmm.. sorry, what was the question again?

Re:EARTH TO MAC ZEALOTS:

[Anonymous Coward]

Interesting but not insightful. InfiniBand has one of the lowest latencies of external cabling solutions. But it simply isn't on the same league (performance or price wise) as the hardwiring you find closest to the nodes on those top three systems (IBM Blue Gene/L, SGI Columbia, NEC Earth Simulator).

In short, the hardware isn't there.

This is not discounting InfiniBand. This is just not pretending that Mellanox or Quadrics or Myrinet can do more than they really can achieve.

Google up those Earth Simulator "Processing Node" cabinets. Mmmm, funky. Wouldn't mind one (empty) cabinet as a wardrobe.

Re:EARTH TO MAC ZEALOTS:

[UnknowingFool]

System X is not a single system image machine. It will not work well for problems requiring low internode latency and high internode bandwidth in comparison to a real supercomputer.

Christ, clusters are not the end all and be all of high performance computing systems.

I guess I don't get the arugment because many of the other entries are clusters and not not single image either. Of course, given OS and architecture differences, all the supercomputers may perform differently in real world applications than the benchmark tests. The point is that System X was built using off-the-shelf components at a fraction of the cost of comparable systems. The entry right above System X is the US Army cluster of Apples (Mach 5). Obviously Apple is starting to be considered for high end computing.

In your face, Amiga! In your face! :)

Re:EARTH TO MAC ZEALOTS:

[fatphil]

And it's not the fastest system in an academic institution either. The technical university of Catalonia's BSC is in the #4 slot.

Who has coffee?

[Saeed al-Sahaf]

It's nice that Off The Shelf boxes like Apple and Intel can make a super computer cluster. When do the stories stop? We know that if you put enough PCs together, you get a very powerful machine. What we should be looking at is cutting edge technology in specialized CPUs. Give me 10,000 vanilla boxes and some good custom software, but give me a cutting edge CPU designed for super computing, that's science. We already know that it is possible to fill a fucking building with Pentiums, or better 68000s.

Re:Who has coffee?

[Anonymous Coward]

WTF? mod fucking parent up

Re:Who has coffee?

[Zork the Almighty]

Bah. At the end of the day, scientists actually want to compile a program and run it, without having to worry about some wacky experimental CPU architecture. Granted, some problems require innovative designs, but if the majority of your problems just require a room full of pentiums, well, whatever.

Re:Who has coffee?

[Westacular]

A quick look at the list just goes to support how in today's market, FLOPS are essentially free while the real concerns for not-embarassingly-parallel problems are memory latency, bandwidth, and energy.

Re:Who has coffee?

[Anonymous Coward]

... give me a cutting edge CPU designed for super computing, that's science. We already know that it is possible to fill a fucking building with Pentiums...

Thank you. I'm just sad that I had to scroll through several hundred comments to find this injection of sanity.

super computations?

[Doc Ruby]

Earth Simulator spent a while at the top of the list - that's a lot of TFLOPs under the curve - a lot of seconds. What did it accomplish while it was king of the hill? How much Earth was simulated?

Re:super computations?

[Cobalt Jacket]

There are prior systems which spent a lot longer there.

Re:super computations?

[Doc Ruby]

It's area under the curve: how much longer, but how much slower? And uptime / CPU time is the "T" axis, not "wall time"; if Earth Simulator has better utilization, it wins again with better "T%" of wall time.

Anyway, that just clarifies the competition. If others have spent more FLOPs as the "superest", what have they done with their time?

Re:super computations?

[Zork the Almighty]

They simulated the entire earth up until the creation of slashdot. Then everything suddenly ceased.

Re:super computations?

[relaxmax]

I knew its the Matrix...

-- rxMx --

Re:super computations?

[roine]

How much Earth was simulated?

The closest they got was 41.9998645234, which is pretty good in my book.

VA Tech Supercomputer

[dretay]

I was down at Virginia Tech last year when I was looking at colleges. They would not let anyone near that computer. Even the guy who was giving the tour was complaining about the limited amount access Tech students were given. The main reason he cited was that the companies developing the supercomputer had technology that they didn't want people who had not signed NDA's to see. Anyway, the point was that while the computer may be owned by the university, students aren't even allowed to see what $5 million of their tuition bought.

Re:VA Tech Supercomputer

[Anonymous Coward]

if virginia is anthing like my university (likely), then that money was not from tuition - it was from profit from spioff companies and research.

Who cares that you went there? Just because you didn't get to look at a bunch of Apple hardware doesn't mean we'll (a) listen to you, or (b) give a flying fuck about it.

Go home to your mother, dick.

Re:VA Tech Supercomputer

[dretay]

The point wasn't that I didn't get to look at the hardware, but that no one was able to do it. The only reason I said I had been there was to back up my position with actual facts and experience. And seriously, if you are gonna do a hostile post, at least use your account.

Re:VA Tech Supercomputer

[ananke]

Unmm, that cluster is in a machine room of a building designed to house a lot of IT/network resources. Of course they don't let anybody just stroll in. It's not like every IT building you've been into was designed as a zoo.

Re:VA Tech Supercomputer

[jdog1016]

It's not even technically on campus. Its at the nearby Virginia Tech Corporate Research Facility. And in any case, you can arrange a tour if you want.

Re:VA Tech Supercomputer

[dretay]

Cool,
I guess a lot has changed since I was down there last year. Thanks fo the link

Re:VA Tech Supercomputer

[monkeyboy87]

Tech has a long tradition of this. We weren't allowed to come near the IBM 3090 (a near super computer in the late '80's) when I was an undergrad there either.

it was housed in the CRC about 1 mile off campus in those days. Probably freed up the room for the cluster when they decomissioned the old 3090 behemoth.

Re:VA Tech Supercomputer

[mikael]

It's the same with other supercomputer sites. I was at a visit to Edinburgh University, and nobody was allowed to "see" the Connection Machine that they had at EPCC. The entire system was surrounded by eight foot high walls which were painted black. The irony was that Scientific American had a special edition that month which covered parallel processing, and had full page photographs of the CPU's, circuit boards and system design. EPCC claimed this was a requirement by the NSA.

Re:VA Tech Supercomputer

[flabbergast]

How is this different from any other school? At Wisconsin undergrads rarely if ever saw the clusters they had, and these weren't particular prestigious either. For instance, I never saw the WAIL lab, and I didn't know anyone else who had either. So, I don't know why you would expect to be able to view VTech's cluster.

Re:VA Tech Supercomputer

[Chiron Taltos]

$5 million of their tuition probably landed several hundred million dollars worth of research grants. Paying tuition does not entitle you to everything a university does with the money. And that's even if the assumption is correct that it was tuition money in the first place.

Can anyone shed light on where the $5 million actually came from?

Re:VA Tech Supercomputer

[doombob]

University of Nebraska-Lincoln's PrairieFire is a lot more open to public viewing. When they initially opened up, there was even an open house and tours given of the facility. The location was only about 3-4 blocks from campus. They even let a student-run engineering publication write a story about it and get up right next to it and everything. I know it wasn't super awesome or anything, but it was impressive for Nebraska.

Re:VA Tech Supercomputer

[jschottm]

The CRC is technically on campus. Half of RB XIV is CNS and half is IS&C. I have no idea what the first poster was talking about with NDAs - I walk (literally) through the cluster from time to time and I've never been NDAed regarding it.

Re:VA Tech Supercomputer

[Anonymous Coward]

The CRC is technically on campus. Half of RB XIV is CNS and half is IS&C. I have no idea what the first poster was talking about with NDAs - I walk (literally) through the cluster from time to time and I've never been NDAed regarding it.

If Virgina Tech is anything like my school, then walking tours around campus are give by undergrads being paid like $6 or $7. Sounds to me like the NDA thing was a polite way of saying no, not now, or no, lets change the subject because I don't know the answers to your impending questions.

Thanks for that post

[commodoresloat]

That was a close one... I was starting to worry that Apple might be dying again.

Erm ...

[phoxix]

What exactly did the Big Mac do anyways ?

I can assume it was put to some sort of use. But I honestly get the feeling it was more to have fun, and look cool (which means more bling bling from sponsors, alumni, etc)

Sunny Dubey

Re:Erm ...

[Zork the Almighty]

Virginia Tech: We got da BLING BLING! Yeah, baby!

Re:Erm ...

[Anonymous Coward]

You have been found guilty of using the term "Bling Bling". You must leave /. now, handing in your Geek bagde at the door.

Good day luser.

Re:Erm ...

[BJH]

Bagdes? We don't need no steenking bagdes!

Re:Erm ...

[Anonymous Coward]

Apparently, it's designed to churn out apple propoganda.

hehehe, IBM knows how to play

[BortQ]

IIRC then IBM just came out with their entry very recently. Perhaps they know how to play supercomputer sniping. It's easy to learn on ebay.

Funny MIPS

[RealProgrammer]

These numbers seem surreal, like thinking about Monopoly money. I'm sitting here at my old PII-300, analyzing the cost/power ratio of machines costing a mere $6M, or as much as $350M. This one cost, uh, nothing.

On any one of those systems, you could emulate a Beowulf cluster of this one, and still have time to play Thermonuclear War.

Re:Funny MIPS

[BlacKat]

I thought it was Global Thermonuclear War. :)

"Would you like to play a game?"

Re:Funny MIPS

[ockegheim]

If anyone wants an excellent flops/dollar deal they can have my old Mac SE/30 for free...

Re:Funny MIPS

[evilviper]

This one cost, uh, nothing.

Not unless you get free electricity.

cluster operating system

[MrMartini]

I get a kick out of the fact that System X runs Mac OS X.

Only with Mac OS X can you get the combination of commercial software (such as Microsoft Office and Adobe Photoshop), user friendliness, no known viruses, best available security, and stability/scalability suitable for world-class superclusters.

Re:cluster operating system

[dretay]

I think you may be mistaken about the viruses

Slashdot has previously covered a http://apple.slashdot.org/article.pl?sid=04/04/08/ 1922237&tid=172&tid=3 trojan and a http://apple.slashdot.org/article.pl?sid=04/10/23/ 0229241&tid=179&tid=172&tid=1 opener malware.

But you're right, MUCH less than windows. I've been using my MAC for about one year now, and I have gotten 3 people in my lab to switch over.

Re:cluster operating system

[Anonymous Coward]

Unless you're talking about network cards, 'Mac' is not an acronym.

Re:cluster operating system

[dretay]

Yah its not like Apple dominates the MP3 market or is the most widely-distributed UNIX-based operating system http://www.apple.com/macosx/features/unix/or is reccomended as the safest OS http://it.slashdot.org/article.pl?sid=04/11/02/172 2237&tid=172&tid=179&tid=190

Seriously, everyone seems determined to put Apple down when they are in fact producing quality products.

Re:cluster operating system

[Anonymous Coward]

lol, that's so funny, really. normaly I get so tired of crapy trolls. but your rejoinder was quite amusing! now if only slashdot posters would all be so cool...

Re:cluster operating system

[Anonymous Coward]

It whitens your teeth and makes your farts smell of roses, too.

Re:cluster operating system

[dretay]

Umm, according to wikipedia http://en.wikipedia.org/wiki/Apple_Macintosh it is a nickname, and I never said it was an acronym.

Re:cluster operating system

[Polarix]

The point is you're not allowed to capitalize it. Apparantly.

