ackthpt writes "Code named Red Storm, Cray and Sandia National Laboratories (US Dept. of Energy) to build a 100 Teraflop super computer employing AMD's Opteron (Hammer) processors. Alluded to in the WSJ (non-free-as-in-beer subscription required), also in Infoworld, and Reuters."
makes you wonder.. what if the transformers was based on real events and they went into hiding in the 80's and now are pusing the cpu development to replace their lost information on cpu's to produce super ai to get themselfs off this planet.
makes you wonder if apple is formed by some sort of organic-hippie-never-to-be-sold transformers, beos is obviously the beast-transformers.
what about amiga? and where does linux end up in this scheme.
old dosbots weren't obviously too good at doing two things at once.
-- world was created 5 seconds before this post as it is.
"what if the transformers was based on real events...and where does linux end up in this scheme."
The Constructicons were probably the Linux bots. They were small parts a of a giant robot called Devastator that was very good at the task at hand. (usually de-construction.) Devastator also had a limited vocabulary so he wasn't the most social. Few Decepticons knew how to give him the right commands to get him to do what he wanted.
Ok I'm not going to use any words that even resemble beowulf or cluster in this post anywhere!
Hopefully AMD will be able to produce a quality product that far down the road. They've been bleeding money at an alarming rate and some say they're focusing too much on marketing instead of engineering- much like 3dfx did back in the day. Let's hope this pans out- my framerate could definitely be more killer.
The mighty have fallen
by
zmalone
·
· Score: 5, Informative
This sort of thing must just be braking all the classic Crayons hearts. I mean, people were getting upset when Cray started building the T3 series Alpha based stuff, nowadays they are cooperating with Dell and making AMD based clusters. At least they have a new vector machine coming out soon.
Re:The mighty have fallen
by
ackthpt
·
· Score: 2, Funny
Relax, they could always overclock them and use liquid nitrogen cooling.;-)
You can't imagine how this deal has to leave Intel smarting.
--
A feeling of having made the same mistake before: Deja Foobar
Re:The mighty have fallen
by
Anonymous Coward
·
· Score: 0
They know HP/Compuqe haven't got the means to produce quality Alphas. Bet-hedging is all this is.
Re:The mighty have fallen
by
Anonymous Coward
·
· Score: 0
Crayons? I thought those were things that you drew pictures with???
Re:The mighty have fallen
by
anzha
·
· Score: 4, Interesting
Yes...and no. What we have been upset by is that people have been trying to shoehorn in all problem sets to MPPs and clusters. There are problems which do so, and do so well.
HOWEVER! Not all do by any stretch. Certain problems map well onto certain architectures.
The second reason is that quite frankly, clusters are boring. Rack, after rack of parts I can buy at Fry's or as a workstation just doesn't have much interest for us. I mean, where's the excitement in thousands of PCs...It's kewl for about 30 seconds and then you have to deal with teh headaches of keeping it up and running...
I'd love to have dozens of interesting architectures running around, not just vector, cluster, and MPP. If five of them could be spun out of slashdot - yeah, right - or anywhere, then we'd be very happy campers.
-- Do you know why the road less traveled by is littered with the bones of the unwary?
Re:The mighty have fallen
by
dubiousmike
·
· Score: 3, Funny
This sort of thing must just be braking all the classic Crayons hearts.
Never thought I'd ever be linking to the Enquirer. I feel dirty.
Re:The mighty have fallen
by
Anonymous Coward
·
· Score: 1, Insightful
True true true.....
Different systems architectures are suitable for some and not suitable for other kinds of computing problems. It bothers me when people (particularly on Slashdot) say we need to run a Linux cluster for everything. There is a place for every kind of solution. You don't see people throwing away all their big iron! Why? For some requirements (reliability, scalability, I/O throughput) they are the best solution to the problem.
The place I work at uses pretty large (about 3500 nodes) of Linux for Seismic processing. Works brilliantly. However, for some other applications we run, a cluster is bloody pointless because the software can't take advantage of the parallel processing....
Re:The mighty have fallen
by
Anonymous Coward
·
· Score: 0
Indeed. In a way, your post reminds me of the guys who think they can/should use Linux for every embedded application, when the same job can be done with a 1-5$ microcontroller.
To any readers who might think "Linux everywhere": you will find an endless fountain of joy by working on microcontroller projects and coding in assembly! Start today!
Re:The mighty have fallen
by
Anonymous Coward
·
· Score: 0
Yea, that would be an endless fountain of joy for you, wouldn't it.
Re:The mighty have fallen
by
Anonymous Coward
·
· Score: 0
Lol, yeah, I'm sure Intel is just crying, considering they remain profitable while AMD contiues to lose more and more each quarter.
AMD is a passing fad. Are they even talking about 300mm wafers yet? 90nm? I haven't heard any talk of it...
Oh wait, forgot, this is/. AMD can do no wrong.
I retract my comment.
Re:The mighty have fallen
by
SpaceJunkie
·
· Score: 1
Go on then- whats your *original* idea for a supercomputer architecture/technology? Is it feasable, viable or usable- could *you* spec it enough to build it?
Who would write software for such a radical architecture? "port linux" I hear you shout, as much as I love Linux-how much would you have to do to adapt it to such an alien original architecture? How much would have to be spent in terms of time, resources and money on full R&D, porting, hardware, manufacture etc... I admit it would be interesting to see new stuff- but it aint that easy. And although clusters are boring, they seem to work and at relatively low cost. Its probably not great slashdot news, but if it works - dont fix it....
It's more than that. Networks of workstations are good at solving problems that are very coarse grained. For fine grained apps, they are poor because of the latencies of the communication medium (100Mb Ethernet has very high latency) and also by the relatively low bandwidth (100Mb Ethernet is about 10MB/s depending on manufacturer or cards, your network, etc). Latency critical apps usually have many small transfers from node to node (and are also finely grained usually) and a Beowulf cluster is not good at those types of things, in general.
For fine-grained apps, you usually have to go to a 'big iron' type machine -- ones with specialized interconnects or shared memory or hybrids.
Cool.. but
by
Anonymous Coward
·
· Score: 5, Funny
I wonder if it will be as fast as the computer William Shatner and Priceline.com use?
I seriously doubt it would be that fast. Even though, I am just hoping it is hi-tech enough to make the cool bepp and boop sounds like the Priceline Supercomputer does in the commercials. Does anyone know how many Teraflops teh Priceline Supercomputer is? Or is it faster than we can even measure?
--
If you had nuts on your chin, would they be chin nuts?
Just imagine...
by
Anonymous Coward
·
· Score: 5, Funny
...building a supercomputer as a cluster of commodity off-the-shelf personal computers, interconnected with a local area network technology like Ethernet, and running programs written for parallel processing out of those!
Re:Just imagine...
by
Anonymous Coward
·
· Score: 0
...building a supercomputer as a cluster of commodity off-the-shelf personal computers, interconnected with a local area network technology like Ethernet, and running programs written for parallel processing out of those!
Sounds like Google?
Re:Just imagine...
by
Anonymous Coward
·
· Score: 0
. . . how describing a beowolf cluster gets modded higher than simply yelleing out beowolf cluster.
I was wondering when this post would appear!
Linux aside, this cray sounds like a cluster in-of itself. I wonder what the difference is between a super-computer and a cluster? The case?
UT (and every other game out there right now) assigns itself (its processer affinity) to the first CPU available. It makes absolutely no use of SMP or parallel execution whatsoever.
So it'll run on CPU 0, and the other 90 bazillion will sit idle.
Interestingly enough, all the gamer kiddies saving up for a shiney new Prescott based P4 with hyper-threading will see no advantage either, for the same reason.
AFAIK, Doom III will actually make use of concurrently executing threads, and there's rumor of a new UT2k3 exe that will, as well.
Programming for parallel CPU's is a whole new ballgame, and the rules are still being written.
--
I don't need no instructions to know how to rock!!!!
Quake3 and its offshoots (mostly) supported smp, good for online servers but for your homebox, the video card is what it needs to pump out the FPS.
I personally went from a dual box to a single cpu (price/ghz), and I miss how smooth windows was on a dual box. Any dialog box, or hardware (floppy/cdrom) or network share can pause your system. Hyperthreading will not increase your speed on games, but it will make your whole (win) OS smoother, which is VERY noticable.
UT...makes absolutely no use of SMP or parallel execution whatsoever.
Just tried it under OS X on a dual 1GHz and top shows the CPU spiking at 112%. Something's letting it have part of the second processor (or maybe those guys at Westlake are really, really efficient programmers).
If you're running on an SMP system, Q3 will use one thread for game calculations and another for the graphics system.
As you implied, not all Q3-engine games use threads. Wolfenstein, for example, doesn't (ugh!). Still I think SMP would be useful in this scenario: on linux when playing wolfenstein, the game takes up a whole processor which leaves X another processor to play with.
Programming for parallel CPU's is a whole new ballgame, and the rules are still being written
Where did you get that idea? Have you seen an operating system recently? Or a database? Or a web-server?
If you meant to say that programming multi-threaded games is a whole new ball-game, then you're wrong there too. A computer game is a simply a soft real-time system, and that'a a pretty well understood application domain.
If, on the other hand, you meant to say that most games programmers don't know how to write effective efficient multi-threaded apps, then I might believe you. They can, however, learn from the engineers that have been doing this kind of stuff for years.
