Slashdot Mirror
Drooling Over VA Tech's 1100-Node G5 Cluster
Mr. Slurpee writes "Virginia Tech's 1100-node dual 2 GHz Apple G5 Terascale Cluster is getting racked up and ready to roar. If you're a penniless geek like me, at least there's some tech pr0n for us to drool over. There's 1100 of them ... think they could part with one?" Update: 09/22 02:55 GMT by T : Matt submits a link to this full mirror of the photos, writing "The page owner's comment on the original mirror being taken down due to bandwidth? 'Bring it on!'"
Imagine a beowulf clu...oh, wait.
Sometimes i wish smaller schools had some pull so we could get projects like these...or even quarter, nay, a n eigth the size! problem is we don't have the immense research backgrounds and it holds us back.
JediLuke
-Do or Do Not, There is no Try
here.
Oh God.
Imagining each one of those came with just a little bit of Steve Job's Reality Distortion Field, someone from NASA might want to head over there and make sure that some kind of tear in space/time doesn't occur right there. With that many G5s, we don't know what level of destruction could happen.
Using full sized cases seems like a rather inefficient use of space to me. But I guess those cases are all fairly full - the heatsinks in those things are enormous. Wish PCs had heatsinks like that, then maybe mine wouldn't be so noisy.
-kidlinux.
This presentation contains content that your browser may not be able to show properly. This presentation was optimized for more recent versions of Microsoft Internet Explorer.
Why make a website "optimized for IE", when the content of the said website is of interest to people who are probably not running IE or Windows?
"Backups are for wimps. Real men upload their data to an FTP site and have everyone else mirror it." -- Linus Torvalds
These photos are from the Virginia Tech Supercomputer Cluster composed of 1100 PowerMac G5's. These photos were originally hosted at this link but access was removed, presumably due to excessive bandwidth usage.
:P Let's see how much power they really have!
Hmm. Something tells me that they should hook up a T3 line to each one of those G5's
----
Go canucks, habs, and sens!
do you think it would take this nifty cluster to correct the barrel distorition from their wide angle lesnse?
FooGoo
People who bite the hand that feeds them usually lick the boot that kicks them
They've got 1100+...where's mine? I ordered a Dual 2.0 GHz G5 in July....still no sight. Supposed to ship on Tuesday....but online time will tell....
Sigh...Maybe they'll loan me one if mine gets delayed!
PS--anyone got the rest of these pics? There were a TON of them...Mirror? COMPLETE?
> Surely it's a pretty expensive option Apple won the contract based on being the lowest priced option to give them the power they wanted.
insert all your g5 are belong to us joke here
I can just feel that some admin will get a deja vu very soon.
Probably just a small glitch in some system.
Proud patriot and republican voter.
What I really would like to know is how they install and configure all those machines. Their method of doing that will be very useful for even the (relatively) smaller networks that don't necessarily have to be clusters.
For example, I've yet to figure out a way to effectively get a computer lab with 30 eMacs installed and configured the same way. DHCP/Netboot is slow because we only have 100mbit switches. Split CD images are slow, and Jaguar doesn't yet have free software that does that yet (besides the dd of course). I'm not sure how to keep them all updated either.
I really hope they describe how they maintain the operating system on them.
$2,999 x 1100? 3.3 Meelion dollars? I hope they got Apple's .edu discount.
He had to admit they didn't have any in stock, and weren't expecting to get any from Apple for some time.
I guess I know where the dual-G5 systems are all going. Ah, well, it's all for a good cause. I hope.
Welcome to the Panopticon. Used to be a prison, now it's your home.
If I was working on that project I'd be like "Hey guys you only need 1099 right? I mean, it'll be okay if I just take ONE, right?"
Small mirror with a couple pictures if you click here
Kick in the Head
Somehow I think this cluster could beat me at Chess....especially the GNUChess that ships with OS X....
Here's why. Some of the more pertinent points:
Dell - too expensive [one of the reasons for the project being so "hush hush" was that dell was exploring pricing options during bidding]
Sun (sparc) - required too many processors, also too expensive
IBM/AMD (opteron) - required twice the number of processors and was twice the price in the desired configuration; had no chassis available
HP (itanium) - ditto
Apple (IBM PPC970) - system available with chassis for lowest price
GPL Deconstructed
It's a bit of a cocktease to post this link right now...Most of the mac community sites linked to the pictures at Virginia Tech's site but brought it down. Try clicking on the "pictures" link on their site and you'll se that they chmod 0'd the whole site so that the bandwidth usage won't peak out again
The pics at chaosmint are a small selection of what was originally on the site.
But to be on topic I'm suprised that Apple didn't get them Xserve G5's for the cluster. While the desktop G5's look cool it's really unneeded to use up all that space.
Pink at LANL has the following:
:)
1024 nodes
2048 cpus
1024 power cables
1024 Myrinet network cards
2048 fiber cables (8.8 miles)
3072 Myrinet switch ports
4096 sticks of RAM (2 Terabytes)
7168 fans
1 hard drive
1 CDROM drive
Not only do they have pictures of its assembly, they have movies.
Check the web page for more stats and better quality movies.
Oh, yes, it's unclassified
I haven't actually tried it yet since I don't have access to enough Macs, but I imagine it's something you would start and let happen overnight... I mean, that's more or less how Apple does it in their own stores, wipe and restore overnight, I think. Or at least after the store closes and before the next opening day.
GPL Deconstructed
Good thing VA Tech got their 1100 before all of the full-price-paying customers who ordered these as soon as they were announced.
64 bit does not make it twice as fast . Having dual 2ghz processors is not as fast as having one 4ghz(which unfortunately doesn't exist yet) Although having 2200 processors might be faster than 1 4ghz :).
PS. I'm waiting around for those 80gHz diamond wired cpu's. Then I can play Doom3!
Karma: -2^0.5 . Mainly due to the imbibing of dihydrogen monoxide
I have been using both Macs and x86 systems extensively for all manner of purposes since 1998. Neither really holds much of an architectural advantage when it comes to a specific type of function, and any slight advantage that is had by either tends to be so small that it isn't noticable. One thing is for certain: x86 hardware is far cheaper for the same amount of power.
Example: Photoshop 7 runs just as fast and efficiently on my dual 867MHz G4 with 512MB of RAM (booted into OS 10.2) as it does on my AthlonXP 1800 (1.533GHz) system with 256Mb of RAM running Windows 98SE, Windows 2000 Pro SP4, or Windows XP SP2. Video playback is identical (both systems have identical 64MB Nvidia graphics cards). Differences in compiling times are negligable.
Don't buy into the marketing hype. A $2,000 Macintosh system will run just as well as a $900 self-built x86 system. Everyone that has to use both platforms daily knows this. I don't have a preference for either architecture or operating system - just take my word as someone that has had to do extensive product testing on both x86 and Macintosh for years.
64-bit only means that the CPU can use memory addresses up to 64 bits in length. Got nothing to do with general computing power. IPCs (instructions per cycle), Mhz, bus speed, cache size and speed, SIMD implementation - these generally determine 'speed'.