Re:cluster operating system

[Zork the Almighty]

Hey I own one of those Apple quality products, specifically an iBook. Yes it is great, but give me a break. Like some jackass is going to run Office or Photoshop on a cluster. Jeez.

Re:cluster operating system

[Zork the Almighty]

Hey, if I'm going to troll or flamebait someone - at least I'll do it with my real name. Oh, wait...

Re:cluster operating system

[MrMartini]

Hey I own one of those Apple quality products, specifically an iBook. Yes it is great, but give me a break. Like some jackass is going to run Office or Photoshop on a cluster. Jeez.

That remark was a fairly typical straw man argument.

http://en.wikipedia.org/wiki/Straw_man

Re:cluster operating system

[MrMartini]

Neither of those was a virus.

The first one was incapable of spreading itself, and ended up being an embarassment for the software company who brought it up in the first place; it seemed that they were basically trying to scare Mac OS X users into buying their antivirus software.

The opener malware required legitimate administrator access to install; It was clearly not a virus.

The number of known Mac OS X viruses stands at zero.

Re:cluster operating system

[Anonymous Coward]

It's not a strawman, knob. Those apps were mentioned in the original post.

Re:cluster operating system

[MrMartini]

It's not a strawman, knob. Those apps were mentioned in the original post.

This is really funny - obviously you need to click the link I provided and read about straw man arguments.

In my original post, I took the position that Mac OS X is strong in a suprising number of ways; it has top security, no known viruses, is scalable and stable enough for superclusters, yet it is also very easy to use and has commercial software available for it.

Zork Almighty then refuted the argument that using commercial software on a supercluster was useful, but I never made that argument; he was taking a contrary position to an argument that I did not make, which is a common form of straw man argument.

You have made yet another straw man argument, since you are taking a contrary position to the argument that those apps were not mentioned in the original post, but I didn't make that argument either. So you too have made a straw man argument. Good job, now go read up on straw man arguments using the link I provided!

Re:cluster operating system

[Anonymous Coward]

You're also not allowed to spell 'apparently' like that.

Re:cluster operating system

[lauterm]

You should be a politician.

Re:cluster operating system

[Anonymous Coward]

Thank god for that upgrade! My mouth was getting tired, and it's going to be a long wait for the G5 powerbook.

Re:cluster operating system

[Oddly_Drac]

"Only with Mac OS X can you get the combination of commercial software (such as Microsoft Office and Adobe Photoshop)"

I'll pass on the GUI and stick with FreeBSD, thanks.

Re:cluster operating system

[fitten]

In my original post, I took the position that Mac OS X is strong in a suprising number of ways;

it has top security,

The "top security" designation is arguable. While there have been editorials espousing the security of Mac OS X, there are some who claim that the security is dubious because of a number of factors. The easiest and first argument is that the marketshare of Mac OS X is small enough to make virus/worm writing unproductive to the aims of virus/worm writers.

no known viruses,

This is simply false, as linked and stated by others.

is scalable and stable enough for superclusters,

OK, but this is sort of a weak argument. A supercluster is really just a network of individual nodes. Any OS that is stable and can use a network reasonably efficiently fits this bill. Even Windows, with the right scheduling and communications software, can be used in a supercluster (see Cornell).

yet it is also very easy to use

This is also weak because "easy to use" doesn't normally come into play on a cluster. No user sits at the console of any of these machines, typically. Perhaps if one were to be configuring a new node on a workbench before actually inserting it into the cluster, but typically these things are managed remotely and/or by software designed to manage the cluster as a whole. Usually, a scheduler (perhaps a batch scheduler) controls which applications are run on which nodes and many (if not most) supercomputing type applications are not interactive with a user. Once the initial parameters are provided, the parallel application will be run and it will compute until some predetermined end condition is met. Unless the application is a graphical visualization done in realtime, no user input is required (or desired). Many applications output data into files and those files are what are used by other visualization tools after the application has finished. I have seen a few visualization applications that would render data that was periodically output from a parallel application, but it was a seperate task not included in the application itself and the visualizations were done on timestep data or the like.

and has commercial software available for it.

As with the previous argument, the types of commercial software that were mentioned would rarely, if ever, used on any node in the cluster. Most often, a node in a well run cluster is controlled by a scheduler (perhaps a batch scheduler) and the processes that run on it are written by the scientists/developers who are doing research. The commercial applications that might be run on a cluster are schedulers and possibly inter-process communication libraries, not Word or an MP3 player or something like that.

Re:cluster operating system

[MrMartini]

fitten:

I mentioned that it has top security. Here's the slashdot post that caused me to think so:
http://it.slashdot.org/article.pl?sid=04/11/02/172 2237&tid=172&tid=179&tid=190

You've claimed that there are known viruses for Mac OS X. Provide a single example.

Look at the OSs used for superclusters - generally either Linux/UNIX/Mac OS X. I just looked at the Cornel Windows supercluster at top500.org, and I see that it has 640 2.4Ghz Intel Xeon CPUs, achieving 1503 Rmax. System X has 2200 2.3Ghz IBM G5 CPUs, achieving 12250 Rmax. Since System X reaches 8.15 times the performace with 3.44 times the number of CPUs, obviously something is holding those Intel CPUs back, unless you think that 2.3Ghz G5 CPUs are 2.37 times faster than 2.4Ghz Intel Xeons.

I mentioned that Mac OS X is easy to use, and you made a straw man argument that ease of use isn't important for cluster nodes - obviously I never claimed that it was. Read about straw man arguments - good to know about!

I mentioned that the availability of commercial software for Mac OS X is a strength, and you made a straw man argument that commercial software is not important for cluster nodes; you are taking a position against something that I did not say, for the second time.

Re:cluster operating system

[fitten]

I mentioned that Mac OS X is easy to use, and you made a straw man argument that ease of use isn't important for cluster nodes - obviously I never claimed that it was. Read about straw man arguments - good to know about!

Yes... Straw man arguments are using points that are irrelevant to the topic at hand.... Like "ease of use"... which is unimportant to a cluster. You might as well claim that 32 bit color is a wonderful feature that these clusters have because they use Mac OS X... completely irrelevant because these machines won't have graphic displays.

I mentioned that the availability of commercial software for Mac OS X is a strength, and you made a straw man argument that commercial software is not important for cluster nodes; you are taking a position against something that I did not say, for the second time.


Again, your Straw Man argument that a wealth of commecial software is available for Mac OS X that runs on these clusters is another example of your using some point that is irrelevant to the discussion of clusters because the clusters won't be using these commercial applications.

If you really want to Straw Man it out to the limit by listing irrelevant "features" of Mac OS X as pertaining to the topic at hand, namely cluster environments, we could list:

- The color of the case is pleasing to the eye.
- The style of the case gives you the warm fuzzies.
- You like the fact that Apple has six characters in its name, whicl Intel only has five.
- The processor calculates predominantly in big endian mode.
- The distance between the case backing and the motherboard is approximately 1/2".
- You saw an elephant at the zoo the other day but you didn't feed it peanuts, but you later looked up "elephants" on Google using your Mac.
- The file size of your favorite editor on the Mac is exactly number of bytes.
- etc.

Then, when someone says that none of these matter for the subject at hand, you can call those Straw Man arguments as well.

Since you seem to have recently learned what "Straw Man" is and enjoy using it to the limit for anything anyone says to refute what you say, you should also realize that you are perfectly capable of initiating Straw Man arguments as well.

Re:cluster operating system

[MrMartini]

Yes... Straw man arguments are using points that are irrelevant to the topic at hand

Wrong, that is not what a straw man argument is. Go read about them. A straw man argument is misrepresenting someone's position, refuting that position (that you've essentially fabricated), and then concluding that you've refuted their original argument. You continue to have an argument with a misrepresentation of my original statement (which was regarding the versatility of Mac OS X), and this indicates that you are severely irony-challenged. It was a simple point I was making - how could you fail to understand it?

what if...

[Suburbanpride]

They made a beowulf cluster of all the top 10?

But..

[Anonymous Coward]

It doesn't matter. The brushed aluminum case makes it at LEAST 3 times faster.

Earth Simulator has liberal bias

[Anonymous Coward]

The Earth Simulator computed that global warming will cause major climate change in the next 50 years.

Clearly it suffers from liberal bias.

Re:Earth Simulator has liberal bias

[Anonymous Coward]

"The Earth Simulator computed that global warming will cause major climate change in the next 50 years."

...and it consumed so much power it skewed the real result nearer to 42.

Re:Earth Simulator has liberal bias

[Anonymous Coward]

It also computed that 53% of that temperature increase was caused by running more climate simulations on the Earth Simulator...

Re:Earth Simulator has liberal bias

[evilviper]

No, it's clearly just infering something based on a very limited set of facts, and limited knowledge of the planet.

It would have a liberal bias if it said SUVs, and civil rights violations were to blame.

It would be conservative bias if it said stem-cell research, and gay marriages were to blame.

Excellent comparison?

[FiReaNGeL]

On the "excellent comparison" page... wow. Excellent? Worst use of tables EVER.

Must...not...stare...at...ugly... colors!

Honestly, can someone actually read anything from these? Ugh

Re:Excellent comparison?

[irc.goatse.cx troll]

View-> Page styles-> No style.

Or alternativly install the EditCSS (or webdevbar) plugin and just add a rule of * {background-color: white;color:black;}

Sounds right, actually

[MooseByte]

"I'm not sure if the costs can be accurately compared in this way. [ ... ] Otherwise you can start comparing stuff like SETI which I'm sure is the world's cheapest supercomputer"

Actually that sounds like a perfectly valid comparison, SETI included. In bang for the buck SETI deserves to win hands-down in that scenario, and fairly. System X deserves its place as well.

Re:Sounds right, actually

[Anonymous Coward]

Except that not all supercomputers are created equally. Sure, it may be cheaper, but that doesn't mean it can solve the same problems as the custom built ones.

Bandwidth and latency issues play a role in what a supercomputer can accomplish with its theoretical FLOPS rating. Not all problems can be split up to work on a SETI style cluster.

Re:Sounds right, actually

[Anonymous Coward]

Depending on application, System X's 'bang for buck' could actually be an unmitigated loss compared to IBM's offering. Any problem that isn't embarassingly parallel, for instance, would destroy System X in a value comparison with even crazy-expensive Cray machines. On such a problem even SETI, were it forced to utilize all nodes, would be orders of magnitude slower than a dual opteron workstation (if not, it might be faster if any single node (or tight cluster of nodes) is faster than the aforementioned workstation and all work is delegated to it, idling the rest).

System X only has respectable bang for buck if you limit the application domain to problems System X was specifically designed to solve effectively, which, incidentally, aren't the same set of problems that BlueGene/L can solve effectively. So no, it is not a valid comparison. It's like saying that the space shuttle transport is less cost-effective than a fleet of pickup trucks with an equivalent combined cargo weight capacity.

Re:Sounds right, actually

[Anonymous Coward]

SETI doesn't run Linpack. Sorry.

Hey, who'll start a Linpack@Home client project on SourceForge? Peferably somebody who has lead a somehow similar project before. We have enough time to develop the client and spread it, in the nearly six months before the final run for the next Top500 list :-)

Re:Sounds right, actually

[timts]

a PC cluster will definitely the cheapest solution, think about this, my p4 2.8G HT dell 400SC from dell's clearance for 400SC is $124+tax AR, if you get 1000 of these, no mac or other vendor can beat the performance/price ratio, btw, mac's price is always pretty high comparing to other vendors, any way, and performance is far lower than its claim.

we did some tests before, mac's website posted comparision between its G5 and P4 is merely cheating, (optimized G5 score versus way slower than unoptimized p4 score).