Re:UT
by
Anonymous Coward
·
· Score: 0
Has anyone ever managed to even start a Q3 based game in SMP mode? I never have, neither in Linux nor in WinXP. Instant crash/freeze. This is on a dual Athlon MP. Q3A, RTCW.
Re:UT
by
Anonymous Coward
·
· Score: 0
That makes about as much sense as sticking your dick into a pickle slicer.
Games aren't written to the hardware level. You see, there is something called an OS, which is an abstraction layer between the program and the hardware. If the OS supports MP in whatever function is called, the calling program will be taking full advantage of any number of CPU's that you have in your system.
Might want to check out what Cray and Sandia
by
anzha
·
· Score: 5, Informative
Cray and Sandia say it is a 40 tera*OP* system, not a 100 teraflop one. See what Cray says here and what Sandia says here
The really interesting thing is not the processor, but rather the interconnect which seems to be very similar to the torus used in the T3E.
I've been wondering if it was time to buy AMD's stock yet. I was going to wait until they were closer to the release of their new 64 bit chips, but I might have to jump on the hypewagon a little earlier.
Its "free-as-in-speech" vs "free-as-in-beer". The first takes free to mean a basic right, while the second just refers to something that has no cost.
Re:OT but...
by
Anonymous Coward
·
· Score: 0
The free-as-in-beer term doesn't make a lot of sense if you ask me. Beer isn't free, the recipe (source code) doesn't come with it when you buy it. I suppose the buzz is free with the beer, but why else would you drink it? It seems to me that free-as-in-pee makes a lot more sense:)
"Free as in beer" means free in the same sense as "FREE BEER!!!!" signs, i.e. it doesn't cost you anything. Of course, such signs often lie, but that's not the point. Free as in speech means you can do basically whatever you want with it, a la GNU GPL, which says essentially:
"you can copy this program as much as you want, edit source code, redistribute it, reverse-engineer it, sell it, etc, but anything you give/sell to other people based on this must be licensed under the GPL, except that you can put warrantees on it, too."
You might have to pay for a GPLed program (eg RedHat Linux), but you can go and give free-as-in-beer (or even not free-as-in-beer) copies to all your friends, as long as you give them the GPL too. Free-as-in-speech is also known as copyleft.
-- I hereby place the above post in the public domain.
Return of Vector Processing
by
drhairston
·
· Score: 4, Interesting
"Cray Chairman and CEO Jim Rottsolk said Red Storm reflects Cray's strategy to deliver high-efficiency, high bandwidth supercomputer systems. "Red Storm embodies the same design philosophy as our new Cray X1(TM) vector-based product in a highly cost-effective superscalar architecture and will be a key initiative for Cray."
Ah, I remember my days on the venerable Cray Y-MP, optimizing my programs for vector processing. I am unsure how Cray has managed to make a combined parallel-vector machine like the Y-MP out of PC chips provided by AMD, but I do not envy the programmers who must now begin the task of vector-optimizing their code to take advantage of this beast.
I had hoped that this idea died with Cray. Apparently not.
Dr. Hairston seems to be the superintendent of public schools in Baltimore, not of community colleges like he claims.
Re:Return of Vector Processing
by
Boone^
·
· Score: 4, Interesting
Japan's Earth Simulator is created from NEC SX-6 Vector processors, proving that not only hasn't the "idea" died, but it's doing very well thank you very much. Besides, who said that a machine made with AMD Opteron processors has anything to do with vectors?
Re:Return of Vector Processing
by
certron
·
· Score: 1
"Besides, who said that a machine made with AMD Opteron processors has anything to do with vectors?"
Re:Return of Vector Processing
by
Anonymous Coward
·
· Score: 1, Interesting
I am unsure how Cray has managed to make a combined parallel-vector machine like the Y-MP
The X1 looks more like the T3 (composed of Alpha chips) than the Y-MP. That 3-D mesh, highspeed interconnect approach to latching all the processors together is what Red Storm and X1 share. [ It isn't the actually same interconnection network. Just similar in design philosophy.]
I'm curious as to what OS they are planning to run. Linux? A port of UNICOS (cray unix)?
do not envy the programmers who must now begin the task of vector-optimizing their code to take advantage of this beast.
The X1 is for all ready tuned vector code. In order to fully leverage Red Storm the programmer is likely going to have to customize the code (for for a MPP machine). More than likely Red Storm will run code that has been tweaked for ASCI Red even faster. That's one likely primary objective.
Heating issues?
by
Spazholio
·
· Score: 5, Interesting
Now, I know that my little ol' Athlon runs hot as a mother, so I can't imagine the cooling necessary to keep this baby running at an optimal temperature. Last I heard (and I could be mistaken), Crays were cooled by bring submerged in liquid nitrogen, and more recently with some sort of liquid plasma cooling (don't ask me, I have NO idea how something like that would work). Does anyone have any information on how they're going to keep this thing from incinerating itself the moment it's turned on?
Simple, Liquid nitrogen(with a lot more than the normal Athlons would need, a LOT more).
I've heard that the Opteron will not run as fast as the T-bird did. It will run more like the new T-bred B(sounds the same, don't it?).
Re:Heating issues?
by
stratjakt
·
· Score: 2, Funny
I'd imagine some sort of phase-change system (same as an air conditioning unit uses).
For 90 mill, I doubt you'd see a bay bus with a bunch of Delta 80mm fans..
--
I don't need no instructions to know how to rock!!!!
Re:Heating issues?
by
ketamine-bp
·
· Score: 5, Informative
for the AMD CPU's, they use Liquid N2 with phase-change system (as described in prev. posts).
for the part of 'plasma cooling', it's similar (in non-scientific term) to laser cooling, which relates to absorbing momentum. (you may want to find some information on the plasma section of http://www.arxiv.org/ if you want to know 'bout that.)
Re:Heating issues?
by
fgodfrey
·
· Score: 5, Informative
Crays have never been cooled by liquid nitrogen or any other super-cold liquid. The explanation for this that was given to me is that the super-cooling causes too great a temperature change to keep the parts reliable as they have to be warmed up every time they are turned on.
The Cray 1 and Cray X-MP were cooled by a freon-cooled cold plate. The Y-MP, C90, T3D, and T3E have a chilled liquid called Flourinert (some derivative of an artificial blood plasma, I believe, which is made by 3M) cirulating through a cold plate between boards. The Cray 2 and Cray T90 were cooled by being immersed in a vat of Flourinert. The Y-MP/EL, J90, and SV1 are all air cooled. The X1 (aka SV2) is cooled by spraying Flourinert onto the chips.
I believe, though I'm not 100% certain, that this system will be air cooled, presumably by lots of big fans:)
-- Go Badgers! -- #include "std/disclaimer.h"
Re:Heating issues?
by
Dukebytes
·
· Score: 5, Informative
Good question - can't imagine the heat that it would generate.
Found this on the web -
"Keeping it cool - The development of Cray cooling technology allowed each technology
generation to increase the circuit board density.
"Someone (perhaps Gary
Smaby? I truly don't remember) once said that Cray Research was primarily a refrigerator company."
Cray-1: Single sided boards clamped to copper plates placed in aluminium
racks that had cooling fluid in tubes.
XMP: Double side sandwich boards clamped to twin copper plates placed in
aluminium racks which had cooling fluid in tubes.
Cray-2,3,4: Immersion cooling. The CPU and memory boards sat in a bath of
electrically inert cooling fluid.
YMP, C90, T3d LC, T3e MC: Double-sided circuit boards clamped to hollow
aluminium boards in which the cooling fluid circulated.
El,J90,T3eAC,SV-1: Blown air cooling.
T90: Immersion cooling. The CPU and memory boards sat in a bath of
electrically inert cooling fluid."
I was up close and personal with an older Cray once - it was basically a tower of CPUs and very very short cables - and a whole bunch of cooling "units" srounding it. They were built into something like bench seats - the tech that was showing us around said they put those in so the guys could sit down and rest once in a while in peace:).
Duke
--
FreeBSD: Nothing runs like a daemon with a pitch fork.
Re:Heating issues?
by
subgeek
·
· Score: 4, Informative
one thing the opteron will have going for it is a heat spreader. intel uses this technique on the p4. it allows better heat transference and a larger surface area so the temperature is more spread out. one of the reasons athlons are hot is because a lot of work is done in a small area.
another is silicon-on-insulator (soi) where a layer of glass insulates each layer of silicon from the next. this allows lower voltages to be used because there is less interferance.
but even after all of that the opteron might run cooler than your athlon, but probably still run hotter than an intel chip at the same clock speed. check out the article on www.tomshardware.com (to lazy to look up the link) and take a look at the basic opteron heat sink requirements. must have copper contact area. bolts to mobo, not to socket (probably to cover for the added weight of the copper).
but that's just the regular opteron set-up.
wait, i forgot, amd fire comments are funny, right? maybe you didn't want discussion. too late now.