(tig)
Ignorance and prejudice and fear
Walk hand in hand
It's not inverted; the Power Mac's USB, Firewire etc. jacks are on the left side of the front panel, so it's right-side up. See a photo of the Power Mac.
www.firastudios.com
Heh. So it does, at first glance. But I'm not sure if it would be practical to store 1100 g5s + racks on a floor made of your everyday celling tiles.
... or just look at that photo and notice that there are fans at the top. Either that or the guy is walking on the ceiling
Karma: Could be worse (could be raining)
I hope they were able to run these without video cards. I can't imagine 1100 brand-new sweet ATI video cards sitting idle for years...
It's rightside-up. The G5's drives are at the top of the case, ports at lower left. I'm curious why they don't face the computers.
HIV Crosses Species Barrier... into Muppets
As a Professional * Information Technology Location Analyst and Physical Security Specialist I need to use my professional abilities to make a professional analysis of the situation for my professional collegues so that we may put forth a professional solution to this problem.**
* - I really, really hate people who make gratutious use of the word "professional" as some sort of elitist mark of supremecy
** - I would like to run in there, see if the machines are locked down, and grab as many as I can hold.
(And yes, I'm just joking, I don't want to steal anything from them and I neither have the plans nor the means to do so, it's just a joke people)
All those silvery slick conforming cases remind me more of something from the book 1984. Ironic considering there advertising campaign of twenty years ago.
Interesting project though, and for those wondering what it will actually do: (from VT press release)
Virginia Tech researchers are already active in a number of areas that will benefit from the new supercomputing facilities, says Kevin Shinpaugh, director of research and cluster computing for the university. These include: nanoscale electronics, quantum chemistry, computational chemistry, aerodynamics through multidisciplinary design optimization, molecular statics, computational acoustics, and the molecular modeling of proteins.
And I thought it was going to run something with a cute name like iFiniteElementAnalysis...
One of the reasons VTech went for a G5 based cluster WAS price-performance...The mac option is cheaper then a PC aption and easier to install and maintain then Linux says the slide show. I might not fully agree, but thats their reasoning.
Sig (appended to the end of comments you post, 120 chars)
"Reality is just a convenient measure of complexity" -Alvy Ray Smith
Reliability.
No one in their right mind would try to argue that one couldn't build a home-grown system for less. But with optical ports? FW 400 and 800? Gigabit ethernet? USB 2.0? And with said home-grown machines, when the NIC goes bad in one, or a memory slot goes bad in another, who do you call? The NIC or mainboard manufacturer? So you what, keep a list of all your machines, give 'em i.d. numbers or whatever, itemize the guts and who made what (mainboard, NIC, RAM, CPU, HDD, etc.) of each, and hope to make sense of it all when stuff starts to fail? Me, if I was in charge of it, it would make sense to me to farm it all out to one company, and then when something breaks there is one number that I have to call.
Also, lets not forget that this is probably going to be used for research, and if it involves vectors, then AltiVec is the SIMD for you.
Of course, being human, my opinion is suspect.
(tig)
Ignorance and prejudice and fear
Walk hand in hand
I really don't get what you can get out of a single fancy expensive processor when a couple of less expensive chips can do the job .
Except that two fancy expensive processors are going to run faster than a couple of less expensive chips. Those are dual G5s.
Yeah, theoretically you could double the number of machines then... but I doubt you'd have any price advantage at that point, after doubling the number of cases, motherboards, etc. etc. Along with the cost of twice as much real-estate to pay for.
Sorry, but the 64 bit does matter with general computing power. It means that 64 bits of data are being brocessed at once instead of 32 bits. Otherwise, you are perfectly correct.
Sig (appended to the end of comments you post, 120 chars)
NO! It won't beat the ES. It won't even come close. The ES is a very tightly coupled system designed with heave I/O throughput from the ground up. This cluster isn't. And, ES up until recently was faster then top 20 US super computers combined. I doubt this VA-Tech cluster is faster in any dimension.
now supporting:
cmdrTaco for president '04
michael for oval office intern summer '05
I'm really feeling for the poor slobs who have to lift 1100 of those beasts onto shelves. G5s are heavy!!!
Ah, vivid memories of the cover of Softtalk magazine, with a picture of the Apple II assembly line with hundreds of machines. Just imagine... 200 * 64k = 12.5 MEGABYTES! That would take 90 floppies to store all that data!
Now some statistic pr0n:
There were about 5 1/2 million Apple IIs sold, so at an average of 64k each (just a guess), that would be 343 GB of memory total. Adding up the couple of computers in the office (it's a 4 person company), we're about 1/70 of the way there. Assuming 2 140K floppy drives per computer, that would be 1.5TB of disk storage -- that would be 6 hard drives, and they would occupy less space than a single pair of old floppy drives.
HIV Crosses Species Barrier... into Muppets
Nope. That's the ceiling. The fans would be blowing outward to help along heat convection. The pans (bottom) of each cab would also either have a fan blowing in or just simply a vent that would suck cooler outside air into the cab, through the perforated shelves past each machine and out the blow hole at the top. The reason the air is taken from the bottom is because cooler air is down there. Heat rises etc etc.
I run Contiki you insensitive clod!
Anti-social? My code is just platform-specific.
Truly amazing, how many of you ever thought you would live ling enough to see Apple win a contract based on price?
We once had a 'discussion' in an apple chat channel about errors made by the Apple Calculator. I just hope they are not visible when you want to do calculations on that cluster... 3.083 - 3.014 = 0.0690000000000003 is what appears on the paper tape from the calculator if you try this calculation. The display shows 0.069, but that is only cause it rounds off. I wonder where that 3*10^-x (too lazy to count) comes from, but can imagine that stuff fed to a supercomputer could give really wrong results if even small calculations are bogus!
Don't worry about them. All the lifting will be done by volunteers hoping for a chance to touch the g5s. The sore backs will be worth the chance to brag about this to their grandchildren.
I suppose I'm just anti-Mac trolling, but reading their slides, I can't help but get the impression that this university's RFP was intentionally slanted toward apple. Some things that seem interesting to me:
- It's clear from other posters that Apple did some shuffling to get inventory that wasn't available to the general public to this school. Not surprising, but indicates Apple was making considerable accomodations.
- I simply don't understand how full-size "commodity" Mac's could be cheaper than something like an IBM BladeCenter (especially in infrastructure (Switch, Rack, Space) costs). There's a lot of talk in the slides about the "required configuration" and the chassis... it sounds as though they placed some kind of hardware requirement on the proposal that could only be met by Apple. It just doesn't *sound* right, you know?
- One of the slides implies that they chose OSX over Linux because there isn't enough support for Linux. But this is a supercomputer! They're not running Photoshop on this thing. right? Aren't most supercomputer apps written fairly specifically for the machine they're running on? I really am asking the question here... is there something I don't understand about high-end cluster computing? I again simply don't understand how, at the super-cluster level, one could say Linux is poorly supported.
Of course, the university is well within their rights to buy from whoever they want. Their claims just seem questionable to me, at least from the brief slides.
I, for one, welcome our new 64-bit overlords!
Hi, your troll sucks ass. Perhaps you should elaborate on how the CPU is inaccurate rather than one stupid application.
Actually, the site is still up and more bountiful if you happen to be on VT's Campus. There are a lot more pictures of it available on that same link of the entire construction effort.