Re:Sounds right, actually

[Lars T.]

we did some tests before, mac's website posted comparision between its G5 and P4 is merely cheating, (optimized G5 score versus way slower than unoptimized p4 score).

Doing what exactly? Or is it a secret?

Re:Sounds right, actually

[the_2nd_coming]

umm, the same can be said about all these top of the list machines.

you give them a problem that is not in their problem domain and they will be embarrassingly slow. try playing doom 3 on Blue Gene.

Re:Sounds right, actually

[Anonymous Coward]

Uh, perhaps you missed that column on the chart marked "interconnect." For $124/box you are not getting infiniband.

I'm also not sure where you are getting a 400SC (PowerEdge Server?) for $124/box; when I search Dell's website, the lowest I see at the Dell Outlet is $744 for system identifier E6R3B791.

They don't list network card, or even mention it on the specs, which should give you some hint, but what I can find says "embedded Gigabit NIC (Intel)." So for 6x your quoted price, you are getting some midrange interconnect.

Re:Sounds right, actually

[Anonymous Coward]

Exactly. That is why cost comparisons between the various entries are largely pointless.

Re:Sounds right, actually

[AviLazar]

Except SETI power is not consistant. It is hinged on being connected to other computers that belong to remote users. The SETI program ceases to function once the computer is removed from idle state or turned off - while these other supercomputers are constant (barring things like malfunctions or breaks). Plus - it would be hard to compare computers located all over the internet to a computer sitting in one location - which might be criteria for these tests.

Re:Sounds right, actually

[timts]

doing the exact benchmark Mac claimed on their false comparison between Mac cpu and p4.

Re:Sounds right, actually

[timts]

of course that's just for one node, did I say it's a cluster? but each Mac node will definitely cost a hell lot more.
well, mac fans cant be mac fans if they check the price, right?
I said cleary in my previous post that's dell's clerance for 400sc, before they get out 420sc. yeah, refurb stuff is not as cheap, amazing? that's called a deal... only for PC users. :D

Re:Sounds right, actually

[Lars T.]

Who is this "Mac" person you keep talking about?

CPU benchmarks

[3770]

You can compare CPU benchmarks here.

AMD is beating the crap out of Intel.

Re:CPU benchmarks

[Anonymous Coward]

This is the best specint AMD has to offer (yes, the fastest operon is slower):
AMD Athlon (TM) 64 FX-55 1 core, 1 chip, 1 core/chip 1750 1854

Here is the best intel:
3.4 GHz, Pentium 4 Proce 1 core, 1 chip, 1 core/chip with HT Technology enabled 1667 1705

Here are the best specfp numbers for AMD:
AMD Athlon (TM) 64 FX-55 1 core, 1 chip, 1 core/chip 1741 1782

Which just edges out the best Pentium
3.6GHz Xeon) 1 core, 1 chip, 1 core/chip 1700 1721

But gets beaten by a low end I2
1300 MHz, Itanium 2 1 1808 1808

Also by a Fujitsu SPARC
PRIMEPOWER900 (1890MHz) 1 core, 1 chip, 1 core/chip 1510 1803

Gets trounced by the second-best I2 available
1500 MHz, Itanium 2 1 2161 2161

And gets demolished by the top end POWER5
1900 MHz, 1 CPU 1 core, 1 chip, 2 cores/chip (SMT off) 2576 2702

Re:CPU benchmarks

[BJH]

Why don't you try comparing the prices of those CPUs too, eh?

Re:CPU benchmarks

[Anonymous Coward]

Ahh the mating call of the loser. This thread is about ass kicking, not money.

So how about this: AMD slightly edges out Intel in integer performance but gets demolished in floating point performance by an old Itanium 2.

Re:CPU benchmarks

[Rufus211]

Usefull site for getting usable spec numbers is ace's SPECmine. Try SPECint2000 or SPECfp2000 and check "CPU MHz". The opteron's fare quite nicely.

move towards clusters...

[pagal_paanda]

After looking at the G5 cluster, would we see a shift in supercomputers made out of clusters? Anyone for Athlon 64 based clusters?

Why does slashdot keep posting these stories?

[The Pim]

Is there some fantasy supercomping league I don't know about?

Re:Why does slashdot keep posting these stories?

[Anonymous Coward]

No, there's a fantasy Mac league.

Re:Why does slashdot keep posting these stories?

[Anonymous Coward]

No, the story just gives them an excuse to plug Apple products. If it wasn't for that, this stuff would hardly get any mention at all.

Just another PR and propaganda piece, just like their write-ups saying the ipod being better whenever mentioning other non-apple mp3 players.

Re:Why does slashdot keep posting these stories?

[jdog1016]

Because it interests the readership?

Re:Why does slashdot keep posting these stories?

[Anonymous Coward]

I am at SC04, there are a lot of people here. And most of them probably don't belong to a fantasy football league. This is the next best thing.

I'll trade a SGI for 2 crays.

Re:Why does slashdot keep posting these stories?

[dfj225]

Yes, the League of Extrodinary GigaFlopers.

Re:Why does slashdot keep posting these stories?

[Mostly a lurker]

You insensitive clod. This kind of review is essental to much of /.'s readership in deciding on their next computer purchase!

power costs?

[CaptainPinko]

adminstration and maintenance similar perhaps... but what about power?a few watts per core adds to a lot more heat PLUS the cost of cooling. i think it would be interesting if they printed a FLOP/$ per annum for each of the top 500. the cost of acquisition being spread evenly over the lifetime of the cluster.

Re:power costs?

[daveschroeder]

That's easy...the Xserve G5s consume a LOT less power (and therefore generate less heat, resulting in lower cooling costs) than any competitive products (Xeon, Itanium2, Opteron)...and this was true even when they were using the 970 (as opposed to the 970fx they are using now).

Several of the researchers at Virginia Tech have referred to this in various news stories numerous times - one estimate was over two times less power than comparable systems.

Re:power costs?

[10Ghz]

What makes you think that 970 requires less power/cooling that (for example) Opteron does? I mean, 2.5Ghz G5 needs frigging liquid-cooling, whereas even the fastest A64/Opterons manage just fine with regural heatsinks and fans.

Re:power costs?

[Anonymous Coward]

I don't beleive your estimates of the power usage are quite correct. A 2.5Ghz G5 (90nm) has been shown to have a peak power in excess of 80W, which puts it in the same ball park as Xeon, Opteron, and Itanium, althoughb it may be a bit lower.

Re:power costs?

[Anonymous Coward]

Your facts and measurements have no place in Apple Land. Let's get some happy feelings and warm fuzzies going, people.

Re:power costs?

[Anonymous Coward]

That's not true.

Opterons heat dissipation is ~56W for a standard chip (40ish on the sub-2Ghz models).

The standard G5 is in the same ball park (51W at 1.8Ghz), although the fastest G5s run significantly hotter and all of them compare poorly at a performance/heat ratio.

However XServeG5 systems don't use standard G5s. They use the newer .09 micron version, which dissipated a very low 24.5W at 2Ghz.

But then, AMD has low-power Opteron models for rack mounts as well. And they eat up even less power than those G5s, 10-20W (again with a more favorable heat/performance ratio).

Opterons are also larger chips, with a notably larger surface area with which to disperse the heat they generate.

Furthermore, Opterons don't require a north bridge chip (it's built-in)... so any heat that the G5's seperate logic chip(s) dissipate should be factored in as well.

It may seem a small point, but when you have hundreds of them crammed next to each other, anything that makes managing their heat easier helps.

how much?

[interactive_civilian]

blockquoth Doc Ruby

How much Earth was simulated?

All of it...

The Matrix has you, Neo.

;)

Moderator Hypocrisy

[Anonymous Coward]

There are maybe.. two posts (counting this one) about the pitiful inter-node performance (when compared to a real supercomputer like a Cray) and they get knocked to Redundant/Troll/Flamebait/etc. almost immediately, but when some Mac fans come by and salivate over the VT cluster, they almost all get modded up without really adding anything to the discussion.

Don't get carried away with LINPACK flops/cost!

[Richard Mills]

You raise good points, and the team at Virginia Tech did do something remarkable. That said, cost per flop of the LINPACK benchmark is interesting but not particularly meaningful. For instance:

"Another example is PNNL's 1936 processor Itanium2 cluster: 3.5Tflops less performance than System X, for $25 million"

What is not captured by the LINPACK scores is that PNNL's machine will absolutely spank the BigMac cluster at what the PNNL machine is intended for: running computational chemistry codes such as NWChem. A lot of the cash for the PNNL machine went into large memories and fast I/O that simply does not show up in the LINPACK benchmark. Furthermore, there are a lot of very high-profile scientific publications that have come out of the computational chemistry abilities of the PNNL machine. That's something else extremely important that doesn't show up in the rankings.

There are a lot of similar examples, but the PNNL one is one that I know something about, so I chose it. Basically, I'm saying to not read too much into those cost comparisons. It really is comparing Apples to oranges... er, HPs in this case. =)

Re:Don't get carried away with LINPACK flops/cost!

[aalobode]

We need to ask several categories of questions instead of bashing Apple or the folks at Va Tech for their accomplishment, or for that matter one another.

First: What is the exact nature of the connection between Apple and the processor. Let's compare Apples to IBMs and G5s (which are not designed or produced by Apple but rather by IBM and Motorola) to Itaniums or Opterons or Fooballs.

Second: What OS does it use? If it is BSD or Mach underlying the GUI, then we need to compare that to Linux/UNIX/ Foo or whatever. Separate the UI from the parallel machine because that is not being used.

Third: What are the exact nature of the cost in terms not used in marketer-speech. Does hardware get donated or sold at a discount? Are salaries included and if so what kinds (college kids vs. someone who has a mortgage and a car and a few moderately inexpensive vices)? What about fixed costs (i.e. is depreciation on the building included, and what about profit margins?)?

My basic point is that early instances of these monsters are not intended for profit or public consumption. They are somewhere just past the proof-of-concept stage. The "mine-is-bigger-than-yours" comparisons only go a short distance. After that, one has to see the true contribution to the theory and practice of parallel computing. Let's hear from the
parallel computing community.

All the best to all of the best!

Did anyone see...

[lukateake]

... the No. 1 entry under the "Not Ready In Time" header? Oak Ridge is planning something in 2008 that has an estimated R-Max of 10,500. (And, no, that's not a typo!) It's a little short on details, unfortunately.

Surely someone here knows a little more. Details? Is it for real?

Fair to compare cost?

[jmcmunn]

I can't get the cost comparison link to work for one reason or another. But I was wondering if they were comparing current day cost, or the cost when the machine was built?

I mean, the cost of processors has gone down significantly in the past several months, so a machine built a year ago with the same speed processors as a machine built today would cost much more.

Not to mention that some of the machines on the list are most likely second (or greater) renditions of an earlier super computer, only with more procs/Ram or whatever, so those could be cheaper just because less R&D money went into them...

Just a thought, I can't see the link myself so I thought I would ponder some thoughts at those who could.

What is with the Apple fan-boyism?

[Anonymous Coward]

Now Apple markets good computers. Tend to be on the expensive side, but they are usually high quality.

The Power970 is decent enough in itself. The opteron is more powerfull, but is also much more energy hungry. The Intel Itanium is nice but it's very expensive. etc etc

But what is this worship of Apple? It makes no sense.

Story 1: Earth simulator.. blah blah blah., but Mac cluster!