Re:Heating issues?
by
Anonymous Coward
·
· Score: 0
Now, I know that my little ol' Athlon runs hot as a mother
Yeah, your mother is pretty hot
Re:Heating issues?
by
Anonymous Coward
·
· Score: 1, Insightful
layer of glass
Actually, nowadays they use Silicon dioxide. They used to use sapphire, but the cost was too prohibitive, so the idea was temporarily scrapped. Silicon dioxide is much cheaper than sapphire, and is typically what they use now.:)
Re:Heating issues?
by
yknott
·
· Score: 4, Interesting
Flourinert is a whole bunch of fun. You can actually buy it for about 200 bucks a gallon. A few people have tried overclocking their computers by immersing their computers in the stuff. Check it out here and here
Re:Heating issues?
by
stak
·
· Score: 2, Interesting
>some derivative of an artificial blood plasma, I believe, which is made by 3M)
So if you were to shoot it, it would bleed?
Re:Heating issues?
by
Anonymous Coward
·
· Score: 0
"...a layer of glass insulates each layer of silicon from the next"
Umm, no. There's only ever one layer of silicon. The rest is metal.
Re:Heating issues?
by
Hoser+McMoose
·
· Score: 3, Informative
Heat spreaders aren't really that great when it comes to actually making the chip run cooler, all they do is exactly what their name suggests, they spread heat.
The reason why this is a good thing is that modern processors do not generate their heat evenly, so you might have the bottom left 25% of the chip generating 50W, while the other 75% of the chip is mostly idle, only generating a few watts here or there. This is a bad thing for a variety of reasons, beyond the obvious fact that one area has to be cooled more than others. When it comes to actually dissipating the heat though, what you want is to get this heat into the furthest corners of the heatsink as quickly as possible. This is why copper inlays are a good thing for heatsinks (they spread the heat away from the hot core towards the edges of the heatsink faster than aluminium does), however even aluminium conducts heat a lot better than the P4's heat spreaders do.
This sort of thing is, presumably, less of a problem for the Athlon as compared to the P4 simply due to die sizes. The P4 is a MUCH larger die (nearly twice the size) when compared to the Athlon. So, while the distance between the hottest and coldest points in an Athlon are always going to be quite small, they can be relatively large on a P4.
As for fixing the heatsink to the motherboard instead of the socket, this is a damn good thing if you ask me. AMD originally thought that this would be how things would be done with the Athlon, and specified 4 holes around the socket to do just such a thing. You can even buy heatsinks that attach this way (my Athlon is cooled by just such a heatsink, an Alpha PAL 8045), however the vast majority of heatsink designers just flat out ignored these holes and just attached to the heatsink. Intel was a bit more forcefull, and absultely required ALL heatsinks to attached either to the motherboard or even to bolt to the case itself (the first P4s, those that came in a socket 423 package, required a special case that the heatsink bolted on to). Intel's currently design of using clips to clamp the heatsink down to the motherboard is a fairly good one IMO, though it does put an awful lot of strain on the board. AMD's plan for the Hammer/Opteron is similar, and it looks like they're going to try to avoid some of the motherboard bending/strain, however it remains to be seen just how well it will work in practice.
In any case, it remains to be seen just what the power consumption of the Hammer processors will be. The only thing that is for certain is that they will use a LOT less power than their Intel counterparts (which, in this particular case, are the Itaniums, which use roughly twice as much power as the hottest running Athlons). IBM's Power4 chips also pump out huge amounts of heat, so relative to these competitors chips, AMD's Opteron could be a fairly low-power solution.
A one time competitor of X-MPs and Y-MPs was the ETA-10 () from ETA Systems. It's boards were immersed in liquid nitrogen. The old John Von Neumann Supercomputing Center at Princeton used to run one.
Heat spreaders aren't really that great when it comes to actually making the chip run cooler, all they do is exactly what their name suggests, they spread heat.
that's basically what i was saying when i said, "it allows better heat transference and a larger surface area so the temperature is more spread out. one of the reasons athlons are hot is because a lot of work is done in a small area." i was trying to convey that a given amount of heat energy will produce a higher temperature if contained in a small mass (current athlons) versus a larger mass (future opterons with heat spreaders.
though i may not have been as eloquent as you. thank you for helping clarify the message.
and i apologize for botching two aspects of soi. if you want to learn something, post on slashdot. if you are wrong, you'll find out.
yes i was wrong, Anonymous Coward. it is layers of interconnects, not silicon. i'm an idiot, though. thanks for contributing to the solution, but there's really only so much you can do for a lost cause such as myself. but i'm not a coward when it comes to admitting i have made a mistake.
Re:Heating issues?
by
BasharTeg
·
· Score: 3, Funny
I'm just waiting for Tom's Hardware to do a video with one of these supercomputers in the event of "total cooling failure." Of course, total cooling failure would mean something like the cooling system springs a leak, all the cooling liquid runs out on the floor, the copper cooling panels "fall off" (just like those damned heatsinks are always doing), the laws of thermodynamics are modified by God, and Q decides to change the gravitational constant of the universe bringing the earth unusually close to the sun. This all will likely cause the Opteron CPUs to smoke and burn out, giving Tom an opportunity to point out that if the Pentium 4 were used, it would just slow down to 10 frames per second when playing Quake 3.
Doh! Gotta end this post, the damned heatsink just fell off my Athlon again. Those wacky fucking heatsinks always jumping off.
Re:Heating issues?
by
delus10n0
·
· Score: 3, Funny
Yeah, my heatsink just commited suicide too!
Apparently gripping onto all three tabs on each side of the socket just isn't enough!
We need to devise a solution where the heatsink and CPU are permanantly joined! Yeah, that's the ticket!
Just don't hit the heatsink, or you'll rip the die clean off the board.
-- Not All Who Wander Are Lost
Re:Sandia's reliance on supercomputers make me ner
by
Astrorunner
·
· Score: 1
Just so we can nuke 11 million chinese with one warhead instead of only 10 million?
Cry me a river.
Oh well. Code named red storm..?
by
ketamine-bp
·
· Score: 2, Funny
from the infoworld report: The machine, code-named Red Storm, will require over 16,000 microprocessors to achieve that performance level, according to a researcher quoted in the Journal report.
This can be rephrased easily into
The machine, named code-red Storm, will require over 16,000 microprocessors' performance level to archive that, according to a researcher quoted in the Journal report.
Is it going to be running an unpatched version of Windows on an OC-48 line?
/me cringes in horror
-- "Evil will always triumph because good is dumb." -- Dark Helmet
Re:Oh well. Code named red storm..?
by
Anonymous Coward
·
· Score: 0
Sadly, Code-red seems to be back (I know I Know Offtopic) Looking over my apache logs we seem to be in for a new bout. But I do wonder what is the sound of One Hundred Teras Flopping??
Once again, AMD missed their chance. They should have named the K8 after the male lead on the robot soap-opera that Bender watches on Futurama:
"Calculon! We thought you were dead!"
-- We reserve the right to serve refuse to anyone. -management
Yay!
by
Anonymous Coward
·
· Score: 0
Maybe now AMD won't go the way of the buffalos... or was that buffalo? No, not buffalo: Apple.
Re:Sandia's reliance on supercomputers make me ner
by
garcia
·
· Score: 2
resume nuclear testing b/c we really don't know what they do...
Nuclear testing is sorta pointless. Nuclear weapons cause long-term irreversable destruction and human death.
I really don't see a need to restart testing.
These countries that have nuclear weapons programs currently, aren't going to be deterred by us testing them. They are going to be deterred by us using them (which is really not going to happen in the forseeable future).
Mod parent down please.
Water cooled
by
djstrehl
·
· Score: 3, Interesting
This will be really cool because I would imagine this will be water cooled like Cray's other computers.
I can see the online adds now.. "Personal water cooling setup. As seen on the Cray Red Storm"
See my post above. Crays in the past were cooled by 3 things, none of them water: Freon, Flourinert, and air. Circulating large amounts of water through a system that uses as much power as ours do is, well, a bit on the risky side:)
-- Go Badgers! -- #include "std/disclaimer.h"
next generation == last generation
by
jukal
·
· Score: 3, Interesting
Indeed, things do not look very good for AMD. Is there any AMD believers to explain how they will survive through next year? A $254 million loss in a quarter is not very convincing. Apparently, the have had to take a huge risk with putting the money in the design of this "new generation". Is it good enough?
Re:next generation == last generation
by
Mikeydude750
·
· Score: 1
I hope so. I don't want to see such a good company go down, and die while Intel contiunes to dominate the market.
Re:next generation == last generation
by
Anonymous Coward
·
· Score: 3, Funny
A $254 million loss in a quarter is not very convincing
Find me a US company that didn't lose money last quarter. And anybody who uses Arthur Anderson doesn't count.
Re:next generation == last generation
by
jukal
·
· Score: 2
> Find me a US company that didn't lose money last quarter. And anybody who uses Arthur Anderson doesn't count.
Here's one. I don't know whether they use AA or not:) SANTA CLARA, Calif., Oct. 15, 2002 - Intel Corporation today announced third-quarter revenue of $6.5 billion, up 3 percent sequentially and flat year-over-year.
Third-quarter net income was $686 million, up 54 percent sequentially and up 547 percent year-over-year. Earnings per share were $0.10, up 43 percent sequentially and up 400 percent from $0.02 in the third quarter of 2001.
Re:next generation == last generation
by
Anonymous Coward
·
· Score: 0
Yeah, it's called laying people off to match income.