"Got nothing to do with general computing power."
Power is work/time. 64bit operations are 2x the work of 32bit ops.
IBM/AMD (opteron) - required twice the number of processors and was twice the price in the desired configuration; had no chassis available
Y'know, I saw this presentation a few days ago. I wasn't there, I saw it on the net. Anyway, this bullet point stuck out then - like, what are they talking about?
For one, how come it required twice the number of processors? From the benchmarks I've seen Opterons normally whup the G5, or are at least very competitive on paticularly G5 optimised code. Certainly not out by a factor of two, anyway.
And no chassis? What the hell does this mean? You can get 1U, 2U and 4U beast Opteron boxes from the likes of, well, IBM for one. As mentioned above.
It's not even like the kinda ropey nature of 64 bit Linux comes into play either because, well, there is no 64 bit OS X - unless VT know something we don't (which is always possible).
So, yeah, I think someone decided to buy all the G5's made for a month and just set up the project to make it happen. This "achitectural options" thing is horseshit.
Dave
I write a blog now, you should be afraid.
Listen, get a grip - when a component goes bad in a home grown system, you throw it away and buy a new one. Nobody cares - it's cheaper to just spend a hundred bucks or whatever then it is to dick with manufacturers. If you home grow, you home repair. Duh.
$45 per U Colocation Special
I wish I had mod points right now... you sir -- YOU owe me a keyboard, for I have just spit out my Mountain Dew.
Karma whorin' since 1999
Does it run Quake III???
- It's pretty clear to me that Apple didn't divert anything. If you look at the numbers, the VT order accounts for about 1% of all Dual G5 orders. That's hardly enough to cause the delays that people are seeing in their ship dates. Notice the slide states Apple offered an "early september" ship date, but Apple initially promised customers a mid-late August date. Given when those talks between VT and Apple were likely taking place, that means that Apple had intended to fill other orders first, and had a special allotment for the VT order.
- I don't know a whole lot about a blade center, but there doesn't seem to be a place to plug in the high-speed interconnects. Also, it runs on Intel chips that run hotter and do less work than the G5, especially when AltiVec gets involved, which is usually why you build a computer this size; vector processing. I'm also guessing the required configuration needed resale value to students at the end of life for the project/system.
- That's absoloutly true. When you need technical details about Linux you have to dig. When you have a question about OS X's guts, I'd guess you call Apple and have a conference call with all the coders (at least at this level of purchase/prestige). Could you imagine trying to get Linus, and all the other code writers for Linux and the supporting libraries and utilites on the phone at once?
Article X: The powers not delegated... by the Constitution...are reserved...to the people
SCO Sucks
No, the 'big fan-things' are called XDVs from Liebert (http://www.liebert.com/dynamic/displayproduct.asp ?id=1077&cycles=60hz). The mount on top of the rack.
Agreed!!! Near as I'm aware, apple doesn't offer a rackmount solution, and franky it's a pain in the butt to mount a g5 class motherboard in a standard rackmount case. I have much respect for Apple but strongly disagree with their policy on computer sales, basicly it really isn't an option to buy mac motherboards... so for the average person looking for a reliable resource for mac boards... you gotta buy the turn key solution and chuck the cases / drives / whatnot.
Apple probally should offer a rackmount solution, or at the very least offer some barebones systems for the enjoyment and pleasure of those who want to use this platform in the industrial enviroment or for the benifit of hardware hackers. Apple cases are spiffy but no replacement for a solid rack by any means.
There is no sanctuary. There is no sanctuary. SHUT UP! There is no shut up. There is no shut up.
Exactly.
Learn to read, VT told you
Because the G5 systems were the cheapest AND fastest
There is no rackmount version of the G5 yet. That would be an upcoming G5 Xserve that has not been announced yet.
Plus, I guess when this cluster needs upgrading, they can sell off the Dual G5s which should hold their value for quite a while, as they are just stock G5 workstations.
It is quite the fashion statement :)
(Excuse the blurriness and poor lighting - crappy cam and crappy dorm lighting)
Macs appear in films because of price, Apple pay for Macs' to appear in films.
Maybe VA got a 'special deal'?
thank God the internet isn't a human right.
All I can say is if you are getting write speeds that slow, when was the last time you defragged your hard drive? I've never experienced that slow of a write for a 17 Mb file on any of my macs. Is your hard drive full?
http://www.apple.com/xserve/
Configure one machine, use that one to produce a dupe, repeat with all currently configured machines as parents until out of unconfigured machines.
Still a huge job.
Ummm, when you have a question about Linux, you open up the kernal source tree.
Well, the same is true with Apple's OS, at least on the level these machines should be running at. Please don't tell me they're wasting process time running a GUI on all those headless boxes...
A Good Intro to NetBS
http://saveie6.com/
So you have 1200 identical machines. There's one NIC vendor, and 1200 NICs. There's one motherboard vendor, and 1200 MBs. You keep a stock of a half dozen of each component on hand.
This isn't a situation where you're going to hold a concert and have everybody bring in a random clone box for the cluster to get a discount on the admission price, ya know....
A Good Intro to NetBS
With an Open Source program, you don't NEED to get to talk to the original programmer for any reason at all. All you need to do is hire your own programmer to go through the code, and understand it. Then when a question arises, you can simply refer to your local expert, or look through the code yourself to locate the problem.
That is one very good point that I don't think anyone else has brought up (including VT, strangely). Macs are knows for having a significantly higher resale value than just about any other system out there. Just look for Macs on ebay to see the amazing prices.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
MacOS X on a recent Mac is a nicely stable and rather efficient (if slightly quirky) Unix, at least that's been my experience with a 700Mhz 384MB RAM iBook. It could well be that you need at least that much RAM before it starts to shine.
The IBM floating point is fast and almost too accurate (Consistency between runs on different architectures is sometimes more important than those last few bits of accuracy). Going to single precision tends to slow it down (extra step in throwing away information).
The bus in the new machines looks awfully nice on paper.
In a cluster: Reliability, Fast networking, fast bus and fast/accurate floating point are mighty big draws.
I don't think anybody will know for sure if this was a good move until it's been running awhile but it's not an unreasonable thing to try.
> cost vs. performance (purely)
Excuse me, but they just threw freedom out of the window...
In the slides, they say Linux hadn't enough support... hello? Support by whom? Apple should give them any specs they request, and I know for a fact that the GNU/Linux PPC developers, at least the kernel and Debian ones, are killing for a chance to have a go at these babies.
And what about rack-mounted units? Also not having ECC... looks like they wanted the toys and needed excuses to get them.
Leandro Guimarães Faria Corcete DUTRA
DA, DBA, SysAdmin, Data Modeller
GNU Project, Debian GNU/Lin
nice generic flame. i usually see it with *BSD instead of mac.
d'oh
My mistake... but still tower cases are not well suited to that enviroment
There is no sanctuary. There is no sanctuary. SHUT UP! There is no shut up. There is no shut up.
All it took was Apple's deep, deep discount for the marketing hoopla this represents for them.
This is gonna be a bullet point in every Apple advertisement for quite some time. It's damned cheap publicity.