Story 2: SGI supercluster.... blah blahblah, But Mac cluster!

Story 3: Blue Gene cluster, 65000+ cpus... blah blah blah, but Mac cluster!

Realy? Who gives a fvck about the 7th place computer, and who gives a damn about cost analysis at this point? What about the Top5?

Did you know that Blue Gene is PowerPC?
Did you know that Linux now runs the majority of top super computers...

Did you know that Blue Gene proccessors only run a 700mhz??!!!

Did you know that #4 is 3564 Power970's running at 2.2 ghz? And that beats out 4000+ Intanium2's running at 1.7ghz?

This is a Geek site.. what about OSes?
By ranking:
1. Linux, 2. Linux, 3. Unix, 4. Linux, 5. Linux, 6. Unix, 7. OS X, 8. Linux, 9. Unix, 10. Linux (most powerfull x86 btw), 11. Unix, 12. Unix, 13. Linux, 14. ?, 15. Linux, 16. Linux, 17. Linux, 18. Linux, 19. Linux, 20. Unix.

Were is the most powerfull Windows computer? Well there is one cluster that is probably still on the top500. I dare you to find it, though. It's probably around #200 or #300, which is stil freaking fast.

Ok, So the big Mac is still #7. That's great, but there are 6 wonderfull computers that have all sorts of great technology that your completely ignoring because Apple wets your pants.

Did you know that Blue Gene will eventually have over 65,000 proccessors??

Re:What is with the Apple fan-boyism?

[Anonymous Coward]

"This is a Geek site.. what about OSes? By ranking: 1. Linux, 2. Linux, 3. Unix, 4. Linux, 5. Linux, 6. Unix, 7. OS X, 8. Linux, 9. Unix, 10. Linux (most powerfull x86 btw), 11. Unix, 12. Unix, 13. Linux, 14. ?, 15. Linux, 16. Linux, 17. Linux, 18. Linux, 19. Linux, 20. Unix." The real point is that the Linux machines cost more...

Re:What is with the Apple fan-boyism?

[MacDork]

Ok, So the big Mac is still #7. That's great, but there are 6 wonderfull computers that have all sorts of great technology that your completely ignoring because Apple wets your pants.

Fine, I'm willing to talk about number 6. ;-)

Re:What is with the Apple fan-boyism?

[Anonymous Coward]

Ok, So the big Mac is still #7.

So what are the top 6? The Whopper? Gordita Supreme?

Re:What is with the Apple fan-boyism?

[PingPongBoy]

Did you know? Did you know?

How the hell did I know? It was slashdotted!!

So you'd better be right

Re:What is with the Apple fan-boyism?

[Anonymous Coward]

It's not quite slashdotted. When I get a database error, I just try the links again and it usually works.

Re:What is with the Apple fan-boyism?

[Oddly_Drac]

"But what is this worship of Apple? It makes no sense."


BURN THE HERETIC!

Re:What is with the Apple fan-boyism?

[xenoandroid]

Apple worship because we all knew they were going to die and are overpriced and whatnot, then we see their products being used to make one of the most powerful clusters in the world for very cheep.

Apple worship because it's a smack in the face to those who still continue to bash Apple for reasons that no longer exist.

"OS 9 sucks!"
"We're on OS X now, and it's unix-like"
"oh...um...Well one button~"
"And yet I'm still more productive on it than my Windows box"
"Well um, I want linu-"
"You can install that too."
"Well, they're slow and-"
"I suggest you actually try using one before saying that."
"They're overpric-"
"Really? I didn't think $900 was that expensive for a mid-range machine."

We do the Apple worship thing just to fustrate the anti-mac crowd even more.

Re:What is with the Apple fan-boyism?

[xenoandroid]

frustrate...

Re:What is with the Apple fan-boyism?

[Mornelithe]

Where are the mid-range machines for $900? I looked at the online apple store, but the closest I could find are eMacs for $800, and iBooks for $1000. Neither one of those are really what I'd consider mid-range machines today.

Re:What is with the Apple fan-boyism?

[xenoandroid]

Those iBooks perform better than any PC laptop of the same price I've ever seen. I've never used an eMac personally but I'm sure there isn't that much of a difference.

What specs do you think mid range machines have?

Re:What is with the Apple fan-boyism?

[Mornelithe]

Well...

All of Apple's computers, except the really high end ones, come with 256 MB of RAM standard. That's low these days. 512 is pretty standard, and if I were building a computer, I'd put 1 GB in, because that only costs about $150. Upgrading from 256 MB to 1 GB costs $225 for a Mac, though. The video cards are also on the low end.

When I was pricing recently, I came up with a setup that goes something like this:

Athlon 64 2800+
1 GB RAM
200 GB hard drive
GeForce 6600

The price of that was around $900. It doesn't include a monitor, but I can buy a 17" CRT (which the eMac has) for a little over $100. So are you saying that the eMac with a G4 1.25 GHz, a Radeon 9200, and a 40 GB hard drive will outperform the above machine (both would have a gig of RAM for equivalent pricing)?

Looking at the Mac store, I'd call the iMac a mid-range machine, and it starts at $1,300. Upgrade the hard drive and memory to what I'd consider acceptable, and it ends up being like $1,600.

I can't really fight on the laptop front, though. I really have no desire for a laptop, so I'm not in touch with their pricing. I could do some research, if you wish, though.

Re:What is with the Apple fan-boyism?

[Anonymous Coward]

What you don't know is that they can't actually run HPL on the entire cluster because their hardware is not stable. When they run one job on the whole cluster, one node locks up solid (different node each time). That's the main reason they've never submitted a Top500 number.

This is with Gigabit Ethernet, not even a fancy interconnect like Myrinet or IB.

Only on Slashdot

[Anonymous Coward]

Only on Slashdot would the article be about the G5 cluster instead of about the new, faster machines. In 3 years, when the VT cluster drops to #400, will you still be posting "news" about how pretty and how adorable and how appley it is...?

Can we please restrict the Apple ads to the banner? Thank you.

Re:Just imagine

[BottleCup]

It seems like every time there is a new supercomputer in a Slashdot headline, the first few responses are about running Doom 3 or some other FPS. Those computers are better off used for among other things to create predictive complex simulations like global weather, seismic activity, the global economy etc ... etc. Using such a beast for a simple FPS would be such a waste.

Re:Just imagine

[The Master Control P]

Hook it up and I think that 'anonymous' may just be displaced as the top contributor...

It's still running

[Rui del-Negro]

> How much Earth was simulated?

Well, I've noticed a vew glitches (disappearing keys, poor AI in girlfriends, crazy presidents in some countries, etc.), but I'd say most of the Earth has been running reasonably well.

Re:It's still running

[Anonymous Coward]

Have you tried leaving the city yet?

Sla....

[kgbspy]

DB Error: connect failed

Maybe they could run the top500 website on it, using up some of those spare cpu cycles...

Re:Just imagine

[mlk]

Using such a beast for a simple FPS would be such a waste.
So true, we should use a complex, awe-inspiring game, which will push the limits of any machine. I suggest Nethack.

Damn!

[Anonymous Coward]

That is impressive. Earth Simulator got beat by 2! count 'em! two! supercomputers after standing supreme for a few year.

This just in...

[Intocabile]

Supercomputers become relatively slower as faster computers are built. PowerPC based supercomputers the exception to the rule, are no exception.

Not on the list

[leapis]

DB Error: connect failed

Apparently, the top 500 list is not actually hosted on one of the top 500 machines.

VT ongoing cost of ownership evaluation...

[MacDork]

200 pizzas a week. ;-)

Re:VT ongoing cost of ownership evaluation...

[LittleBigLui]

You USians and your funny units... how much is that in Libraries-of-Congress?

Re:VT ongoing cost of ownership evaluation...

[kevingolding2001]

Yes, but how many volkswagens would be needed to hold that many books?

Wait for the HL2 benchmarks before buying

[Anonymous Coward]

Yawn...you're better of waiting until HL2 comes out before running out and buying one of these. We've all been burned once buying hardware in anticipation of HL2...

Steve Jobs: "Where the heck is Apple MACH 5?!!!"

[Anonymous Coward]

Steve Jobs to COLSA

"Bill? Yea it's Steve at Apple, hey we need you guys to run the Linpack test and submit the results.

I know the ARMY is having fun with their 3132 processor LAN Unreal Tournament 2004 frag fest, but this is buisness!

YES I KNOW THE MAC IS A AWESOME GAMING MACHINE!"

"hello?"
take a pretty picture

Any idea on the OS that run these machines?

[kalki]

Why is there no parameter (of supercomputers) describing the OS?

It will be nice if the OS of the TOP 500 super computer are listed as well. Does any one point to the information that lists the OS of the Supercomputer?

I'm still hanging in

[serutan]

My computer is number 44,286,551 and I'm gunning hard for position 44,286,550.

Re:I'm still hanging in

[Anonymous Coward]

You're not going to get me! I just bought a RAM-upgrade.

Re:Just imagine

[Lifewish]

I was wondering: you reckon this thing would have enough processor power to run Longhorn when it comes out?

LINPACK is a terrible benchmark

[Anonymous Coward]

With all due respect to the TOP500 publicity machine, the LINPACK benchmark is a terrible measure of the usefulness of a computer. Very few algorithms scale as n squared in flops versus n in i/o. The entire TOP500 effort is seriously flawed because essentially no applications behave like the LINPACK benchmark. It is a shame that this single number is used in such a blatantly political fashion.

Re:LINPACK is a terrible benchmark

[99BottlesOfBeerInMyF]

no applications behave like the LINPACK benchmark

Actually, few applications behave like the Linpack benchmark. But if you want to do those things, like render video or compile genetic sequences the Linpack is just fine as a benchmark. It is certainly not the be-all end-all. I think the reason for it's popularity is that it is a single number that can be easily measured. No one is interested in watching a race where they calculate the height of each person, the time in which they finished, their mass, their top speed, and their wind resistance, and try to give an accurate count as to who did the most work in the least time. Most people just want to know who got to the finish first.

Obviously, if you have job you want to get done, you will need to look at all the options for accomplishing that goal. PR should not play a big factor in your decision. Unfortunately, that is not always the case. Prior to the Virginia tech cluster, most people would never consider deploying a cluster of Apple's as a possible solution. Maybe the Linpack is just PR, and not meaningful for the majority of cases, but to my mind, it was at least PR that accomplished something useful, and challanged a number of prejudices.

Software

[Mardak]

It's not just about how fast the systems perform in linpack. The machines should be calculating something useful, and if you feed it inefficient code, it'll be going nowhere fast.

Apple has created software development packages specifically designed for their G5's with optimized code for the 64bit architecture such as complex math functions.

So not only is Apple providing a cheaper and power efficient system for academic institutes, they make it easier for professors and assistants to create the software to run on those clusters.

inane modding

[Zork the Almighty]

Why do people always mod up bland and unoriginal humor. Honestly. I have to post shit like this from time to time so that my user page doesn't make me look like a troll. If you want to see something funny that I posted, read this guy's comment and then find my reply.

Re:Just imagine

[michaeldot]

No, for that you'll need the "Hell Simulator."

#4 is also academic

[az4 h0th]

Just a minor comment...

I work at UPC and there has been a lot of hype here for machine #4, which is (or is going to be) a >4500 PPC970s machine running linux (nice work, ibm). I disagree with the claim that the Virgina Tech cluster is the first academic supercomputer. As far as I'm concerned the Technical University of Catalonia (UPC) is also an academic institution.