Re:next generation == last generation
by
afidel
·
· Score: 2
GE, but then again with GE Capital they can basically print money =)
-- There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
Re:next generation == last generation
by
A5un
·
· Score: 1
Why, M$ of course.
As a side note, I believe RedHat broke even as well.
Re:next generation == last generation
by
tshak
·
· Score: 2
Are you serious? AMD may not be able to lose that much quarter after quarter, but you have to remember that this is a bad economy. Also, and this applies to all of you newbie investors as well, IT'S ONLY ONE QUARTER! One bad quarter after many good quarters is not the end of a company. Also consider that AMD's total assets are around 5.5billion. Not whole lot, but not a little either. Finally, Net Earnings, which is way more important then revenue, have been almost neck in neck throughout 2001, with Intel riding stronger throught the economic downturn. The bottom line? Intel spends a lot more per $ profit.
AMD is riding it's success on it's proven superior technology. Sure, Intel has the cash to sell undersell their products, but AMD doesn't take as big of a hit because their production process is so much cheaper. Although the P4's have become a lot better, the Athlon is still a superior chip. I think that the 64bit sector will welcome either player, considering that niether one has a dominant position to start in. It's almost an even playing field for AMD, and I think this will attribute to more success, not a demise.
--
There is no longer anything that can be done with computers that is nontrivial and clearly legal. -- Paul Phillips
Re:next generation == last generation
by
aminorex
·
· Score: 2
Intel has a lot of lines of business. If you consider each line of business, each product class, individually, Intel is likely to lose money on 64-bit CPUs for a long time to come, while AMD is likely to make money on them very soon, just on the basis of currently disclosed bookings, not even considering the established performance issues and the expected importance of 64-bit x86 extensions in the Linux server space. Intel is making some money on the P4 line, and a LOT of money on glue chips and embedded RISC CPUs, but Itanic is a terrible looser, as it always has been. Itanic 2 is likely to be the last generation, unless they are dead set on throwing good money after bad until their engineers can pull the bacon out of the fire with some as yet uninvented virtuoso trick.
-- -I like my women like I like my tea: green-
Re:next generation == last generation
by
Anonymous Coward
·
· Score: 0
They didn't release the Hammer to the market yet.
Re:next generation == last generation
by
Anonymous Coward
·
· Score: 0
They also make just about everything. Buying GE stock is like a mutual fund. Airplane engines, industrial equipment, generators, lightbulbs, every type of plastic from makeup bases to car dashboards, microwaves, stoves, blenders, X-Rays, CT, PET, and MRI scanners. Capital does make a lot of money, but recent divisions of the financial modality were made to make these reportings clear. The company makes money, because even though some businesses experience short term losses, long term growth throught strong acquisitions makes up for this. There are very few fields where GE is not number one or two. I grow tired of people thinking that everytime a company does well, it is because of fancy bookkeeping. You can't maintain double digit growth for so long by massaging the books.
Re:next generation == last generation
by
Anonymous Coward
·
· Score: 0
AMD only makes procs and chipsets at the moment. Intel has their fingers in a hell of a lot more honey-pots. But ok I admit they must have made some money. They can afford those lame commercials.
Re:next generation == last generation
by
Brian+Stretch
·
· Score: 4, Interesting
Simple. AMD crammed as much bad news as possible into Q3 and held up shipping their shiny new Athlon XP 2400+ and higher chips to distributors until the first day of Q4. Accordingly, Q4 revenues are going to be much higher than Q3, AMD's net loss will be considerably lower, and they'll hang in there just fine until the Opteron ships in early Q2 (late Q1?) next year.
AMD also took the unusual step of accelerating their changeover to 130nm and the new Thoroughbred Revision B core that those neato new 2400+ and higher chips use and letting old inventory burn off during the resulting downtime during the last two quarters.
I say "unusual" because Intel did just the opposite. They dumped lots of crippled 2GHz Celeron processors onto the market rather than shut down their old 180nm fabs and they brought lots of new 130nm capacity online. They have no prayer of finding buyers for all the chips they now have the capacity to build and the sales channels are choked with rapidly aging Intel inventory. Their ASPs are eroding and the Xeon line that sustains their profitability is going to get Hammered in about 6 months, assuming no Tier 1 OEMs grow a pair and start offering AMD Athlon MP servers and workstations before then.
Soooooo, AMD's future looks pretty good, depending on how badly Intel panics at the mess they've gotten themselves into.
Re:next generation == last generation
by
mgoff
·
· Score: 1
Re:next generation == last generation
by
Anonymous Coward
·
· Score: 0
.sig: > If Pro is the opposite of Con, then what is the opposite of Progress?
Regression.
If Mikeydude750 is the opposite of intelligent, then what is the opposite of illogical?
Not only the fastest, but also the hottest
by
asscroft
·
· Score: 3, Funny
computer on earth!!
1. cluster lots of opterons 2. place popcorn on top 3. sell popcorn and cycles 4. profit
I'd like a beowulf cluster of those...
does it play quake?
yeah, but how much longer before it is DRM-enabled to run only MS....
etc..
-- because I have been enjoined by this Holy Office to abandon the false opinion which maintains that the Sun is the centre
Re:Not only the fastest, but also the hottest
by
Mikeydude750
·
· Score: 1
Okay, enough with the whole "beowulf cluster" thing. It would be cool, but it's not going to happen.
Anyways, I hope AMD doesn't do the DRM thing, because that would just piss the user off. And when you piss the user off, you might not get his business again.
I've heard that these big machines are way to big to play Quake, but I'm not sure about that.
Re:Not only the fastest, but also the hottest
by
stantron77
·
· Score: 1
That's way easier than the silly laser they used in Real Genius.
-- "Good people do not need laws to tell them to act responsibly, while bad people will find a way around the laws."
- Pla
Unicos VS Linux
by
Anonymous Coward
·
· Score: 0
Have they ported Linux, or one of the BSDs to it yet, though?
A lot of people don't like coding for Unicos, and think how fast Tux Racer would run on it:-)
temporary setback
by
GunFodder
·
· Score: 5, Insightful
AMD has been suffering recently because they are focusing their efforts on Opteron, and meanwhile Athlon isn't getting any younger. This is a temporary setback; if Opteron is any good then AMD will be performance competitive again, allowing them to sell at higher price points and get a better margin.
AMD was in the same funk back when Intel released the Pentium II and AMD was still working on Athlon. Once AMD got Athlon out the door they started doing a lot better.
The main difference between now and then is that the whole tech industry is in a slump. AMD needs more than for Opteron to be a screamer in the benchmarks - they actually need people to start buying cpu's again.
On the other hand, even through these tough times in the industry, AMD has had better revenues from cpu's than they did during their boom time when the stock hit a pre-split high of 95 bucks. It's the flash industry that is really killing AMD, revenue on flash has been dreadful since 2000.
I don't think it's much of a setback. It's more simple business. While the "who's got the bigger..." comparisons can be fun, it's not always profitable to have the fastest chip on the block. During these economic times especially, it's much more important to deliver a very fast CPU at a very good price. Very few people need a 2.8Ghz Athlon right now, and simple business dictates that there's no reason to be producing them yet. AMD has refocused on future platforms that will be released to sectors who A) actually need the performance and B) have the cash to pay for the performance. By the time the 64bit platform makes it to the desktop, the general consumer may have software that actually needs the extra performance. For now though, the sub 2ghz Athlons at sub $100 prices are more then adequate.
--
There is no longer anything that can be done with computers that is nontrivial and clearly legal. -- Paul Phillips
In what sense is Athlon not competetive with P4? An Athlon 2600+ will stomp a P4-2.53GHz, and with DDR-2700 will parallel a P4-2.8GHz for about $200 less per chip. That's the top-of-the-line, in both cases. I don't see that as uncompetitive in the least.
How about thermal protection? I don't want my processor to fry my motherboard or itself! Another area is DIVX encoding (or codecs that are optimized for SSE2)
-- Space may be the final frontier, but it's made in a Hollywood basement.
--Red Hot Chili Peppers, Californication
Re:temporary setback
by
Anonymous Coward
·
· Score: 0
The real problem is that AMD isn't making 2600+ chips in any kind of volume. The majority of their processors are 2100+ and lower. That hurts, especially with the market in a slump.
"An Athlon 2600+ will stomp a P4-2.53GHz, and with
DDR-2700 will parallel a P4-2.8GHz for about $200
less per chip"
No it won't. Not even close. Not "in some apps" not "with this configuration", not in any way shape or form. Even if you get a 2800+ with 3200, 333FSB OCed to 3200 RAM speeds, it is barely on par with a 2.53
That and considering the fact that the 2400 is just now available, and it won't be until 03 that you can get your hands on a 2600 or 2800.
I like AMD, they give better price/performance, but be realistic, they are far behind in the 'king of the hill' race.
notice in this POV-Ray test, the P4 needs a _1 gigahertz_ clock frequency lead to pull even/ahead (depending on ram type) with the Athlon.. that my man is brute force floating point power in action:)
The 2600+ OTOH, IS barely competitive with the 2.53Ghz P4, the 2800+ has a faster bus though, and that seems to balance things out, with the 2.8Ghz P4 still having an edge when bandwidth is the bottleneck, and the Athlon having an edge when raw computational power is the bottleneck.