A Good Intro to NetBS
How many times are you going to post this same stupid message? If seen it for about two years now, it's some sort of ancient Microsoft Form Letter from 1997.
An 8600/300? I doubt that you even know what that means. PPC604's kicked ass in their day.
It's mandatory to wash your hands before returning to the land of Dairy Queen.
176 TB is for the entire cluster
I remember years ago that it was difficult to get spare parts from IBM for PC-XT machines. You could go to the Ham Radio Swapmeets and there would be people selling stripped XT cases for a good price, because they were associated with a company that needed spare parts, and it was cheaper for them to buy IBM XT boxes and strip them for the parts they needed. It seemed nuts at the time, but, then, if you were servicing IBM equipment at the time the money was good servicing the machines.
A Good Intro to NetBS
Re: Processors
Perhaps for their benchmarks, the G5 was 2x the performance of the Opteron. Have you taken into consideration the Altivec processor, which happens to be 128bit in size? Any vector processing will be enhanced greatly by the powerful nature of the G5 in general, and especially when using Altivec optimized code. Couple this with IBM's XLC auto-vectorizing C compiler, and I wouldn't be surprised if Altivec did wipe SSE2/3D!Now; it's been discussed before that Altivec is a superior solution to MMX/MMX2/SSE, and SSE2, so there's no reason to doubt that when you pump up the FSB from 167MHz->1GHz, pump up the CPU from 1.4GHz->2.0GHz, on the PowerPC architecture, that Altivec doesn't become the most powerful SIMD solution in commodity computing.
Re: Chassis
It may be a time of research vs time to market discrepancy; IE, at the time VT was requesting bids, there were no Opteron chassis announced or available, whilst Apple may have had at 95% completion, barring an actual press release and announcement. Like, simultaneous to the release of the G5 there are no IBM PPC 970 machines, yet both companies use the same CPU.
Re: OS X
Yeah, there is a 64 bit X. It's called OS X Panther, and there's a 64 bit aware X called 10.2.7, and the libraries for Altivec have been 128bit for years now, so all 10.2.7 really added was... 64 bit pointers and memory addresses, really.
To recap: Altivec makes a big difference. Having immediately available machines makes a difference. Having a lower price point per performance per machine makes a difference (each node, including AC + networking + ram only costs about $4,727, which is $1,600 lower than an identically specced stock dual G5 with 4GB of ram!), as well as supportability of OS X vs Linux or, heaven forbid, Windows 2k... And yes, OS X for these machines are at least 64 bit enough to address 8GB of ram, and the OS has *always* been able to manipulate 128 bit data, as well as 64 bit data.
GPL Deconstructed
The boxes were purchased at the standard academic discount price from Apple. Not even a volume discount. It's said so in at least two of the articles I read about this story.
But the truth usually doesn't stand very well against the 'obviously true', does it?
-fred
Sign #11 of Slashdot overdose: You see the phrase 'moderate Republican' and you wonder if that would be a +1 or a -1.
Except VT isn't going to be running SPEC[int|fp]+[_rate]*2000, so optimizing against those benchmarks isn't sufficient.
When all is said and done, it's been shown that for many vectorizable programs, Altivec still spanks SSE/SSE2/3d!Now, and anything else offered in the commodity market. Couple dual 2GHz G5s with Altivec and IBM's XLC autovectorizing compiler, and I think VT probably does have quite a powerful machine, more so than SPEC scores can quantify.
GPL Deconstructed
...after a good ten minutes of googling, I can't actually find anything out about it.
Basically, he made it up, because he doesn't like the idea that something like this might be possible.
-fred
Sign #11 of Slashdot overdose: You see the phrase 'moderate Republican' and you wonder if that would be a +1 or a -1.
Funniest thing I've seen all day.
And I'm a Mac guy, too. I wouldn't mind wandering through that room for a while myself... though I probably would keep my pants on.
-fred
Sign #11 of Slashdot overdose: You see the phrase 'moderate Republican' and you wonder if that would be a +1 or a -1.
Truly amazing, how many of you ever thought you would live long enough to see Apple win a contract based on price?
I am typing these very words on a contract won based on price. When I was searching the market for a new laptop with all the qualities I wanted (repeat: ALL, including such factors often omitted by PC users as battery life, general robustness or silence), low-end iBook was actually the CHEAPEST option.
Only true if it can do 2 32bit ops in parallel. Which IIRC, it can't.
Hrm, this same logic should apply to medicine, cars, houses, and just about anything else that we as people have access to Original Source... yet notice how as a society and culture we tend to specialize and rely on experts?
I would be that, due to economics and division of labor, it is more productive and cost effective for VT, and many other places, to rely on a third party (IE Apple or Red Hat) to support their OS, while they themselves support, say, their business/product/venture... in this case, VT's research, while Apple supports the hardware/software. Makes sense, doesn't it? Each party does what it does best, so that the end result is more spectacular than if VT decided to, I dunno, devote resources to replicate what Apple already knows. Or at least has the resources to know.
GPL Deconstructed
No Argument, Apple hardware is nice. I just want the Apple Zealots to go away. My experience with these people has proven the typical Steve Jobs groupie to be as behaviourally stable as a feral scientologist dieting on methamphetimes & rice cookies. Hopefully Jerry Lewis will be around next year to raise money for Steve-Jobs-Groupie-Syndrome... We need a shot for these people.
That's the whole point of the PPC970's Altivec unit, isn't it?
Here, scroll down a little to see the CPU performance on RC5, and note how a dual GHz G4 is 2x as powerful as a dual 1.533GHz Athlon.
Extrapolate then to a dual 2.0GHz G5 vs a, I dunno, dual 2.0GHz Opteron... why wouldn't it be, if not 2x faster, at least as fast, if we want to be generous and assume that AMD somehow managed to figure out how to increase the performance of the Opteron over the Athlon by more than 2 (in order to take into account the fact that the G4->G5 increase is 1.0GHz, but the Athlon->Opteron increase is only 500MHz...)
GPL Deconstructed
So they reimage about 5gb worth of OS + Applications per machine, and assuming it's a decently fast FW drive at 20mb/s means about 5 minutes per machine.
So an iPod with 30 eMacs in a lab should take about the same, since an iPod clocks in at 16mb/s (I own one) but there shouldn't be nearly as many applications/software on an eMac as on a G5... so 150 minutes, or a little over two hours a night.
And at standard lab-rate of $5/hr, that's $10 to reimage the entire lab.
GPL Deconstructed
The only thing I found of interest is some BONUS apple pr0n pics:
Look at the rack on that one!
Even bigger rack!
What the G5 sees
Looking in.. notice no nose prints/drool marks on the window(yet)
Pulling "facts" out of your ass to make your "point" just makes you look like an "idiot".
I am a believer of momentum and curves.
Could you possibly use a few more quotes in your next reply or were you just confused about what you were saying. Looks like another Steve Jobs groupie puzzled by their own words.
You are half-right. You see, you are mixing terms. Yes, I agree that relying on a "expert" is a good thing, and I pointed out that anyone can become an expert.
On the other hand, the post I was replying to, insisted that access to the original CREATORS was required (eg Linus). With Mac OS, which is closed-source, that really is required. With an Open Source OS, anyone can become an expert, and the the same job instead.