Anyway. we now got europe's fastest supercomputer. That's what matters. ha! ;-)

Re:#4 is also academic

[Anonymous Coward]

Me caguen en el campus nord, que me echó en la fase selectiva...algun dia me pasare y le prenderé fuego a la capilla ;)

Un abrazo a los que aun seguis ahi...y a disfrutar del lomo con queso :P

Re:#4 is also academic

[danigiri]

Yeah, but as the Barcelona Supercomputing Center Site states: "the Spanish Government, the Catalan Government and the Technical University of Catalonia (UPC) signed an agreement to create a new supercomputer center that will host the MareNostrum supercomputer". (Emphasis mine)

I mean, though I would not like to demean UPC's involvement and fantastic job, from the BSC, IBM and the Top500 sites it is inferred that IBM built it, not the UPC, and not by all means a bunch of pizza-fed students and some enterprising professors (like VT). I also remember that it is IBM who built it from the press releases and TV "infomercials" at the time.

Have you looked at the cost figures!?! $6M (VT) v.s. $86M (MN)! Ridiculous! Yeah, MN is more powerful but a swindle, IMHO. If we had just grabbed the money and gone for VT-clones we would have 5 or 6 top 15 supercomputers and have several million left to keep the beasts running for years. It's my taxpayer money too.

Apart from very non-embarrassingly-parallel problems, the 5-6 XServe-cluster super-cluster would be much more efficient. In fact, the efficiency ratio figures hint that MN is a colossal waste of money. For 8 more in R-max we have spent $80M more!

I'm not an expert in supercomputing, but I am also versed in CS and this MN looks too an expensive toy compared to a whole bunch of VT-clones. I have yet to hear any compelling arguments to the contrary (except that we get a shiny #4), in fact we could be like #4 at $10M or so if gone the VT route.

Re:#4 is also academic

[Tiosman]

> Have you looked at the cost figures!?! $6M (VT) v.s. $86M (MN)! Ridiculous!

The price of this machine is $20M. I don't know where they got this figure of $86M, but they should not invent numbers if they don't know.

If you want to speak about something ridiculous, speak about the $6M price tag for the VT cluster. This machine is a big PR operation, that's it. the VT guys ask us to sell them an interconnect for $300 a node, all included. This was below production cost, we refused. Others did not refuse to lose money to get visibility...

> Apart from very non-embarrassingly-parallel problems, the 5-6 XServe-cluster super-cluster would be much more efficient

What do you mean by more efficient ? The upgraded VT cluster gets 60% of efficiency on Linpack, MareNostrum gets 65% of efficiency on a larger machine.

BTW, MareNostrum is supposed to be 2200 nodes. They use only 1782 nodes for the Top500 runs, ie 70% of the final machine.

> this MN looks too an expensive toy compared to a whole bunch of VT-clones

Price is not a reliable comparaison points. It's hard to know the real price tag on a machine, without services or other things and the money/PR tradeoff messes up everything.

Re:#4 is also academic

[danigiri]

I looked up the price in the supplied link, I looked for confirmation but couldn't find any other price figures. You mention a price of $20M, please include link(s).

Regarding the visibility, if VT managed to get lower prices in exchange of publicity, I say it is a smart move on their part. Everybody is entitled to do this and can do it, so it's a level field. MareNostrum's builders should do it as well. Component maintainers and suppliers have to value what they can win by giving discounts to important clients.

Hey, they do it at my local gym, by working at a big corporation I am given a good discount, that in exchange of the good publicity they expect me to give about them at work.

By efficient I meant in price/returns terms. VT is much cheaper and quite powerful even if the $20M figure were taken into account. Yeah, I know about the law of diminishing returns, but the VT-MN disparity is too big IMHO.

It's hard to know the real price on anything. Take for example potable water, quite cheap at malls but it can be argued that it's real economic and enviromental costs are quite higher. For the sake of comparison we can only stick to published costs. I am sure that at the UPC they'll leech work time from undergraduates and PhD students as well, just like everybody else, again level field.

beleaguered

[koi88]

Let's say ist all together: beleaguered
(if you know apple for a long enough time, you certainly know this word)

Re:beleaguered

[mastahblastah]

NOOOOOOOOO!!!!!!!!!!!! Not the "B" word !!!! ARRRGGHHHHH!! It Burns!!!, It Burns!!!!!!!!!!!

Re:two times less

[jbarlow]

Ok, this is just one of those things that bothers my proofreading brain. What exactly is "two times less" supposed to mean? Literally, it means that they should be shoving as much power back into the grid as these comparable systems. 'Cause "one time less" would be zero.

Try "half as much." Damn, I hate grammar nazis like me.

Not to mention that BG beats BigMac in flops/$

[joib]

The cost he quotes for the Blue Gene ($200 million) was the cost of some government contract that included BG/L, ASCI Purple (a huge cluster of POWER5 servers) and some R&D as well.

Recently IBM announced their commercial prices for BG machines (see e.g. theregister.co.uk or news.com.com). Prices start at $1.5 million (1 fully equipped cabinet). Using this price and published linpack figures one arrives at about 2.9 Mflop/s/$, compared to the maximum value of 2.2 Mflop/s/$ he quotes for the best apple system.

Add in the fact that the BG uses much less space and power than a comparable xserve cluster, that it has a faster and lower latency network, and we have a winner.

Re:Not to mention that BG beats BigMac in flops/$

[Anonymous Coward]

Recently IBM announced their commercial prices for BG machines (see e.g. theregister.co.uk or news.com.com). Prices start at $1.5 million (1 fully equipped cabinet). Using this price and published linpack figures one arrives at about 2.9 Mflop/s/$, compared to the maximum value of 2.2 Mflop/s/$ he quotes for the best apple system.

Actually, that 2.2MFLOP/s/$ figure for VT's cluster isn't the best for an apple XServe based system. The University of Maine built a 512 CPU cluster of XServes for $680K, or 3.012MFLOP/s/$, according to the price comparison's chart, edging out a BlueGene cabinet.

Of course, now I get to be flamed by a bunch of mac haters who think pointing out a factual error in your statement means I don't know anything else about clustering and will blindly chose a mac above all others.... *sigh*

Re:Not to mention that BG beats BigMac in flops/$

[Lars T.]

Since his point was PowerPC beats the crap out of x86 and Itanium, yeah, I'm convinced he will be disappointed it takes a PPC system to beat a PPC cluster.

Re:Not to mention that BG beats BigMac in flops/$

[iammaxus]

This is not a flame:

That UoM 512 CPU one is significantly smaller than the top 5 super computers. You should probably only be comparing in class. Costs do not increase proportionally with size of the computer.

Re:Not to mention that BG beats BigMac in flops/$

[kaldari]

>That UoM 512 CPU one is significantly smaller than the top 5 super computers.

Exactly how small is it? Smaller than a breadbox? Are you trying to say that UoM's supercomputer is significantly "less powerful" than the top 5? Actually, w/ 512 XServes I don't imagine it would be that far behind - probably around 30th place if it were on the list. Not bad for under a million.

Re:Not to mention that BG beats BigMac in flops/$

[oudzeeman]

I work on the Maine Xserve cluster, and while we didn't submit linpack numbers, I can tell you we would be FAR from 30th place if we had. It would be hurt significantly due to the fact that it uses GigE and linpack is very sensitive to the network (our code is much more forgiving when it comes to latency).

Re:need?

[mosb1000]

Your key mistake here is your use of the word "needs". The data I've seen indicates that the G5 draws an equivalent amount of power as comparable Intel and AMD systems. Also, the G5's in the x-servs are air cooled. I think they mostly liquid cool the dual 2.5 Ghz G5 just to keep the noise down.

PPC970 best LinPack performer

[taharvey]

Look at the TOP500. Do the calculations yourself. Opteron is NOT more powerful, just the opposite.

VT X G5 2GHz 8 Rpeak/processor
NCSA Xeon 3.06GHz 6.12 Rpeak/processor
LANL Opteron 2Ghz 4 Rpeak/processor
PNNL Itanium 1.5 GHz 6 Rpeak/processor

Its the same no matter what supercomputer cluster system. PPC970 kick some serious whoopass.

If you want to know WHY people get all excited about system X, look at performance per processor and performance per dollar.

Re:PPC970 best LinPack performer

[Anonymous Coward]

Way to prove the original poster's point..

*spunk* *spunk *spunk* Apple! *spunk* *spunk *spunk* Mac! *squirt* *squirt* *squirt*

Christ, lay off the Kool-Aid for a while, ok?

Re:PPC970 best LinPack performer

[Anonymous Coward]

Umm.. if you actually look at the top5 instead of getting all a-little-bit-to-happy about Apple hardware you'll notice that #4 (as I thought I pointed out) is indeed running Power970 proccessors.

Also the fastest computer is a PPC 440 proccessor..

Also because the PPC 970 can outperform in certain benchmarks does not make it more powerfull all around. It just makes it faster at that specific benchmark.

And I would also like to point out that Power970 is a IBM cpu, not a Apple design. AND I would like to also point out that the Opteron is spawn of AMD + IBM, the vast majority of the technology that is present in the PowerPC is present in the Opteron. Don't make me google for bunches of links, just take my word for it. What the Power970 has that the Opteron has is the Altivac portion of the CPU, which is what gives its impressive linpack performance.

It's nice for benchmarking but it doesn't have practical use for everyday tasks. What the Opteron was designed to do was to provide a fast CPU for servers, which it does very well and outperforms the Power970 in most benchmarks, even when compared in a single cpu configuration (opteron) vs dual cpu setup.

Also I would like to point out that #4 computer runs Power970 proccessors + runs Linux 2.6.

Linux is also slightly more efficient.
The Spanish power970 computer.

I'll use rough math to show it:
Barcelona Supercomputer:
20530 RMAX GFlops / 3564 proccessors / 2.2 ghz =
2.62 GFlops per Ghz aviable.

System X
12250 RMAX GFlops / 2200 proccessors / 2.3 ghz =
2.42 GFlops per Ghz aviable.

So are you going to argue to me that the Apple's Power970 machines offer better per proccessor performance vs IBM's own Power970 machines? Or that they offer much better cost performance then IBM's low-end 700mhz Power440 cpus?

There are lots of tasks that are much better served by single image machines vs cluster.

Look at #2. That's a Itanium2 machine. It runs 512 cpus in a single machine image.

That's right, 512 cpus for ONE linux 2.6 kernel.

You know those jokes about "what if I make a beowolf cluster of supercomputers"?

Well that's a beowolf cluster of 20 super computer machines with 512 proccessors each. It can do tasks that would be very inneffient to carry out on a cluster.

Low cost is nice, but you can't use the basis for one artificial benchmark as the ultimate grade of a supercomputer's power.

That would be like me taking a single benchmark of Quake3 FPS and comparing a Power G5 vs a Dell gaming machine and declaring the Dell gaming machine as the "most powerfull PC of all times".

So I ask you again.

Why the concentration on a Single Power970 machine, while ignoring the other Power970 machine not to also ignoring the Power440 machine, which happens to actually be the most powerfull machine aviable.

Re:PPC970 best LinPack performer

[Kuad]

RPeak == meaningless statistic. Machines can achieve anything between 40 and 90 percent of their RPeak. Its only value is in determining theoretical maximum performance.

Linpack is also *very* happy on the PPC architecture. Opterons regularly spank PPC chips on more common workloads.