Actually, the XP2800+ won't be available 'till Christmas AFAIK. Nevertheless, that's not my point. There's no reason for AMD to strive to build a 2.8Ghz Athlon when no one needs it. That doesn't mean they don't try and build faster chips (see: Opteron), it just means that they don't release as many faster consumer grade chips because the market doesn't demand it.
--
There is no longer anything that can be done with computers that is nontrivial and clearly legal. -- Paul Phillips
That's like saying "What about Applications written for Windows XP? They run better under Windows XP instead of Lindows.
You can't reasonably expect AMD to beat Intel's proprietary instructions with a processor that doesn't have them; unless they manage to make a pre-processor that decodes the instructions and does it faster than the Pentium 4...
-- That what was all this school was for... to teach us how to solve our own problems. -- janeowit
Units nitpicking
by
Anonymous Coward
·
· Score: 2, Funny
Since the expression "100 teraflop" contains no time related unit, it probably refers to the total number of flop that will be delivered by the system running at 40 teraflop/s before its predicted meltdown after 2.5 s.
I have seen the expression "Free-as-in-Beer" a few times on Slashdot, yet I have no idea what this means. I have scoured my head to think of what it could mean, but can't figure it out.
What does "Free-as-in-Beer" mean?
Free beer, to me, would seem to be the greatest thing in the history of the world!
...to quote the great John Belushi from the movie Animal House..
..go ahead, take one...don't cost nuthin'...
the code / app is free to use...i still retain ownership but you can use it as you'd like....my take anyway....
-- "Hatred is the coward's revenge for being intimidated"...George Bernard Shaw
Re:Free-as-in-Beer
by
program21
·
· Score: 0, Redundant
Free-as-in-beer means it doesn't cost anything.
-- This has been a test. Had this been a real emergency, we would have fled in terror and you would not have been informed.
Re:Free-as-in-Beer
by
Overt+Coward
·
· Score: 5, Informative
"Free" is an overloaded term. In the typical context it can either refer to cost ("free" = "no cost") or to rights ("free" = "no legal restrictions").
To differentiate, many postings here at./ and elsewhere use "free speech" vs. "free beer" as examples of whether something is free from restriction or doesn't cost anything, respectively. So "free-as-in-beer" refers specifically to cost.
Re:Free-as-in-Beer
by
Anonymous Coward
·
· Score: 2, Informative
Free-as-in-Beer means that you don't have to pay for it, just like when someone gives you a beer.
This is to differentiate it from Free-as-in-Speech.
The first usage of Free is to mean "Gratis" where as the second means "liberty." English just uses the same word for both concepts.
GNU stuff takes the "Liberty" meaning as being more important than the beer meaning. That is why you can copy a Redhat ISO without any problems. You are at Liberty to do so. Of course you can also pay Redhat if you want to.
free-as-in-beer (as opposed to free-as-in-speech) refers to the cost. It's free!
Free-as-in-speech refers to the EULA. Some Free-as-in-speech EULAs include the GPL, the BSD license, and the MPL. These licenses vary in their degree of freedom granted to the end-user, but are generally considered free, especially compared to EULAs tied to many commercial products.
Free beer? I'd rather have cider myself....
-- meh.
Re:Free-as-in-Beer
by
Anonymous Coward
·
· Score: 0
And if I piss in a bottle and give it to you for free, will you drink it? You will never see Linux the same way again.
You have stumbled on one part of the core culture of/. Info about 'free-as-in-beer' is available in the faq. See also; 'free-as-in-speech', 'imagine a bewoulf cluster of these', 'FP! w00t!!', etc.
-- "A language that doesn't affect the way you think about programming, is not worth knowing" - Alan Perlis
These terms arose because of the multiple definitions of the word free in english. Free as in beer is used to classify anything that you don't have to pay for to use. Any sort of freeware would count as this. With the GPL (and other public licenses), software was also free, not just in the $ sense, but you were allowed to modify it. Because liberty software sounds like a company, free softwarewas reused, but to differentiate it the term is free as in speech. Meaning you are free to modify it. Personally I like the method of terming modifyable software libre and freely usable software gratis from Spanish. You can check the jargon file, or probably e2 to find more info on who originated the terms, and read more about each.
-- Degaussing scares the bad magnetism out of the monitor and fills it with good karma.
Re:Free-as-in-Beer
by
Anonymous Coward
·
· Score: 1, Insightful
It comes from a quote
"Open source is free as in speach, not free as in beer" ~RMS
It's a reply to the people who think OSS is all about getting free stuff (most of them are pre-college age)
Re:Free-as-in-Beer
by
Anonymous Coward
·
· Score: 0
s/speach/speech/
Re:Free-as-in-Beer
by
Anonymous Coward
·
· Score: 0
But only older and more hard headed OSS zealots can't see that free as in speech MEANS free as in beer. It always has and it always will. Red Hat CAN charge if they want but they also HAVE to give it away free AND they cannot stop anyone else from re-labeling it Green Hat and giving it away free.
Re:Free-as-in-Beer
by
Anonymous Coward
·
· Score: 0
free beer refers to something that will not cost you any money, i.e. you can drink without paying, but the beers recipe is still the intellectual property of the brewer. there could still be a hidden "cost" that has nothing to do with the immediate time period, or monetary.
example: microsoft could give their operating system away for free (as in beer). this eventually locks you into a monopoly with microsoft, and they can change their minds later and start charging. This can happen because their product is not free (as in speech)...they maintain all Intellectual Property rights, and can change their minds about how/what/where/when they distribute/make/control their software products.
you are powerless in that case.
free as in speech means that while something may or may not cost you money up front, in the long run, there is no IP rights being held over your head.
linux is free both ways.
though some people "buy" linux, they still gain the "free as in speech" benefits.
Framerate
by
Anonymous Coward
·
· Score: 0
I know, you need that 100 frames/sec when playing doom 3 even though the human eye can only see up to 50 frames/sec.
Re:Framerate
by
Anonymous Coward
·
· Score: 0
I thought it was more around 60Hz. It's sort of amazing that even the human eye is digital:P
It can feel up to 100 FPS really. I can definetly feel the difference between 50 FPS and 100 FPS. Not like i will miss any detail at 50FS, but at 100 you can be sure you eye "refresh rate" will ALWAYS cat a new frame everytime.
Some people may not notice beyond 60 FPS. Some people do, and 100 feels better for them. More real. Of course, you mostly notice under FPS because they happen at very high speed and everything moves when you make a turn.
Some people may not notice beyond 60 FPS. Some people do, and 100 feels better for them. More real. Of course, you mostly notice under FPS because they happen at very high speed and everything moves when you make a turn.
Show me someone who can feel a difference past 85fps, and I'll show you someone who thinks analog sounds better than digital. (hint: what's your monitor refresh?)
-- "We returned the General to El Salvador, or maybe Guatemala,
it's difficult to tell from 10,000 feet"
85 hz has a nice balance. The things is some people can feel the difference. How much you can feel depends on the colors in the (let's say) movie and how much things are panning/sec. Of course, you will need to to have a good monitor that can display 100 hz.
When you talk about games the thing changes quite a lot but for practical reason. If a card benchmark against game X shows and average framerate of 120 FPS, it's likely you'll have at least 80 or 100 in the slowers more dense part of the game. f it shows 83, at some point it will certainly display less than that. This is a practical thing, not that you can see 120 FPS, just that beign able to play a game at 120 average is a good thing for these reasons.
Anyway, 85 hz is pretty good, some people may notice an improvement at 100 hz, but it's not a significaty improvement.
In practice, it may depend. The key variable is if the FPS are time constant. That is, a movie played back with a good decoder, and assuming it has 86 Fps, you can only see 85 FPS.
For a game, it depends. If for any reason you cannot deliver say, frame 40 of second 1, and you can afterward "process" 46 more frames, you'll only see 84 real frames.
This is what I am talking about. A practical thing. Of course, you will never see more than 85 frames with an 85 hz refresh.
Anyway, some people can tell the difference between a game at 85 hz and 100 hz. It really is irrelevant if you have 85 hz or 100hz. What you are impling is that people that (example) have 15 hz monitors could not tell the difference between 15 hz and let's say 60 hz. Of course they can, you just need a better monitor.
Notice that I said framerate and not average framerate. It's not a difference between theory and practice, it's a matter of average versus instantaneous.
-- "We returned the General to El Salvador, or maybe Guatemala,
it's difficult to tell from 10,000 feet"
I know, I was just making the distinction because everyone usually quotes average FPS and don't know it's instantaneous FPS what matters. That is, many people think they can feel 100 FPS when they cannot (but some still clearly can).
and come over to my house, and I will demonstrate.
-- LETS DECOMPOSE & ENJOY ASSEMBLING
WSJ story here
by
CathedralRulz
·
· Score: 5, Informative
I'm a subscriber. Here's the article. I hope it convinces you,too, to subscribe.
AMD's New Opteron Chips
Are Tapped for Red Storm
By DON CLARK
Staff Reporter of THE WALL STREET JOURNAL
Endorsing the technology of one of Intel Corp.'s key rivals, Sandia National Laboratories and Cray Inc. plan to build a massive supercomputer using a soon-to-be-introduced line of microprocessor chips from Advanced Micro Devices Inc.