That would be just fine... Out-sourcing expertice is just fine. But as I said, you don't require access to the original programmers, when the software is open source.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
I'd recommend some basic computer maintenance.
- run disk tools... have it repair everything. Do this from the Boot CD. Insert Boot CD... hold down C key while machine boots until you see Welcome To Macintosh. Once it boots, locate Disk Tools in the Utilities folder and run repair on all available partitions.
- run a current version of Norton Disk Doctor if you have it and have it auto fix all errors.
- Pony up for some RAM. 64 megs has been puny for a long time, especially for graphics. I recommend a min of 128megs for Mac OS (classic) based systems. Preferably more if you do graphics.
- Check your Virtual Memory control panel. Set it to no more than double the installed RAM.
- if you have Norton, use the defragger. I'm sure the drive's never been defragged.
- reboot the machine
Should work great after that. My old 8500 still works fine. I can copy several gigs on it in the 17 minutes or so.
Or do you just want to bitch?
The real answer is that the problems that are going to be solved with this cluster are easily parallelizable. That's the IDEA, right? 1100 machines, each running one chunk. Well, the G5, and more specifically the Altivec vector processing section of it, is SO MUCH better for processing big bites of easily parallelizable data at a time than any of the alternatives that it can run rings around any Intel or AMD machine you care to name with fewer than double the number of processors. (And in the cases of some particular kinds of calculations, it beats those, too. But you can't count on that for all your problems.)
We've seen this before a number of times... I seem to recall a gene sequencing program that was running five or six times faster on a G4 than it was on a Pentium IV of the same speed. And then there's SETI@home, which runs much faster, cycle-for-cycle, on the Mac, and doesn't even USE altivec. (Though I believe it does take advantage of the 'multiply-and-add' instruction of the PPC, which is another nice little feature.)
Altivec is an astonishingly clean and usable interface for an amazingly powerful vector processor that is, in 99% of the Macs out there, underutilized to the point that if it suddenly disappeared, most people wouldn't notice any difference at all. It's kind of a pity, really.
Basically, Intel came out with MMX (and all the later developments) in order to have a talking point on a slide presentation about their processors, about the time when competitors like AMD were starting to come forward: functionally, an awful mess, and impossibly difficult to program. (In fact, for the first few years, Intel would send programmers out to work with companies to implement MMX, because otherwise none of them would bother.
AMD came up with something that was a little less hacked together in a very short period of time, as a response to Intel. But it still wasn't pretty, at least partially because of the limitations of the archetecture, and the performance wasn't *that* much better than just doing without.
Apple (who really designed a lot of the basics themselves when it comes to Altivec, so don't think this was a Motorola invention) said, 'Hey, wow, we need something like that, in order to compete.' First they decided on a coprocessor, but that didn't fly any better with the PPC than it did with the older Macs (840av, 660av) with DSPs in them. So they sat down and came up with a really *good* spec for a set of multimedia extensions. And they've only gotten better since.
I've toyed with altivec code, and I can tell you that in one application that I wrote, one instruction (vector permute) did the work of ten or more non-altiveced instructions on four times the data per cycle. Mind you, I just did it for fun, I don't know enough about parallel computing problems to come up with anything useful... but there's some interesting stuff under the hood.
Of course, nobody is going to believe this, because as fashionable as it is to like MacOS X on slashdot these days, nobody wants to admit that, for *some* subset of problems, Mr. Jobs's reality distortion field might not be quite as much of a distortion as you might think...
-fred
Sign #11 of Slashdot overdose: You see the phrase 'moderate Republican' and you wonder if that would be a +1 or a -1.
np: Aphex Twin - On (Reload Mix) (On Remixes)
"I'm not anti-anything, I'm anti-everything, it fits better." - Sole
VATech received the grant from an 'anonymous donor' for the project.
Funny how you didn't cite a source for this, mister AC.
Oh, that's right, because there isn't one.
Mod me down and I will become more powerful than you can possibly imagine...
So... does it go without saying that you should mirror the image, somewhere we in "the masses" can see it? :)
There's a difference between the way something is done, and WHY it is done that way. The source code will tell you the first, but usually not the second. In many respects, code is NOT documentation.
Nope. Not wasting time. Unless you are actually doing something, the WindowManager is idle.
Article X: The powers not delegated... by the Constitution...are reserved...to the people
towers work just fine, most are front to back cooled, just like a rack mount
also the hardware has more room too for more heat disapation.
if you have the rack space, its not a bad idea. I imagine that the racks they are on are cheaper than "real" racks. And I know a g5 tower will be cheaper than a comperable g5 xserve, whenever such a beast exists.
-Tim
-I just work here... how am I supposed to know?
Nice example of the ad hominem argument. Nevermind the fact that the point he was making was correct. Go stick your head in a pig, or back in a pig, as the case may be.
political_news.c: warning: comparison is always true due to limited range of data type
Well, kick down then! Surely you know how to set up a mirror =]
political_news.c: warning: comparison is always true due to limited range of data type
I'm still bleeding from the adds and cookies zinging by on that page that had three poorly-produced photos.
Hundred bucks here, hundred bucks there... random hardware driver causing incompatibility with rest of cluster over here, shit, should'a gone single-source with a warranty, huh?
political_news.c: warning: comparison is always true due to limited range of data type
In a single word: altivec
Google it if you don't already know what that means.
political_news.c: warning: comparison is always true due to limited range of data type
Unless you want to look at a disgusting picture of a girl in a bathtub eating her feces. No, I'm not kidding, that's one of his links, and it's quite possibly groser than goatse.cx.
political_news.c: warning: comparison is always true due to limited range of data type
Of course, being human, my opinion is suspect.
Haw haw! Because, I am not human, and hence my opinion is inherently superior to yours!
with the g5 solution you get 4g of free ram???
that certainly isn't a fair comparision then, just a special deal comparision.
world was created 5 seconds before this post as it is.
17600 FANS!!!!!
Maybe they used screws... (Looking at the G5s the picture semms ok)
Yeah man I don't understand the big deal with G5s for this kind of application. I'm sitting here in front of my 1100-unit dual-G5 cluster at my freelance gig trying to copy a 17M file from one folder to another and it's taking over 20 nanoseconds. My Cray at home would be done with this already, and even my beowulf cluster of TRS-80s wouldn't take this long....
Or just keep half a dozen spare boxes on the side. If something goes bad, swap out the whole box, and fix it whenever.
May not be perfect, but it's still the best government we have. For now.
[notices "HAIL 64-BIT" sign taped up, tears it down]
Oh, yes, by the way, the spacecraft still in extreme danger, may not make it back, attempting risky reentry, bla bla bla bla bla bla.
We'll see you after the movie.
But that is only usefull if they use 64bit integer math. My guess is that they need theese computers for their fpu power which always have been atleast 64bit, even on 32bit processors.
ALmost all problems that require integer math, can be solved with 32bit integers, so doing 64bit integer math, will just shuffel 32 zeroes around.
Between the on-campus nuclear reactor and the supercomputer cluster, I'd keep an eye out if I were Tech's cross-state rival, University of Virginia. I'd say the Hokies are just one diabolical dean away from becoming an evil university bent on world domination. And five bucks says they start in Charlottesville.