Re:PPC970 best LinPack performer

[Smurf]

You make very good points and I agree that it's absurd to ignore the other machines, some of which advantage System X in more than raw power.

But going directly to your question: Why the concentration on a Single Power970 machine ?

A possible explanation: It is the first machine of the list that is made up of individual servers that some people would actually have in their small company. And its previous encarnation was made of workstations identical to the ones that several people have on their desks. You can't help but get some emotional attachment to the VT cluster.

I know it sounds silly to you and that it's not a technical justification. But certainly the other machines look more impersonal and distant to the average user (specially the Mac users). I can understand that, and I don't even have a G5.

Re:G5s spanks the other processors, look at Rpeak!

[Anonymous Coward]

Huh? The argument is that LINPACK is not the be-all end-all of supercomputer performance, so you go and quote LINPACK scores as a counter-argument?

What i really like is

[imsabbel]

How MACItiots who are like "Macs arent overpriced, you are paying for quality, users of pcs are people who cant spent money for a real computer",ect for years now all unisolo see the virtues of the cheap mac clusters and how they are better than everything because they are so CHEAP.
Well, a good answer would be using their own argument against them: Who buys MAC-clusters buys CHEAP, who buys power5/itanium/opteron clusters buys premium quality...

Re:What i really like is

[AusG4]

Does anyone know what this guy is talking about? Who is the wise prince? What is happening?

Re:need?

[10Ghz]

Your key mistake here is your use of the word "needs". The data I've seen indicates that the G5 draws an equivalent amount of power as comparable Intel and AMD systems. Also, the G5's in the x-servs are air cooled. I think they mostly liquid cool the dual 2.5 Ghz G5 just to keep the noise down.

uh-huh. If G5 runs so cool, then surely they could have kept the original cooling-system for the 2.5GHz model, instead of going for an complicated liquid-cooling system? Really, why did they move from heatsink/fan to liquid-cooling? AFAIK the original G5's were already quiet. And looking at reviews such as this seems to suggest that the G5 does indeed run very hot.

And looking here and here I can see this:

2.5GHz G5: 75-85C during load
2.2GHz Opteron: 48C during load

G5 runs cooler? Hardly.

Fan boys read only what they want to see!

[mdemirha]

Well, I really am getting sick of this Apple fan-boyism. What is up with that? People here scratch their heads all day long and try to find some calculations that may show that Apple higher than the others even though the comparison is a just an apples to oranges comparison. One writes that Flops per $ for apple's are better? How the hell can they make a real comparison? Why dont they look at the Flops per CPU? What are the other hardware in those systems? Why dont they compare all the machines? Do they have the same HDD/RAM/..etc other parts. If not (and it is not) this is a totally useless comparison.

Now another fan boy writes that 3600 2.0 GHz G5's beat 4000 1.7 GHz Itanium 2s. What a comparison! My god!

One other fan boy writes that AMD Opterons beats the crap out of Intel systems. Looking at the link he provided, it is pretty clear that Intel Itaniums beat the crap out AMD systems and then the fan boy defends himself by saying that AMDs are cheaper!

Oh come on now people, be a little more objective! The article says the following: A total of 320 systems are now using Intel processors. Six months ago there were 287 Intel-based systems on the list and one year ago only 189. # The second most common processor family is the IBM Power processor (54 systems), ahead of PA Risc processors (48) and AMD processors (31). # At present, IBM and Hewlett-Packard sell the bulk of systems at all performance levels of the TOP500.

These are much more important numbers than some uber-geek-fan boys calculations. It is apparent that Intel has increased its percentae *A LOT*. AMD also started to putting many systems into top 500. In my opinion, Apple's success in this list is MUCH MUCH LESS than Intel's or AMD's or IBM's or HP's successes.

Be a little more logical, open minded, less fanatic people. Apple is just a freaking computer like any other computers around. It is not some sort of a super/splendid/magnificent/God-Like computer.

Re:Fan boys read only what they want to see!

[zpok]

"Be a little more logical, open minded, less fanatic people."

So we should say "It's really no achievement to have a supercomputer in a pricerange available to institutions other than military. The fact that they use Apple's G5 and OS X, an almost out of the box solution is totally irrelevant. If you like you can build your own courtesy of Virginia University, but who would want a supercomputer that's cheaper than the other twenty first contenders in the list of supercomputers. Remember, they're Apple, so they're crap. And expensive, whatever the calculations say. They must be. They're Apple. I repeat, they're Apple. Crap. Be realistic, don't be a fan-boy."

Fan-boy indeed.

Re:Fan boys read only what they want to see!

[Anonymous Coward]

Maybe you should try reading the posts instead of spouting off on them.

I am the one that said that 3560 Power970's beet out 4000 Itanium2's IN THE RANKINGS.

Dipshit.

plus the #4 machine isn't even a Apple. And Power970's aren't even Apple proccessors. They are IBM's and they are used in some of Apple's computers.

It's like if I talked about intel proccessors and you freak out about me being a Dell fanboy. Get a grip.

Re:Fan boys read only what they want to see!

[Anonymous Coward]

Thats fine .. i think we knew that it was cheaper fucking 10 look at me articles ago. Thats the whole point asshole.

Re:Fan boys read only what they want to see!

[xenoandroid]

Sounds to me like a case of penis envy.

I'm sorry you raise the money to buy the mac you want.

Re:Fan boys read only what they want to see!

[zpok]

Asshole? Nice. Be my guest, by all means, very constructive, adult and erm... smelling of fan-boy stupidity.

I've read pretty interesting posts on the issue, talking where cost, clustering or even simple stupid speed isn't the issue. The vast majority enabled to bring the point across without being 6year old anal retentives in the process.

So why not make the world a less smelly place and either make your point without drooling and pissing around or post your own articles on /. without any mention whatsoever of whatever makes you go agagagaga

Re:Fan boys read only what they want to see!

[Anonymous Coward]

I'm about to swear in the church of Apple and point out that the Virginia Tech cluster hasn't been very good value for money _this_ _far_. Since they bought the machine over a year ago, it has spent months in the testing phase, and was then taken down entirely and rebuilt.

Remember that a HPC (high performance computing) cluster typically has a lifetime of 2-3 years. Viriginia is now finally operational (meaning scientists are using it for production runs), but at least 30% of the lifetime was wasted.

Agencies and Universities pay for these computer so that scientists like my group can use them, not to get a high rank in pissing contests. By and large, in mosts cases it is much wiser to pay a vendor and have it running in a week (Spain) rather than having professors and students spend a year to get it running.

Re:Fan boys read only what they want to see!

[zpok]

Good points, and yes, the Spanish accomplishment is something, muy impresionante por supuesto.

Re:Fan boys read only what they want to see!

[zpok]

OTOH they've done the necessary research to make it a good project for other institutions with the budget and the need for (clustered) HPC. So all in all not entirely wasted.

Still, good point and still kudo's for the Spanish.

Re:Fan boys read only what they want to see!

[Anonymous Coward]

" ×^" - Wayne Teasdale " ×^" = "{mind reboot}" = Creation is the infinite times the infinite to the infinite degree. "#s rule the universe" - Pythagoras "Good people do not need laws to tell them to act responsibly, while bad people will find a way around the laws" - Plato

A supercomputer application

[tezza]

Keeping track of the very high frequency of postings of these supercomputer rankings on Slashdot.

Can I vote for a supercomputer thread so that I can elect to have it not displayed in my preference? I wouldn't want to miss out all the other tasty hardware goodness. I don't mind news about new Supercomputer technology, but whoever holds the most teraflops at a certain point in time is not of interest.

http://apple.slashdot.org/article.pl?sid=04/11/09/ 0126220&tid=137&tid=126&tid=181&tid=1
November 9th, 2004

http://science.slashdot.org/article.pl?sid=04/11/0 6/2239245&tid=136&tid=137&tid=14
November 7th, 2004

http://slashdot.org/article.pl?sid=04/11/03/161424 5&tid=137&tid=139&tid=1
November 3rd, 2004

http://science.slashdot.org/article.pl?sid=04/10/2 7/0147206&tid=137&tid=139&tid=14&tid=106
October 26th, 2004

http://apple.slashdot.org/article.pl?sid=04/10/26/ 0636230&tid=137&tid=3
October 26th, 2004

http://science.slashdot.org/article.pl?sid=04/10/2 0/1727255&tid=137&tid=136&tid=14
October 20th, 2004

Virginia fastest academic?

[shiruf]

Virginia Tech's 'System X' Xserve G5 cluster [...] remains the fastest supercomputer at an academic institution.
Beg to differ: #4 is about 5 mins from home (by bus), in the northern campus of universitat politècnica de catalunya. And, yes, part of the institution, not some loaned space or something. Mind you, one wishes Spanish Universities involved their students a tenth as much. S-2.

Why is nobody copying the Big Mac?

[groomed]

We can toss around numbers all day, but these numbers are only meaningful in so far as they persuade the people who actually build supercomputers to copy the Big Mac design. I don't see anybody doing that, so it appears to me that the Big Mac, on the whole, is not very significant.

Re:Why is nobody copying the Big Mac?

[Chiron Taltos]

You are so correct! Nobody, and I mean nobody, is copying the success of Apple's X-serve at Virginia Tech.

Not only is no one doing it. But this nobody certainly doesn't expect to duplicate the success of the Virginia Tech Big Mac.

Re:Why is nobody copying the Big Mac?

[99BottlesOfBeerInMyF]

so it appears to me that the Big Mac, on the whole, is not very significant.

Maybe you are trolling, but, do you really propose that the significance of any scientific endeavor should be judged by whether or not is is copied? John Smith demonstrated a car that runs on potatoes and can be built for only $500. Still, no one has copied him so his design is not significant. Your statement does not sound very "scientific" to me. That said, There are several institutions copying the design, including several DOD projects. More importantly, you know that Virginia Tech exists, which I doubt more than 1 in 1000 Slashdotters knew before they created the Big Mac. I'd say that they are successful all around.

Fan-boy complex

[zpok]

I've read a few posts on how Slashdot is taken hostage by Apple fanboys.

So what I propose is that even when Apple related posts are few and far between, we should always add one of the following conclusions:
1) Apple is still expensive, even if they're proven cheaper than the alternative.
2) Apple is still slow compared to anything my cousin Ned can build in his back-yard.
3) Jobs is an ass-hole.
4) You know, any objective computer user has to acknowledge the simple truth: Apple sux and anyway is dead, gone and dying.
5) Apple's are sissy.

And if that doesn't prove enough, maybe some "Linux Rulez!"

I'm only slightly kidding here...

MACItiots

[zpok]

Rrrright.

Us dumb, stupid people with our virus- and bluescreen-free machines. Stupid stupid stupid.

Re:MACItiots

[Anonymous Coward]

You forgot games on that list, too.

Re:MACItiots

[zpok]

It's hard to be sarcastic on mac versus PC games.

If you're into games, buy a PC. Sad but true. The coolest, most fun game I ever played was on a PC. I had all the typical PC/Windows horrors on drivers, soundcards, upgrades, viruses and whatmore, but I enjoyed the game tremendously.

And yes, I'm a lifelong Mac user (since '84). I have a PS2, it still doesn't come close to the wonderful gaming experience I had with that piece of junk PC I finally had to bury with a hatchet...

Cheers

HEAVEN TO POSTER:

[Anonymous Coward]

Christ, clusters are not the end all and be all of high performance computing systems.