The development project, estimated in June to cost $90 million, is a high-profile vote of confidence for AMD's new Opteron chip, in a small but prestigious market long dominated by other chip suppliers. It represents a missed opportunity for Intel, which has been targeting its new Itanium line at high-performance computing applications.
Red Storm, Sandia's name for the new supercomputer, also marks a step forward for the U.S. effort at leadership in supercomputers, which suffered a blow this year with the completion of a huge machine called the Earth Simulator by Japanese government agencies and NEC Corp. Where recent U.S. machines have largely been constructed out of components used in commercial computers, Cray is expected to develop special technology for connecting the AMD chips that should make Red Storm suited for more-complex scientific problems.
"This is a move away from commodity components," said Horst Simon, division director of the National Energy Research Scientific Computing Center, a supercomputer facility affiliated with Lawrence Berkeley National Laboratory. "It's very exciting."
Sandia, which does research for the U.S. Department of Energy in Albuquerque, N.M., and Livermore, Calif., has a performance goal of 100 trillion operations per second for Red Storm. It hasn't disclosed most technical details, including the chip selection. But Mr. Simon estimated that the machine will require 16,000 or more microprocessors to hit its speed target, which would appear to surpass the Earth Simulator's current performance.
Sandia said in June that it had selected Cray, a longtime supercomputer maker based in Seattle, to negotiate a development contract. Cray and Sandia officials didn't return calls seeking comment Friday. AMD and Intel officials declined to comment.
AMD could use some good news. The company's Athlon chip line, mainly used in personal computers, has been falling behind the performance of comparable Intel chips. The company reported last week a third-quarter loss of $254 million on sales of $508.2 million, off 34% from the year-earlier period.
Opteron is a high-end member of the new line, code-named Hammer, that is due out next year and viewed by analysts as AMD's best hope for recovery. Like the Itanium, Hammer chips are designed to process 64 bits of information at a time, instead of 32 bits, a capability that helps run huge databases and solve scientific problems.
Intel's Itanium line, developed over eight years with help from Hewlett-Packard Co., is based on an entirely new architecture and achieves its best performance on new 64-bit programs. AMD, by contrast, made 64-bit additions to the original Intel technology used in the past by both companies.
The difference, AMD says, allows Hammer-based computers to run both 32-bit and 64-bit software at high speed. AMD released preliminary test results last week for Opteron -- so far not validated by outside researchers -- that show the chip exceeding Intel's latest Itanium 2 model on one of two widely-used speed measures, AMD said.
Itanium 2, introduced last summer, has already been selected for at least a half-dozen high-performance installations. The Pacific Northwest National Laboratory, another Department of Energy facility, is building a $24.5 million system based on 1,400 Itanium 2 chips. Based on past Sandia announcements, the Red Storm project's stated performance goal is more than 10 times that of the Pacific Northwest project.
Write to Don Clark at don.clark@wsj.com
Re:WSJ story here
by
Anonymous Coward
·
· Score: 0
And I hope WSJ sues your ass for (C) infringement. Their web site specifically states their prices for reproduction of articles and I'm betting you didn't give them a dime.
Just another user...
Re:WSJ story here
by
Anonymous Coward
·
· Score: 0
developed over eight years with help from Hewlett-Packard Co.
Ah, mourn for alpha:(
Could someone explain to me why you'd want to use x86 for a job like this? The newer models (i586/i686) have lots of instructions for graphics type applications, and although you can apply many of them to vector operations, wouldn't it be better to choose something in a RISC, and not have to fool with all the x86 baggage? As long as Linux runs on it, it shouldn't matter. At least some of the hardware is going to have to be custom anyway, because they certainly won't use regular PC motherboards anyway. Small boards with already Linux supported ethernet chips embedded in them, along with a slot for chip and slots for memory would be grand.
As an aside, it's funny to think that the cooling for this thing will probably be controlled by a bunch of pooty microcontrollers chained together:)
It convinced me to let you post stories here for me.
-- "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
you are retarded.
by
Anonymous Coward
·
· Score: 0
man what is up with you fucking slashbots today
And in other news ...
by
Anonymous Coward
·
· Score: 0
The "(US Dept. of Energy) has just built a 100 Teraflop super computer employing AMD's Opteron (Hammer) processors", causing rolling blackouts. Rummor has it that it isn't the power consumption by all of the procesors that is causing the blackouts, rather it is all of the power needed to keep all of the industrial strength fans and nitrogen cooling systems running. Citizens of the area are asked to turn off all their power consuming items (please use candles) so that we can see how this moster machine performs.
Besides, you're all unemployeed... you can't afford power. Watch TV at a buddy's house... it's cheaper!
Re:And in other news ...
by
Anonymous Coward
·
· Score: 0
Taht woudnt of bin funney evne if yuod of speld it rite.
God damn fucking idiots. Why is it that so many computer geeks, people who are supposed to excel with languages, structures, syntax and logic, can't even communicate in their own native tongue?
You are a joke, asshole, and I bet you're going to play that tired old dyslexia card as a pathetically transparent excuse for being lazy and illiterate. Fuck you.
It seems like Linux is the most likely choice for an operating system for this, but they don't mention anything. Does anybody know?
Re:running Linux?
by
rotwhylr
·
· Score: 3, Insightful
AFAIK, Linux doesn't scale above a small number of CPU's. Looks to me that implementing it on a 16,000 cpu computer would require a complete rewrite of the entire OS.
Even high-end UNIXs like Sun's Solaris don't commercially scale much beyond 100+/- cpus, and are generally intended for UMA, not NUMA.
When you step away from standard SMP systems into parallel and NUMA architectures, there are lots of problems to solve that don't exist at the small end of the scale, both at the OS and the app level. OSes better suited to that environment already exist (anyone have more info ???).
-- -- Windows is not simply installed on a computer; it is inflicted.
I imagine that it will be running some form of MPI.
Let's say that they use 2000 8-way Opteron boxes. Thus each kernel will run over 8 processors, with MPI performing message passing between boxes. That's how it works on ASCI White etc anyhow (except they run AIX on POWER3 processors).
It would be quite an acomplishment to implement a NUMA architecture over 16000 processors. SGI can get decent performance with 128 processors (supposedly 512 but I have not used this), and I think Cray's T3E-1350 supports NUMA on up to 2176 processors. Cray is known for great interconnects though -- they can do it if anyone can.
Re:running Linux?
by
Anonymous Coward
·
· Score: 0
No, the Cray systems run the UNICOS, go to cray.com and click on software.
It's a cluster, not a shared memory machine. Linux scales very well to clusters. In fact, it seems like it is the only choice right now for Opteron-based clustering since there is no released version of XP supporting it. Also, Linux is much more widely used for clustering on PC hardware than anything else and has a lot of tools available for it. So, it seems like the most logical choice; I would just be nice to see that confirmed.
Re:running Linux?
by
Anonymous Coward
·
· Score: 0
Yes, UNICOS is the operating system. So?
The T3E has a single global memory architecture, probably similar to NUMA (Non-uniform memory architecture) from SGI, HP and others.
Re:running Linux?
by
Anonymous Coward
·
· Score: 0
...OSes better suited to that environment already exist (anyone have more info ???).
I was initiall thinking UNICOS. However, looking at the specs for ASCI Red ( the predecessor), perhaps they are just going to port the that OS onto the Opteron chips. Since Opterion is x86 compatible that would allow for most of the user software to be moved over without too much of a disruption after perhaps a few kernel level tweaks. May not even require 64 bit mode. ASCI Red only had 256 MB per node you can 6x that and still be well within 32 address space.
I think you are correct in that this is NOT a cluster (the press release hints at this) and so therefore Linux need not apply. It isn't quite NUMA either though. More than likely the want to move the stuff that has been hacked to run fast on ASCI Red to a newer box that will run the same hacked code much faster. (Hence Opteron's backward x86 capatibility being a key factor along with high floating point crunching capacity.).
We have gotten quite accustomed to the very cool and efficient operation of our high density VLSI microprocessors. How soon we forget the days when computers were constructed not of VLSI devices, nor even discrete component semiconductors, but good old tubes! Those babies got HOT! I can't imagine cooling on the scale of a supercomputer is any more difficult than cooling the old discrete component and tube computers had been. I would guess the primary challenge is finding a cost efficient cooling technique that efficiently absorbs the heat from all of those CPU chips.
you are retarded.
by
Anonymous Coward
·
· Score: 0
s s i a
you are retarded.
by
Anonymous Coward
·
· Score: 0
why is everyone so fucking retarded today
Look at FP
by
Anonymous Coward
·
· Score: 0
This is exactly the same as the first post. And yet this is "funny" and the first post is "redundant". Are the/. moderators so fast that they move backwards through time?
What the hell are you talking about?
by
Wakko+Warner
·
· Score: 2, Insightful
Hopefully AMD will be able to produce a quality product that far down the road. They've been bleeding money at an alarming rate and some say they're focusing too much on marketing instead of engineering- much like 3dfx did back in the day.
I donno where you've been, but, "back in the day", 3DFX was the only 3d card manufacturer, and the products they produced were light-years ahead of their competition (who WAS their competition? S3? Nvidia? Remember how hard Nvidia sucked before they bought 3DFX?)
What's even more damaging to your argument is that, back in the day, the only CPU manufacturer worth a damn was Intel. Unless you've been living under a rock for the past few years, you know AMD's current line of CPUs is as competitive and high-quality as anything else in the PC arena.