DecafJedi
my weblog: apropos of something
When I first heard about this I thought these would be custom boards that would contain a minimum of components, SBC style that plug into a back plane and net boot, with tens of nodes per rack, etc.
Never in my wildest imagination did I think these would be the full Apple G5 desktop's, cases and all.
Those fans in the G5 switch on when the system gets hot. Imagine the noise when this cluster is pegged at 100% CPU occupancy!
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
If you believe 100mbits is too slow why not use IP over Firewire. I use it on my MacOSX server and it is screaming fast. You get 3200 mbits.
There is really nothing to do just daisy chain your mac with Firewire cables and configure the new network. On MacOSX client you will need to install IP over Firewire manually.
This could be multicast... provided the network is setup correctly. That makes it a lot easier.
Also, in this kind of setup you often keep a copy of the master image locally on the dive of each machine, on a separate slice / utility parittion... if done properly this lets you re-image in seconds or minutes,
for windows users do as follows
1. Set up 1 machine how you want it.
2. Use Ghost / Drive Image or what ever your favourite drive cloning utility is, to image all the drives in no time at all
3. Go in and change HD, Network, etc names.
4. Smile because you just did something in 4 steps that took a mac user 6
I was one of the hordes of CS majors who helped setup the super computer (grunt work is fun!). VT is using inifiniband cards w/extremely low latency copper cable (forget the name) which acheives the same bandwidth as fiber optics.
Loads of cisco catalyst switches are involved also.
"...anyone can become an expert..."
This speaks right to the point, IMHO. VT may not have the time to put someone on the source to learn it. When a problem crops up, you usually need it fixed NOW; you don't have time for some coder to sift throught the source and try to figure out whats up.
Are you familiar with ALL the source necessary to run a cluster of this size under Linux? Do you know anyone who is? Could you assemble all the coders/designers whos knowledge of the source might be necessary NOW to fix the problem thats shut down a multi-million dollar supercomputer cluster?
Just my opinion, of course.
(tig)
Ignorance and prejudice and fear
Walk hand in hand
One quote for the power consumption of a Power Mac G5 Dual Processor stock is 800 watts. Now 1100 of these - 880kW! 880kW at 120 volts is about 8000 amps. Mind you, this is not including all the glue electronics... I am glad I don't have to pay THIS electric bill!!! :-)
i could but i dont want to spend the huge amounts of money from the slashdot effect again.
Good points, all. But even card manufacturers use different chipsets. And Macs are known not only for their initial price, but their resale value, which tends to hold up MUCH better than commodity PCs.
Plus, I'm one of those scary liberal types the GOP warned you about. Our disposable society generally sickens me. Our landfills don't need any more electronic components thrown in 'em. Build it right, and it'll last. TOC and all that.
Same ol' opinion, still suspect.
(tig)
"We do not inherit the land from our ancestors"
"We borrow it from our children"
Ignorance and prejudice and fear
Walk hand in hand
I thought VA Linux was changing their name again.
--
"Outlook not so good." That magic 8-ball knows everything! I'll ask about Exchange Server next.
This gives them 1101 good computers - a kickass cluster now, and 1100 workstations later.
[100% ISO 646 Compliant]
SVM, ERGO MONSTRO.
I read the article, and it looks like it was a REAL amature decision to go with these systems.
Sure, the systems probably offered good performance for the (heavily discounted) price. However, they extremely large size of the desktop systems means that they're going to spend a lot more money on physical space to keep these systems. They also have rather high power consumption for super computer use due to all the completely useless (for a supercomputer) extras that the G5s come with (eg the high-end gaming video card). They are also likely to run into a number of problems with failing fans. Each of those G5s has 9 fans, and fans are about the least reliable part of any modern computer.
What's more, they now have 4.4 terabytes of non-ECC memory. Given the Soft Error Rate of today's memory, it's likely that they'll be getting at least one soft error every few days in this cluster. Now, maybe they'll get lucky and the error won't affect things, but they are leaving a LOT to chance here! This looks like a MAJOR oversight if you ask me. Ohh, and before someone starts talking about that "Deja Vu" technology, it does absolutely nothing to detect or correct memory soft errors.
Alright, here is a mirror:
J PG
http://www.paullockaby.com/pub/pictures/IMG_0099.
I go to a lot of auctions. A lot of surplus computer equipment appears at said auctions. Lately I've been to a few auctions where all, and I mean ALL, of the 'PC' equipment gets bought up. Then I get to pick and choose what Apple equipment I want, because nobody is gonna bid on it. I would estimate that in my personal experience 80% of the old Apple equipment ends up in the landfill, and 20% of the old PC equipment. Just my personal observation based on the recent market, though.
I love it, personally, because I've been getting into old Apple gear, running NetBSD on some of it. I got a whole pallet of Power Mac 75xx boxes for a dollar a few months ago, along with a pallet of monitors and keyboards for the same price. Lots of them had 64 meg SIMMs in them, so now I've got one hell of a 7500 box.
It's gotta hurt at the recycling center, because there's a hell of a lot of plastic in all those Macs. If all the machines hitting the bottom of the dumpster from the auction place I frequent are even being recycled....
A Good Intro to NetBS
Err, please TRY to find a 32-bit FPU in ANY processor. Trust me, it won't happen. Intel's old 8087 FPU could handle 80-bit data. This chip was used alongside an 8-bit processor.
AMD has been making Intel-compatible processors under license from Intel since the 8088 processor.
2+ million BTUs of cooling capacity using Liebert's extreme density cooling. This system uses rack-mounted heat exchangers with R-134A refrigerant and an overhead chiller unit
(R)Traditional AC systems would have resulted in a wind velocity of over 60 mph under the raised floor
Holy hell, that is quite mad.
I am NaN
Don't put the spare boxes on the side, put them to work in the cluster! If your numbers say you need 100 machines to get the throughput you need, you buy 110, and if two or three die, you still have plenty of horsepower. You yank the dead machine out and repair it, everything else keeps chugging along, and then when you're ready to start the next job, you plug the repaired machine back in.
Very good points indeed. However, even then, lets say that Apple has a 20% (optimistic) and PCs have an 80% installed base of overall desktops. .8 * .2 = .16 or 16% overall and are PCs .2 * .8 = .16 (funny how that works) or 16% overall and are Macs.
:-)
In other words, say 32% of desktops make it into landfills (more actually do; this article claims only about 14% are recycled/donated), half of them PCs, half Macs.
And as Macs become more popular, which I think is inevitable, and begin using more standard parts, which they already do, I think you'll see less and less of them ending up in landfills.
"... I got a whole pallet of Power Mac 75xx boxes for a dollar a few months ago..."
Where? And do you want to unload any? Man, imagine a beowulf cluster...
(tig)
"We do not inherit the land from our ancestors"
"We borrow it from our children"
Ignorance and prejudice and fear
Walk hand in hand
At our school, we have developed an ad hoc solution with ASR (Apple Software Restore), it comes in 10.2 (client): try "man asr" on the command line.