Thank you, but I know already. Father certainly has more bandwidth than Big Mac. (But damn Akamai!) What is strange, people are suggesting that I have a high latency, though. Surely they have a low patience for these things.

Yours, "X"

DB Error: Connect failed

[fullofangst]

Man you guys slashdotted the server.
Top500.org are gonna need to rent time on the BlueGene just to host their site now.

Coup? Cuckoo!

[PingPongBoy]

5.7 TFlops in less than a square metre coming to a store near you, an IBM Blue Gene. If these start appearing at Wal Mart, it's time to redefine supercomputer.

Re:VA Tech Supercomputer (hahahaha)

[mcknation]

That's funny. When the 1st article about Va. Tech building this machine graced the pages of Slashdot someone posted the signup page for students of Va. Tech on Slashdot.

I'm not a student at Va. Tech.

Hell, I didn't even graduate High School...I found it much more fun to follow Bands (and the BEATUIFUL girls that follow bands) around the country.

No-one checked that signup list against currently enrolled students. I showed up and helped build the 1st machine. The one with the original G5's. I helped pull the cable, place boxes and install those ram chip laden Infiniband cards.

Karma and the morals I learned from running around the country eating grilled cheese sandwiches kept me from liberating those ram chips from the Infiniband cards.

NDA? I signed nothing. Futuristic hardware that the public is not to see? Yeah whatever. It was all off the shelf stuff. The Infiniband cards were shipped in from (or at least the boxes they were in) were stamped Israel.

The closest thing to oversite at these pizza parties was an Apple rep. I think he was too warped by the Jobs distortion field to notice what was going on.

Heck I had my camera phone on the whole time. I could have taken pictures.

Do you usually whip out your camera phone and take pictures of the stock on Bestbuy's shelves? Apart from the Infiniband cards and some of the Cisco gear, I've seen everything they used at the local computer store.

I've got a HUGE roll of tinfol at my house if you want to stop by and borrow some.

/-McK

Re:Just imagine

[Anonymous Coward]

which will push the limits of any machine

Nethack on Longhorn!

This is the top 500 tested machines

[Andy_R]

It's not an absolute top 500, so there is not a lot of military simulation hardware there, nor anything that won't run the test suite.

I wonder where google's cluster would rank in the top 500, and how about the machine(s) that run Eschelon?

Would latency be too big a barrier to a seti-style distributed effort getting into the top 500? t500-at-home anyone?

Re:This is the top 500 tested machines

[bhima]

I had the impression that google machines were in various locations. When I search I windup searching in German in the ".at" domain, because that's where I am.

Surely those machines are not in the US?

I guess what I'm saying is that... yes the latency would be too big a barrier

funny or funny?

[ynotds]

Only mods from the red states think sex is funny.

You must've been modded by guys from the blue states or even overseas like me who realise it is Slashdot that is funny enough to ROTFLOAO, or at least /.'s effects.

Getting back to XServe clusters, what really interested me was that some under $1 million made the list. I reckon there might be a business plan or three there.

Re:funny or funny?

[Zork the Almighty]

I think cheap clusters will be old news in a year or two. IBM's Blue Gene designs are so space efficient, I think they will shift the whole market in that direction.

Re:funny or funny?

[Harbinjer]

Exactly. I think Blue Gene will become a commodity system when IBM begins putting it on the shelf.

Re:funny or funny?

[evilviper]

IBM's Blue Gene designs are so space efficient, I think they will shift the whole market in that direction.

You must live in a city. Most places, space is almost a non-issue. If you have to build a wherehouse to save a couple million dollars on your supercomputer, it would be a good option most places.

Re:funny or funny?

[Zork the Almighty]

Maybe you are building a shitty warehouse then ? You know, supercomputer rooms actually have special requirements. Things like temperature and humidity control. You can't just stick these things in a barn full of chicken shit and have them work properly. Large, inefficient designs (for buildings and for computers) increase the costs for EVERYTHING, including actually running your supercomputer. Some of us city folk know these things.

Re:funny or funny?

[evilviper]

A "proper" building to house a large supercomputer (cluster) is still going to cost FAR less than the millions of dollars saved in inital hardware purchase costs.

Re:funny or funny?

[Zork the Almighty]

Well who am I to argue. I think the savings in infrastructure and operating costs, coupled with the convenience of a smaller system (later, when you want to upgrade it) are worth a good premium. But hey, if you have cheap land and resources and no qualms about using them inefficiently, then I concede. SUV's are great for grocery shopping too, so long as the price of gas is cheap.

Mirror Here

[MartinB]

Fortunately, they anticipated the slashdotting, and provided their own mirror. It even has an apposite fqdn.

SHUT YOUR HOLE

[Anonymous Coward]

As Duke Nukem would say just before he blasts your head off.

Can't you see? Intel is dying and so will Windows, so lets celebrate the dawn of Linux & desktops as well as BSD Unix on Mac OS X, shall we?

Free Linux for everone and the money people buy Mac's.

Re:two times less

[pkhuong]

Too true! Twice as little is also accepted i believe. :)

MOD PARENT DOWN

[Anonymous Coward]

For not being sufficiently deferential to Apple.

MOD PARENT DOWN

[Anonymous Coward]

For failing to be sufficiently deferential to Apple.

These are the ones you're allowed to know about...

[JustNiz]

I bet some of the Echelon stuff makes these babies look like .. well.. babies.

BDH

[Anonymous Coward]

Beating the dead horse or what. Yes Apple fanboys we know about your super computer .. yes we know its cheaper than anything.

Re:i work at #9.....

[Cutting_Crew]

and also i am proud to let you know that Mississippi is ranked #3 overall for supercomputing power, behind California and New Mexico. pretty cool stuff.

Not a good showing for Sun

[Anonymous Coward]

Did anyone else notice the stark lack of Sun machines in the list? There is only one SPARC based machine that I can find (there is another based on the v60 platform, but it's intermixed with IBM machines as well). There was a time that almost every E10k out there showed up on the list, how far and how fast Sun has fallen!

Are prices on that comparison adjusted for date?

[Sir_Ahzz]

THat's the big question. IF theywere, then it's an obvious win for macs. But something tells me that they aren't. (note the dates for the athlons are OLD by a year or more than the "freshly upgraded" G5*s.) Anyone care to adjust the system prices based on equivilant date costs for the whitebox systems and re-calculate that table? I'd be interested to see how close the athlon64 CPUs come now. Lies, damned lies and statistics.....

The big news here...

[wondermog]

is that the world's most powerful computer is now located in Minnesota, Y'know!

Re:These are the ones you're allowed to know about

[Oddly_Drac]

" I bet some of the Echelon stuff makes these babies look like .. well.. babies."

They'll be dealing with bandwidth rather than raw computational power, mainly because Echelon is about signal intercept, and they very rarely create these things in isolation. As I recall the NSA used to use computers created for Naval weather analysis until someone outed the Puzzle Palace.

We need a Steve Jobs Benchmark!

[Anonymous Coward]

How fast does it do a 500 pixel Gaussian Blur on a 300MB Photoshop file, convert it to CMYK and rotate it 22 degrees?

(betcha it comes out #1!)

What they didn't mention...

[Anand_S]

They neglected to mention that the real reason for creating a 70 TFlop system was to get Yum to run smoothly.

Re:need?

[Kyusaku Natsume]

It's because the die size is smaller, the G5 has less surface to dissipate heat, that's the reason for the high eficiency liquid cooling.

Yuppie!

[Anonymous Coward]

Apple yuppies .. another weekly installment of hey look at how great the Apple Supercomputer is. In the mountain bike world Apple people would be considered cross country yuppies who can't do anything but cut up everyone elses stuff. Where as freeriders can do lots of different things and appreciate different types of riding. Comes down to Apple people just having more money than brains. They are simply Yuppies in every sense. I personally can't wait for the Apple SUV.

Re:need?

[xenoandroid]

The reason it runs so hot is because the system is designed to be as quiet as possible. I'm sure if you made the G5 fans blow at full blast constantly the temperature would not be nearly as hot.

Some people actually like having quiet but powerful systems.

Re:need?

[znu]

It doesn't make much sense to compare operating temperature differences between machines with different cooling systems. There's a much easier way to figure out how much heat a processor generates: just look at how much power it consumes. An Opteron at 2.2 GHz sucks 89W. A PPC 970fx at 2.5 GHz uses around 50W.

Re:Are prices on that comparison adjusted for date

[bhima]

Yeah, I know what you mean. If only I could buy a PowerMac use it for a year and then pay market prices for it, I'd save a lot of money

Re: IBM knows how to play

[booch]

Check out the 2nd chart in the cost comparison. IBM plans to have a Blue Gene/L system 4 times as fast as their current number 1 system by May 2005. Not only have they sniped the lead, they're planning to completely out-class the competition.

Re:two times less

[Queer Boy]

What exactly is "two times less" supposed to mean? Literally, it means that they should be shoving as much power back into the grid as these comparable systems. 'Cause "one time less" would be zero.

Next you'll tell me the square root of -1 is an imaginary number, sheesh!

Thank You

[Anonymous Coward]

I have just poured hot apple juice down my pants. Thank you.

Supercomputer Window-shopping

[speedbump]

The top 500 list is interesting, and fun to read, but LinPack and other benchmarks only give us a rough feel for how well these monsters perform.

We're not comparing apples to oranges, more like ORCHARDS of apples to oranges...

I play DFT on HPCx

[Dr. Stavros]

In my research, I get to play on the machine at number 27 in the list. Someone here must be able to beat that.

We use it (along with several other machines) to do modelling of the physical and electronic properties of atomic-scale defects in semiconductor materials, using a density-functional theory (DFT)-based code.

They stopped selling the 400SC

[Ayanami Rei]

The i875P chipset-based system was too easy to upgrade into a decent system. :p

However, the 420SC is a _really_ good alternative for a super-compute cluster component. It's got a built-in Broadcom 5703, and you can add an intel Pro-1000 PCI-express dohickey for $100 more. Free upgrade to 512MB DDR2 RAM, an upgrade to 800FSB 2.8GHz PIV and you're looking at like $500 or $600 shipped.
Not bad!
And you get disks for free, although you probably don't need 'em.
And I could see you ripping the guts out of these guys and stacking them with hex spacers to save space.
This would be a great node component for clusters that solve both embarrasingly parallel, and even not-quite-parallel problems too.

Re:They stopped selling the 400SC

[timts]

one thing bad for power edge servers in a cluster is that they are not rack mounted. you probably dont want 1000 machines to be in a room, stacked up. believe me, we had a server cluster like that and we hated it.
on the other hand, that price is unbeatable.

[geeks should put science before faith, rationality before belief. I just spent my moderator point down rate those flamebait and troll from mac/linux fans, it's normal I got flame back from them]

#4 JS20 cluster is the fastest academic computer

[antonb]

The VT cluster is not the fastest supercomputer at an academic institution any more.

Not necessarily.

[Ayanami Rei]

Consider:
420SC w/ 3.4GHz PIV on 925X chipset. Say, 2GB of DDR2 RAM.
$1900, but quite probably less depending on quantity.
Has one PCI 8x slot open. Fill that with a Infinihost III card.

Or do 1GB 533 w/celeron for 1K.
Or do 512MB w/celeron for $320.

It's not a pretty server. But it does the job.
You won't even bother with racks. Just buy el-cheapo garage shelving units from Home Depot. You can get 19" rackmount boxes to put on the top shelves (for the switching/monitoring equipment).