In conclusion, stop being retarded.
- A.P.
-- "Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
Re:What the hell are you talking about?
by
fferreres
·
· Score: 3, Insightful
NVidia focused on engineering. 3DFx lost a lot of their core developers and focused on marketing and/or bets (integrating with board manufacturer, increading marketing expenses, restricting chip sells to their own producers, etc).
I know, because I have an Edge 3D which puts me scene even BEFORE 3Dfx released their first product.
3Dfx had the best product for 3D, and was a small company. They focused more on 3D only. They cut expenses by ditching integration with a 2D core, which meant the procesors could only do full-screen, and even with the arrival of Rush3D, they could only do 3D in ONE window at a time.
They made a lot of mistakes really. Nvidia just kept focusing on engineering at a HUGE LOSS (don't know where that money came from because they REALLY LOST A LOT OF IT, only with TNT2 they managed to get afoot, several years later. All other companies like Rendition, 3DFx, S3, etc didn't survive). Nvidia did survive, money pumped at it by unknown investors (was venture capital at the time).
I am pretty sure that wasn't coincidence. Remember Sega was to use 3Dfx chip. And finally 3Dfx closed, and some years later XBox is born powered by Nvidia.
-- unfinished: (adj.)
Re:What the hell are you talking about?
by
jtshaw
·
· Score: 1
3DFX was the pioneer in 3D graphics. But making the statement that Nvidia sucked before they bought 3dfx is rediculous. The Voodoo 2 was the last product 3dfx made that lead the industry. Nvidia released the TNT (decent, but not quite there yet), and then the industry leading TNT2 well before they purchased 3dfx and that was really what started Nvidia's domination.
If you want to look at the single biggest even that attributed to 3dfx's downfall it was deciding to be the only manufacturer of products with there chips. Buying STB never paid off for them.
Now I don't totally agree with the post you responded too but I look at the situation this way... AMD has definitly built some awsome processors in the last few years, and will continue to do so. But in my mind they have yet to engineer the complete product. The Athlon performance is great, but the packaging, chipsets, ect. never achieved the quality of Intel counterparts.
What we need from AMD is not just a good performing processor, but better packaging (not exposed, easy to crack cores), better heat management, and more reliable chipsets. Having memory that is 33Mhz. faster doesn't mean a damn thing when the chipsets show instabilities...
Re:What the hell are you talking about?
by
Junky191
·
· Score: 1
Sorry, but you must have not been alive during this time period. 3Dfx got killed by strong competition from Nvidia- the GF2 in particular. They were outengineered, plain and simple. Nvidia bought them after they declared bankrupcty and were in tatters.
Re:What the hell are you talking about?
by
Decimal
·
· Score: 2
I am pretty sure that wasn't coincidence. Remember Sega was to use 3Dfx chip. And finally 3Dfx closed
They were? Sega of America wanted to use 3Dfx in the Dreamcast. Sega of Japan wanted to use the PowerVR (NEC, I believe). As usual, Sega of Japan got to pick. There's a bit of a complicated story behind it, but if they had stuck with 3Dfx they may have had the money and product supply to keep the Dreamcast floating long enough to make a profit. But anyway, Sega wasn't "to use" any 3Dfx chip. SoA just wanted to use one.
I hope they are not putting that supercomputer anywhere near Antarctica.. 16,000 Athlons can get pretty hot resulting in some serious polar melting forcing me to evacuate from New Orleans (we are already underwater).
I'm a subscriber. Here's the article. I hope it convinces you,too, to subscribe.
Thanks for the article, but sorry, your valiant effort does not convince me.
-- Tat Tvam Asi
Gratis
by
Anonymous Coward
·
· Score: 0
Why don't you just do as the rest of the world and use the word 'gratis' for something that doesn't cost any money? It fromes from the latin 'gra', which means 'free' and 'tis', which means 'as in beer'.
Hast to be said:
by
A_Non_Moose
·
· Score: 1, Flamebait
Sould it be: Please Hammer, don't hurt 'em or Imagine a Beowolf Cluster of these?
Nah, how about: Please Hammer, don't imagine a Beowolf cluster of these.
(I gave up mod points for this comment? Sheesh, I must be smoking crac....errr...oops)
-- Have you read the moderator guidelines? Well, have you, PUNK?
(and I want a Karma: Gnarly option)
Thanks
by
Anonymous Coward
·
· Score: 0
I was going to subscribe, actually. But now you posted it here, violating WSJ's copyright, I don't need to anymore. Thanks.
reason for "op" instead of "flop"
by
help007
·
· Score: 1
According to a tour guide, the powers that be at the national labs thought that "teraflop" sounded like a really, really bad mistake, so they decided to call a teraflop a teraop instead. In other words, the phrases refer to exactly the same thing.
Re:Sandia's reliance on supercomputers make me ner
by
Anonvmous+Coward
·
· Score: 2
"Please, contact your local senator and representative, explain the dire need for the US to resume nuclear testing to prove that we have a valid, proven deterrent. We may nuke a few more Nevadans and Utahians, but they're dope-smoking whores and heretical polygamists, respectively. It's better that we detonate a few nukes, than rely on unproven simulations and end up being ruled by inscrutable atheistic Chinese overlords in a generation."
-- Escher was the first MC and Giger invented the HR department.
Re:Sandia's reliance on supercomputers make me ner
by
Anonymous Coward
·
· Score: 0
I don't think you understand the point of nuclear weapons testing.. mainly it's to ensure that current weapons are still operational and not, say *malfunction*. It takes a vast amount of effort to keep weapons operational, especially those that contain decades old technology. Without going into the politics of whether they should exist or not, it is in everyone's best interest to make sure they don't introduce problems of their own.
o/~
The sun is a mass of incandescent gas A gigantic nuclear furnace Where Hydrogen is built into Helium At a temperature of millions of degrees o/~
Hey, someone had to say it.
Guess you'd better invest in a good heatsink (even though the speculation is that a 0.09 micron process will produce hammer chips that can survive with passive cooling...)
"Yoho it's hot, the sun is not a place for you and me, but here on earth there'd be no life without the light it gives."
Great song, great song.
builder must be around ten years
by
peter303
·
· Score: 2
Its going to take four years for the machine to reach full speed, and you like to use several years. Only the most established names can qualify- IBM & Intel. SGI, HP(alpha), and AMD are contenders, but I dont know if they have the financial stability.
Don't try to use your magic words on me!
by
Anonymous Coward
·
· Score: 0
Copyright is a figment of a diseased and avaricious imagination.
Re:Don't try to use your magic words on me!
by
CathedralRulz
·
· Score: 1
Hi,
You are right, I went back and checked and it turns out I did violate the copywrite. I thought about it before posting and recall that in limited circumstances that I could post articles, but after going back and checking, it appears I cannot. See here:
(ii) You may occasionally use our "E-mail This" service to e-mail an article from WSJ.com to a few individuals, without charge. You are not permitted to use this service for the purpose of regularly providing other users with access to content from WSJ.com.
In any case, I admit I didn't recall the rules correctly and don't want to give the impression that I don't respect copyright laws. I own every CD I ever ripped, every DVD I ever watched, every software title I have ever played. Being a developer myself, I can appreciate copywright.
I've decided to notify WSJ of what I did ("posted an article on an internet discussion board") and we can see what comes of it.
Re:Don't try to use your magic words on me!
by
bugnuts
·
· Score: 2
If you're not regularly providing articles, who cares? WSJ is big enough that a single article is fine for free use, as it's a tiny fraction of what they print and won't cause them harm. If you signed a contract saying you won't use any other method to distribute, except the "Email this article" function, then you violated a usage license, not a copyright.
Re:Don't try to use your magic words on me!
by
Anonymous Coward
·
· Score: 0
Blanket comdemnation of copyright is a figment of a diseased and avaricious imagination.
Re:Sandia's reliance on supercomputers make me ner
by
Anonymous Coward
·
· Score: 0
Eh, the Utahians actually own most of that desert, in preparation for the day that California slides into the ocean.
Parent is troll, or else trying to be funny but isn't. I'll bite anyway.
Basically, Sandia is betting the safety and sanctity of the Free World on computer simulations... Basically, you're saying the safety of the free world rests on our ability to blow the rest of the world off the face of the earth with nukes. But you know what? We can do that already. We have enough ICBMs to kill the "godless commies" a hundred times over, and i have no idea why we are doing more testing unless it's for some sort of intercepter. Now those have to be modeled first, and you can't test them anyway because that would require an atmospheric blast (bad).
Please, contact your local senator and representative, explain the dire need for the US to resume nuclear testing to prove that we have a valid, proven deterrent. See above, for why this is unnecessary. Also recall the so-called "Nuclear Test Ban Treaty." I wonder what that does....
In conclusion, the parent is a troll or a very stupid bigot. Let's hope it's the former.
-- I hereby place the above post in the public domain.
Re:Sandia's reliance on supercomputers make me ner
by
Tailhook
·
· Score: 1
"Just so we can nuke 11 million chinese with one warhead instead of only 10 million?"
The US isn't developing new warheads or ICBMs. The stock we have we will live with. By the middle of this century, some of these warheads will be close to 100 years old. Sandia is busy simulating warhead decay.