ASR is damned fast ! We restore from scratch 2Gb system disks in 5 minutes. The image file is stored on a standard AppleShare server (100 Mbit/s switch). The speed is due to ASR ability to restore using block level disk access instead of file level, and from compressed images so less data has to travel on the network.
Each client has 2 partitions: one of them has a minimal system so we can boot on it when we want to restore the "main" partition. Everything is done automatically with ssh.
All us Geeks afflicted with interminable penury should merge our truly meager resources and nearly infinite technical skills to build a massive supercomputing cluster!
Hell with the unemployment rate at 30% for techies here in Silicon Valley we could use something to do!
I killed da wabbit -Elmer Fudd
64-bits? The vector unit (Altivec) on a Mac has been 128-bits for a long time.
Remember, a lot of supercomputing is vector-based....heck, even Crays were vector machines. Therefore the 128-bit Altivec is a perfect choice for many types of calculations.
-psy
We use it in our lab here at work and yes it is very very evil
-ravan_a
yeah, but thats a characteristic of ALL fpu's. there isnt much you can do to remedy that, and the problem is prevalent in ALL computers, not just apples', so I dont see how this situation is analogus to the intel fdiv bug. but on the whole, that IS a fairly good explanation of the problem.
one world | many people
http://www.chaosmint.com/mac/vt-supercomputer-2/
"Project started back in February; secret with Dell because of the pricing issues; dealt with vendors individually because bidding wars do not drive the prices down in this case"
BLOW ME DELL!!!!!
HA HA HA HA HA. Jobs must really love reading that.
Just blogged it. It's a quality quote.
This is exactly what I did with my G4 Cube, and we all know how that went over.
"You know why you do not see me styling wit my homies? Because I have no homies!!" -Mojo Jojo
According to you, then, Apple, when it sells direct to a customer through the Apple store, is making less than a 15% margin on its machines. Because that's the typical maximum discount they give, except in cases of clearing up back stock to make way for new models. Even then, I suspect that Apple's margins are 'healthy' enough that they're not selling below cost, but I *know* their margins are more than 15%. It's one of the things people like you complain about, that Apple makes big margins on their machines, 'overcharging' their poor, hapless consumers.
Basically, grow up, There are plenty of nasty, awful things that Apple has done in the past, and a number of them they still do. When you make stupid accusations with no basis in fact, just because it's Apple and you hate them, you make it harder to have a reasonable dialog about what's really going on.
-fred
Sign #11 of Slashdot overdose: You see the phrase 'moderate Republican' and you wonder if that would be a +1 or a -1.
But they didn't become popular, didn't become a common replacement for Intel chips in the real world, until the K5 and K6, and the 6x86 (from whoever made those), etc. Before that, Intel didn't have to worry about them; they weren't much of a threat.
-fred
Sign #11 of Slashdot overdose: You see the phrase 'moderate Republican' and you wonder if that would be a +1 or a -1.
probobly AC ducts. Cold air falls, remember.
That cluster is probably capable of compiling Gentoo from source in only 2 days!
But it doesn't have to be that way. It is really a user-interface issue.
It is clear that what the user of a calculator program wants when they enter 3.083-3.014 is the mathematically correct answer, not the closest approximation to the difference between closest_floating_pt(3.083) and closest_floating_pt(3.014). So the program should do the extra work to make it so whenever possible. In most cases, it is possible.
And the original poster was talking about the fact that the rounding error (from 3.083-3.014) showed up in the paper-tape printout. That would seem to be a bug, pure and simple.
Yes, and anyone CAN become an expert, provided they are willing to do so.
Everyone seems to be focusing on the wrong part of my statements.
Yes, it's a lot of work to become an expert, however, my point is that, with closed-source, NOBODY can become an expert at all, meaning access to the original programmers is absolutely required. With open source, it's not necessary, since ANYONE CAN BECOME AN EXPERT. That doesn't mean everyone that uses the system should become an expert, nor does it mean that any 13-year-old that knows a programming language will be an expert. It means just what I said, and only what I said, not what everyone wants to read into.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
Fine. Let's say they out-source support to RedHat, or somebody else. It doesn't matter at all. It is completely besides the point. The issue was direct access to the original developers, while that is simply not necessary in Open Source software.
That's all there is to it. Nothing else. No more than that.
If your coder needs to "sift through the source", then he's not much of an EXPERT now is he? Of course not.
No I'm not, nor do I need to be. I never said that every programmer can be considered an expert, just that the source is there for anyone with adequate skill to learn. This is not an option with closed-source software at all.
If I was the one responsible for this, you can bet I would. You can also bet that it would be far less expensive, and more reliable than the support contract they have with Apple.
Perhaps a single or multiple in-house developers. Perhaps RedHat or some other Linux co. It doesn't matter. The very fact that it is Open Source gives you infinitely more options.
Slashdot gets worse every day... Pipedot: News for nerds, without the corporate slant
A lot of scientific computing uses double precision floats, which altivec doesn't do. The G5 has enough double precision power to handle it quite capably, but it doesn't use altivec.
When someone might yell at me, it has to be OpenBSD.
What operating system will they be running on these things?This is not a troll, I'm geniunely concerned to know if they will be wasting these lovely resources on OSX ? (Which is not truly 64bit)
Electronic Music Made Using Linux http://soundcloud.com/polyp
"Re: Processors"
...? That's a non issue. The OS doesn't get involved with data in that way. The OS might provide libraries to help, but they don't do anything you couldn't do in assembly by yourself. You can use any (supported, non-privileged) instructions you want in a program. On OS X, Linux, Windows, whatever.
Altivec can't do double precision floats. The G5 has plenty of double precision power, but it's not in altivec. Dunno what they're going to be calculating, that may or may not not hurt.
"Re: Chassis"
Opteron cases have been available for many months. As have the Opterons, in point of fact. Months before this cluster got delivered, you could go to several stores in my city and buy a complete dual processor opteron system off the shelf, and OEMs have been selling workstation and rackmount systems.
They were probably taking bids on this a long time ago, and if it was before Opteron systems were available, it was also certainly before Apple had more than engineering samples of the G5's.
"Re: OS X"
It's not 64-bit, but I believe even the 32-bit versions of the OS could handle more than 4 gigs.
"and the OS has *always* been able to manipulate 128 bit data, as well as 64 bit data."
When someone might yell at me, it has to be OpenBSD.
Just curious, one thing I haven't been fully able to understand is what they mean by 100 nodes? Since the G5s they installed are dual processor, does that mean 1100 Computers/2200 Processors, or 550 Computers/1100 Processors?
"Reality is merely an illusion, albeit a very persistent one " -Albert Einstein
Again, explanation is not an excuse. My point is that, until such time as the user executes a mathematical operation whose result becomes a repeating decimal (divide, not add/subtract/multiply/exponentiate), it's mathematically incorrect to employ the FPU in the first place.
I'm well aware the original poster knew little or nothing about floating point math...my question is, where do we get off talking down to him when *ours* are the incorrect results?
--Dan
Actually, we didn't spend more money for physical space to keep these systems. The whole thing fits into about 3,000 square feet. We have over 10,000 square feet of machine room that we built back in 1989 (sized to hold an IBM 3084 and an IBM 3090-300J plus all the disk drives that went with them). Good thing our director at the time spec'ed out "all the floor space we're likely to ever need".....