And beautfitul things keep coming:

[Ayanami Rei]

See the Ultra-Low-Voltage K7s, the DTR mobile K8s (especially since the 90nm shrink), dual-core opterons, and especially the Dothan Pentium-Ms... ::droool::

And where is Sun? Will Fujitsu provide an UltraSPARC that will make us all say wow again?

Re:need?

[default luser]

Not true. The numbers I have seen from IBM have put the 970FX 2.5GHz at ~65w, full-load.

The 90nm Opterons are showing similar power characteristics at matching performance levels, but that's no surprise; they also take advantage of IBM's advanced process techniques (SOI).

The 970FX is an impressive chip, but it's nothing particularly special in this competitive environment. VT's supercomputer is an impressive undertaking for it's thrift, but the fact of the matter is it could have been done with other systems on the market with similar end characteristics.

Re:These are the ones you're allowed to know about

[Anonymous Coward]

Well the "Beast" in Brussels UN is probably up there also .. but its mainly a database of everyones info to be put to Big Brother uses in the very near future if not already. I for one welcome out new supercomputer overlords? :/

NCSA's Tungsten

[cylcyl]

I'm just amazed that they were able to squeeze so much performance out of a PDA!

Re:Are prices on that comparison adjusted for date

[Sir_Ahzz]

I should elaborate... The grid shoudl be based on "if you bought it today". Those that have supercomputers already won't care about the price/FLOP. Those that would find the report the most usefull would be groups looking to invest in a new supercomputer, and it's misleading for that.

Re:Are prices on that comparison adjusted for date

[99BottlesOfBeerInMyF]

Of course, if you want the numbers to be even more meaningful you should reference the cost of the systems, not today, but the length of time it takes to build them before today. Right? Of course there are advantages to quick deployment that in many cases are just as important as cost, so perhaps you should also add to the cost, the price of hiring out comparable computing resources for the time it takes to build them. Of course all of this is moot since the Big Mac cluster was so much cheaper AND faster to build than anything else in the top 10 it wins any non-biased comparison hands down. (Given that your goal is to cheaply and quickly develop a cluster for a purpose for which the LINPACK test is a good benchmark.)

Google Cache

[marktaw.com]

You do realize that the page you linked to was a giant page with a giant GIF. Linking to the Google Cache version only saved them about 50% of the bandwidth because the giant image still had to be loaded from the original site.

Google may have a complete mirror of the web, but only the text portions (and thumbnails of the images).

Re:need?

[Lars T.]

Really, why did they move from heatsink/fan to liquid-cooling?

Easy. The 2.5 GHz 970FX is a 90nm chip and has a die area of 66 square-mm, the 2.0 GHz 970 is a 130nm chip at 121mm2 and the Opteron / Athlon64 chips are193mm2 / 144mm2 large. So what do you think happens when you generate less power on much less area?

Re:Just imagine

[H8X55]

who did you think 'anonymous' was?

$600,000 System X upgrade was a VERY special deal.

[MojoStan]

comparable to System X's $5.8 million overall price, including the upgrade to Xserve G5s

Don't forget that this is a "one time deal" that no one else can get if they want to build an Apple-based supercomputer.

As this article states, the $5.8 million cost was calculated by adding a $600,000 upgrade cost to the $5.2 million cost of the original PowerMac-based System X. As Dr. Srinidhi Varadarajan said in this article, the original System X cost $3.2 million for cluster hardware plus $2 million for facilities upgrades.

The $600,000 upgrade to System X included upgrading all 1100 PowerMacs to dual 2.3GHz Xserves, plus 50 additional nodes. Note that the fastest Xserves Apple sells to everyone else are only 2.0GHz, so System X got "extreme" versions of the Xserve.

A dual 2.0GHz Xserve "Cluster Node" starts at $2999 at the Apple Store. Since each node has 4GB RAM and the cheapest 4GB RAM upgrade costs $1450 at the Apple store, that makes it $4449 per node. According to this article, Small Tree's InfiniBand cards cost $1095 each, so that makes it $5544 per node (without cables). Therefore, Virginia Tech should have spent at least $277,200 for the additional 50 nodes.

That leaves at most $322,800 to upgrade the 1100 PowerMacs to the special 2.3GHz Xserves. That's about $245 per node, not including any additional costs I can't quantify like labor, additional hardware, and facilities upgrades (if needed).

No one else can buy 1100 dual 2.0GHz PowerMacs and expect to upgrade them all to dual 2.3GHz Xserves (with ECC memory) for only $245 per node (including labor). Comparing the cost/teraflop of System X with non-comparable government-funded, high-bandwidth supercomputers seems silly to me.

Re:need?

[Xabraxas]

uh-huh. If G5 runs so cool, then surely they could have kept the original cooling-system for the 2.5GHz model, instead of going for an complicated liquid-cooling system? Really, why did they move from heatsink/fan to liquid-cooling?

There is a Sony Vaio that is liquid cooled. It doesn't mean anything. It just means they decided to liquid cool it. Apple probably did it because they tend to be more about the whole package than most PC manufacturers, and liquid cooling it cuts down the noise.

MW/h

[scruffyMark]

MW/h would be to work as acceleration is to distance wouldn't it? Seeing as how a watt is already a joule/sec, a MW/h would be 1 million joules/sec /3600 sec, or 277 joule/(sec^2) So, if the thing 'works up' at 6000 MW/h, then after running the thing for a year, it would be cranking out 3600 sec/h * 24 h/d * 365.25d/y * 1 y * 277 j/sec^2 * 6000 = 52,448,731,200,000 j/sec. 52 trillion Watts - not so cheap anymore!

It may be top 10, but...

[Anonymous Coward]

They have built the cluster that is cheap, and has ranked top 10. Very cool.... All the talk is about the rankings and such but what about the machines original purpose? What have they done with it? Nothing?

BlueGene/P ~= 1,000,000 processors

[frenchs]

Last time I was reading about the BlueGene project. The BlueGene/P (P as in Petaflop), was supposed to have somewhere around 1 MILLION processors, and the goal was to achieve a petaflop with it.

BlueGene/L is supposed to, depending on the IBM press briefing you read, top out at 200-300 terraflops.

-Steve

Re:need?

[10Ghz]

An Opteron at 2.2 GHz sucks 89W.

No it doesn't. it has a power-envelope of 89W. Which means that AMD requires MoBo and heatsink/fan-makers to design their products in such way that they can handle a CPU that dissipates 89W. The actual wattage of the CPU is way below that number. AMD just plays it safe and gives the CPU quite a large margin to play with.

Re:need?

[10Ghz]

The reason it runs so hot is because the system is designed to be as quiet as possible. I'm sure if you made the G5 fans blow at full blast constantly the temperature would not be nearly as hot.

And AMD-systems are the same. In my case (2.2GHz Athlon64) the fans automatically adjust their speed according to the temperature of the CPU. The system is very quiet and during high load, I can hear the CPU-fan spin up a bit. So the fans do not "blow at full blast constantly". So you have to find another reason for G5 running so hot, whereas A64/Opteron does not.

Could it be that G5 simply does run hotter than A64/Opteron does? Why is that so difficult to accept?

the floating point operations thing (Altivec)

[kardar]

I think if I understand this properly, it's the G5 processor (and the previous ones too) that have the Altivec engine, which provides a level of "vector" capabilities that other CPU's simply don't have.

While there may be a bias (personally I am sick of the way EVERY FreeBSD, OpenBSD, and NetBSD post is being mishandled - it's unprofessional and destructive, and irresponsible given /.'s place in the IT community :), but I think it's worth mentioning that (correct me if I am wrong) - that the G5 (and G4 before it) have floating point, or "vector" capabilities, which really set them apart from the other PC processors such as the AMD and Intel. And probably better vector capabilities than a lot of RISC CPUs as well.

The G5 is an elegant, very cool processor, it's unfortunate that it's so locked into Apple's proprietary format - it would be so cool to build your own with parts, and have a choice of parts.

It's really an awesome processor. As to whether or not there's a bias, perhaps there is... but I think that when it comes to supercomputing clusters, the G5 has some unique capabilities that set it apart from other processors (unless, of course, you want to buy Crays and things like that). It would be interesting to see if there were some way (and this is way in the future, but just in terms of where things are headed) - it would be interesting to see DragonFlyBSD paired with a G5, or maybe at that point in the future it will be a G6 (who knows?). DragonFlyBSD is working on a way to (I can't find the info right now) improve clustering performance - if that comes to pass, and their PowerPC port comes to pass, even with purchasing a machine from Apple directly, you might have some very cool things going on.

Re:need?

[10Ghz]

Well 90nm Athlon64 has a die-size of 83mm2. And since they must run hotter than G5 does, it must need insane cooling-system just to keep them cool, right? I mean, since the die-size is so small and all? Wrong! They manage just fine with regural heatsink/fan! But how can that be??

So let me ask you again: Why does "cool-running" G5 need liquid-cooling for, whereas "hotter" A64's do not? Yes, G5 has a bit smaller die-size (66mm2 vs. 83mm2), but since G5 "runs cooler" that shouldn't be a problem? But it is, why? Could it be because G5 is NOT the "cool-running" CPU some fanboys make it out to be?

Re:These are the ones you're allowed to know about

[Oddly_Drac]

"Well the "Beast" in Brussels UN is probably up there also"

Only if we're counting fictional computers thought up by conspiracy theorists.

So, no.

"Big Brother uses in the very near future if not already."

You mean filming some no-marks around the clock in the name of entertainment? Or the fairly silly idea that Europe is spearheading an effort to slap everyone into a database. Have you ever seen the EU decide anything? Do you know that the EC meets in Brussels, not the UN?

Re:need?

[Lars T.]

Well 90nm Athlon64 has a die-size of 83mm2.

Where did you get that?

Re:need?

[10Ghz]

here, here, here and here.

Re:need?

[Lars T.]

Very good. And now you show me what cooling systems one actually needs for these new processors out for 3 weeks. Esp. since these are only allowed to reach 70C.

Regular Heatsink/Fan?

And while we're at it, keep them quiet.

See?

[mbbac]

Macs aren't expensive! You get a lot of bang for the buck with them. :)

Re:need?

[10Ghz]

Of course you can buy any kind of ridiculous heatsink/fan/peltier/whatever for your CPU. Just because you CAN does not mean that you must have one. The regural OEM heatsink/fan that ships with 90nm Athlon64 is just that: a regural heatsink/fan-combo. It's not some ridiculous apparatus as shown on your pic. And it keeps the CPU cool, while not being loud. The fan on the heatsink automatically adjusts it speed according to the temperature of the CPU.

So you find the most ridiculous-looking aftermarket heatsink on the market, and somehow assume that 90nm A64 somehow "needs" something like that? Sorry to rain on your parade, but it does not. OEM a&4 does NOT ship heatsink/fan like the one shown on your picture.

Is it really so difficult to admit that G5 is not the uber-cool processor some people make it out to be? Are you people living in denial or something?

Re:need?

[evilviper]

the G5 has less surface to dissipate heat, that's the reason for the high eficiency liquid cooling.

Can you possibly support your claims? Something more than an Apple press release I mean.

A HEATSINK is much more effecient at distributing heat from a tiny area to a large area than water could possibly be. Heatpipes work very well, also.

Besides that, the Intel P4 is a very large die, while the AMD Athlon/Opteron die is much smaller, yet the Intel P4 is the one with the cooling problem, and the AMDs cool very nicely

I've heard this excuse before, but it doesn't pass the laugh test, and nobody has followed up and provided a source of information for this unusual claim.