Nobody cares about the noise. It's in a machine room, not office space. All the offices are on the other side of the atrium.
And I doubt that it can be much noisier than the printer room was when there were 2 IBM3800 printers going full-tilt (400 pages a minute each. No, that's not a typo. Four Hundred. Feed it a box of 3,200 pages of fanfold, and you *might* have time to go pee before it ran out again).
Nope.
The *management* net is gigabit ethernet.
The actual clustering will be done over 10-gigabit Infiniband. (For a good time, figure out what the maximum bandwidth of a PCI-X slot is, and compare..)
There are no announced dual-CPU G5 blades.
And what makes you think that a dual-CPU G5 blade would cost significantly less than a tower G5? (Hint - what's the price tag on a dual-CPU 1U G4 rackmount?)
Tower or blade, the budget covered a bit over 2,000 processors, memory, networking, and all the rest of it. When you consider that some 20% of the budget was *networking* gear (Cisco, Infiniband, all that stuff), the blade/tower distinction didn't matter much.
About all going with blades would have done is made the footprint smaller - and that would have just made the cooling issues even worse.....
Well.. the processor itself is 70W. There's 2 of them, that's 140W. Then there's memory (4G of it), a 180G disk drive, all the support chips (yeah, things like disk controller chips and PCI driver chipsets and memory/cache interface chips are nice to have if you actually want to do any computing), the PCI card for the Infiniband... and pretty soon that 750W power supply isn't as much overkill as you thought.
Nobody cares about the noise. It's in a machine room, not office space.
Yeah, I know. I've worked in many a computer room, including huge rooms full of DEC big iron.
I wonder if they'll drown out the air con.
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
I weep for the state of reading comprehension in this country.
I am a believer of momentum and curves.
It's just VT. It's not VA Tech, VTech, it's just plain VT damnit!
http://www.maximum-cars.com - My little hobbie.
Retail cost for 1100 dual G5's (with 4GB of memory) is $5.85 million alone. Discount that by 11% and you get $5.2 million for the G5's all on their own. Then there's the inifiniband cards, which will run you easily $1000 a port, so that's another $1.1 million or more. Add in all the extras for cabling, the racks, the fans and ducting systems, etc. etc., and you'll quickly see that having a complete system cost of only $5.2 million IS heavily discounted.
As for the Deja Vu thing, I wouldn't have even mentioned it except that the last time someone pointed out this major short-comings of Apple's G5 for this application a bunch of people started screaming that this "Deja Vu" would save the day. Deja Vu is run at the software level on top of the operating system. All it can do is look at what data is there. There's absolutely NO WAY to for it to check to see if the data is correct unless they are running every single calculation twice at the same time and comparing results, and if you're doing that, you're just wasting your time and money when you could have bought a 550 node cluster with ECC memory and got the same result.
Yes, but altivec can't make 128 bit integer math. Well, i don't even think it can make 64bit integer math. Afair it can only du 8,16 and 32bit math.
Martin
>Having dual 2ghz processors is not as fast as having one 4ghz
Right. It's much faster than having one 4GHz processor.
The kernel runs as a 64-bit process while the majority of other OS components remain 32-bit processes. Applications are run as 32-bit processes unless that application can specifically benefit from 64-bit memory addressing
Sorry, but according to pricewatch, dual opteron motherboards start around $400 and the lowend SMP capable Opteron starts at about $250. So you're going to be looking at least $1000 for the cheapest Opteron machine that is somewhat the same sort of configuration as these G5s. Unfortunately, the G5 would be able to kick the snot out of such a machine. Use the 246 Opterons and now you're talking about a machine in the $2000-$2500 range. That's the type of machine these guys would want.
From looking at these pictures, all I kept thinking was: what a bunch of wasted space and under used equipment. As a compute node, what use is wi-fi, bluetooth, USB, audio subsystem, etc. If they have video cards, is the graphics processor & memory going to be used as some sort of coprocessor? If not, that's a waste too.
Apple was stupid to not create a G5 version of their rack mountable XServe machine. It would be a better fit for what these people want to do.
the good ground has been paved over by suicidal maniacs
Look at the heat sinks in the g5's. I doubt they'll be getting that into a 1U anytime soon. I'm betting on a 2U Xserve for the g5s.
Apple just probably didn't want it all rolled out at the same time. They'd have nothing to show off for a couple months.
Mod point free since 2001
The Apple System Restore tool can erase a disk and copy a configuration onto it from a compressed disk image. This is what Apple uses for setting up machines on the production line.
"man ASR' ion Terminal.app to get started.
-Mark
Same image canvas size, but lower image file size:
http://technojunkie.org/berniec/VTpr0n/
Total images filesize before were around 61MB; now down to 17.5MB.
I don't doubt that it would be a while before 1U G5s show up, but a 2U server would be better than just stacking deskside towers on a bunch of shelves. Good point about not having something new for a while. Maybe the university didn't want to wait.
It would be cool if instead of announcing a machine and then waiting a couple months for it to show up, Apple could have gotten the school to wait, secretly install a slew of rackmount G5s, and then have Jobs and the univ big wigs unveil them at a ceremony the next day in the server room.
the good ground has been paved over by suicidal maniacs
For those interested in a lot of FAQs about the project, I'd recommend a visit to Your Mac Life, the weekly net radio program about Macs. Their Sep 10 show has a half-hour interview with Hassan Aref, the Dean in charge of the school and the project. Much useful info.
Having seen this cluster in person last week (I work right across the street), I can say that it is very awe inspiring to walk through. The woman that gave my friends and I the tour mentioned all of the logistical problems that were overcome to house that many computers in one room. I guess I am naive but I had never considered how fast the temperature would increase in that environment. She said that if the AC were to fail that the temperature would increase to around 150F in about 10 minutes. That is incredible. Because of the an extra phase of power was brought into the building to just for the AC units. There is a room (larger than my computer room at home) just to house the UPS batteries. And the UPS is designed to run only long enough for the diesel generator to kick in. The rows of computers were designed with hot and cold isles to help with air circulation. One thing that I did not expect to see was out of the box G5s sitting on the shelves. I thought that they were just using the processors. The woman giving the tour did not know the answers to all of the technical questions that we started asking about file storage, back ups, process management, but it was still an experience that I will remember for the rest of my life. It also puts the ~24 node Beowulf cluster that I have been working on to shame. :)
I couldn't tell it was on til I put my ear ON the case.
Curious. Was it being placed under stress for long enough to fire up some fans?
I imagine in the showroom, people will be very impressed with the noise, but when they get it home and push it will some games, 3D rendering or 2D filters, it might suddenly be a lot louder than the user bargained for.
BTW, I'm not trying to put shit on the G5, I'd love one. I'd also love to know how it sounds when the fans are going hard.
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
for all those G5s, or signed them up for .Mac?
Of course the reason to use 10 gigabit here isn't throughput, it's latency. The individual messages take roughly one-tenth the time from start of send to completion of receive than they would with a 1 gigabit link.
#naabhaprzrag, #sverubfr-000, #agi-fcbafberq, negvpyr[pynff*=' negvpyr-ary-'] { qvfcynl: abar !vzcbegnag; }