Despite popular perception of vast quantities of cheap devices surrounded by evil Generals slobbering for a chance to use them, nuclear warheads are very complex and require intense maintenance by sober professionals. Making sure these weapons actually function is a big part of the freedom you enjoy, whether or not you actually live in Sandia's country.
I donno where you've been, but, "back in the day", 3DFX was the only 3d card manufacturer, and the products they produced were light-years ahead of their competition (who WAS their competition? S3? NVidia? Remember how hard nVidia sucked before they bought 3DFX?)
Yeah, NVidia sucked, that's why they had enough money to buy 3dfx. Yup.
3dfx was way ahead of the other 3d accelerators for the first few revisions, but their "who gives a shit about image quality" attitude, and complete lack of innovation killed them long before nVidia purchased them.
Basically 3dfx had a killer product, and spent all their time doing nothing but making it faster, while nVidia and others concentrated on technological advancements, such as T&L engines, and the like. 3dfx died because they couldn't compete technically.
Hrm... I have a feeling you're a troll, but whatever.
I'm sure most of you who've programmed have felt that way, until you tried to compile and execute your program (real programmers, I'm not talking interpreted baby languages like Basic or Perl), and then you were amazed by your simple syntax errors, and astonished by your modeling misconceptions
Well yeah, but that's just because I didn't bother to prove them....
What part of the definition of science makes you think manually managing memory allocation and deallocation is better than having a smart algorithm for handling it?
What school or university did you attend (if any). Could be usefull so tat I know where not to send my kids.
Perhaps we should name it after something simple, yet powerful...
Perhaps a Norse god, or legendary Scandinavian warrior?
-- -- Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
Re:The mighty have forgoten the first grade
by
Anonymous Coward
·
· Score: 0
Guess he was too busy sucking you cock to spell-check.
Local Politics Needs Heat Spreader
by
4of12
·
· Score: 4, Interesting
It will be real interesting to be at local chamber of commerce meeting where Sandia Labs management gets to meet with managment from another big employer in Albuquerque.
That's right boys and girls.
On the west side of the Rio Grande is Rio Rancho, home of Intel Fab 9. (the same one that got struck by lightning a while back).
-- "Provided by the management for your protection."
that's because you are an idiot
by
RelliK
·
· Score: 2
AFAIK, Linux doesn't scale above a small number of CPU's. Looks to me that implementing it on a 16,000 cpu computer would require a complete rewrite of the entire OS.
Here is a 256-CPU IA-64 Linux cluster, #53 on the top 500 list. And here is CPlant, #50. You can find more Linux boxes on the top 500 list
-- ___
If you think big enough, you'll never have to do it.
Re:that's because you are an idiot
by
bugnuts
·
· Score: 2
Without addressing whether he's an idiot or not, he also said it would require completely rewriting the OS. That's not quite true, but it did require some kernel mods, some custom coding of drivers, a completely new way to handle NFS, and many other changes. It is open-source, so you can have a look for yourself if you're really interested.
Re:that's because you are an idiot
by
Anonymous Coward
·
· Score: 0
I'm not sure who the idiot is..... Those are linux boxes (plural). There is no single linux box within miles of the top 500. Think Cray T3, SGI Origin, not a cluster of N boxes running N copies of the OS.
From the vague description I think Red Storm may be a NUMA box. From the press release at Cray/Sandia:
....were key factors. MPP supercomputers, designed as single machines, are more efficient than "clustered" systems that more loosely link together multiple servers or PCs
In that case this would be a linux box (singular). One linux logical "box" with 16,000 processors isn't going to work so well. Hell, one linux box with 16 processsors isn't going to work so well. We're not talking clusters here. Apples to oranges.
Concievably Cray could port UNICOS. But that would be a hassle, but that's an unique advantage they have.
Re:that's because you are an idiot
by
Anonymous Coward
·
· Score: 0
Yeah, you really showed him! I am now *COMPLETELY* certain that because Linux can support 256 CPU's, it can also support the remaining 15,744 that this Cray is going to use.
Yr 50 31337! i ph33r yr m4d 5ki11z0rZ!
Re:that's because you are an idiot
by
Anonymous Coward
·
· Score: 0
Re:Sandia's reliance on supercomputers make me ner
by
aminorex
·
· Score: 2
> Nuclear testing is sorta pointless.
You'd quickly change your mind if a dud nuke failed to deflect an asteroid.
> Nuclear weapons cause long-term irreversable > destruction and human death.
Well, Doh! That's what they're *for*.
Really, if we're going to keep a stock pile, we should test them. I suggest testing them by building a big glass tunnel underneath Costa Rica connecting the Pacific to the Atlantic, so that China (via the PLA proxy, Li Ka-ching's Hutchinson Wampoa), doesn't control the *only* canal.
Looks like the fine folks from Sweden have already made a 240-CPU Athlon MP super-cluster. It is currently #94 on the top 500 list (linked above). Just imagine what Hammer will do!
Whoever modded the parent post "insightful" hasn't got a clue.
-- ___
If you think big enough, you'll never have to do it.
Re:Athlon MP 2000 cluster
by
rotwhylr
·
· Score: 1
Does your.sig imply that idiots avoid small groups of people?
First of all, we don't even know what the hardware architecture is going to look like, exactly. We're guessing it will have to be some sort of cluster of multi-cpu nodes, but we don't know how many nodes there will be, or how many cpu's per node.
The Swedish super cluster you refer to is a nice piece of engineering, but even it has its limits. If you want to link 16000 cpu's together in a useful way, 2- or 4-way nodes are going to be marginally useful at best. More cpus per node would be one possible answer.
Scaling nodes in a cluster and scaling CPUs in a node are two different things. Last I heard, Linus was working on increasing the SMP scalability of the Linux kernel, but I don't think that more than 4 to 8 cpus per node is available yet. (Is IBM helping him here? They've got lots of experience with this, and they're always talking about Linux...)
Anyway, my thought was that to scale to that number of cpus, they would probably need heavier nodes, and that would require them to adapt an existing SMP or NUMA architecture to fit their needs. Ergo, Linux may not be the best OS for them.
-- -- Windows is not simply installed on a computer; it is inflicted.
Linus was working on increasing the SMP scalability of the Linux kernel, but I don't think that more than 4 to 8 cpus per node is available yet.
The largest amount of SMP AMD has demonstrated for Opteron is 4-way. It's a pretty good bet that Linux can easily handle that.
If you want to link 16000 cpu's together in a useful way, 2- or 4-way nodes are going to be marginally useful at best.
Quite to the contrary. A cluster of 16000 CPUs is very, very useful, without any form of SMP. In fact, some degree clustering is unavoidable for that many nodes. The question is whether SMP is useful at all in that kind of environment, and it may well not be: it actually complicates software (because you have now have to write for two levels of parallelism) and it may not help you much (because achiving the kind of necessary data locality may be hard). If you have the money, a cluster of 16000 single CPU machines is usually better than a cluster of 8000 dual CPU machines.
Anyway, my thought was that to scale to that number of cpus, they would probably need heavier nodes, and that would require them to adapt an existing SMP or NUMA architecture to fit their needs
Well, even assuming that your spurious reason was correct--what OS are you talking about? Come on, tell us!
It's a pretty good bet that Linux can easily handle that.
Actually, it turns out, it's not just a bet, it's a fact. AMD demonstrated
Opteron 4-way multiprocessing using Linux. It seems the only other
OS they have demonstrated is Windows with SMP, but only with 2-way multiprocessing.
Re:Athlon MP 2000 cluster
by
rotwhylr
·
· Score: 1
1 - Each node requires a certain amount of CPU power, memory, and other resources just to run its OS. By using SMP nodes, you get economies of scale in terms of kernel footprint and overhead.
You mention data locality; this is precisely one of the advantages of a more robust node. Fewer (larger) nodes means that there is a greater chance that the data is already local, avoiding the need to go out over the cluster interconnect to get it. Which leads to the next point...
2 - This will require an incredibly robust interconnect. Building such an interconnect that delivers screaming bandwidth and minimal latency efficiently will not be cheap. As node count goes up, so does cost/complexity of the interconnect.
3 - (Disclaimer: I am not a developer, nor do I play one on TV. However, I have the following from a trusted source.) Writing multithreaded, multi-proc aware s/w IS somewhat more difficult than plain vanilla coding, but it's not THAT much harder. I am sure that whatever engineers they involve in a project like this will have the necessary skills.
4 - Actually, I don't know.what OS they could use for this beast; that was one of my questions on an earlier post. I recently spent some time speaking with an engineer who was intimately involved with SGI's high end stuff, and based on what he said, I wouldn't be surprised to see a Cray-related OS used here.
Although, if some of those Cray engineers started toying around under the hood of a Linux kernel... who knows.
-- -- Windows is not simply installed on a computer; it is inflicted.
glad cray ditched the itanium
by
Billly+Gates
·
· Score: 4, Informative
Itanium is a dog. They are a bitch to program in and optimize. THis means the compilers for it suck and will take years before optimized code will begin to take hold. I still prefer risc like the alpha chips because they are easy to program and optimize with. Not to mention very fast. But I understand why cray switched due to the death of the processor. Samsung still sells supercomputers with alpha's even though it will be dead shortly.
Remember that the customers who purchase these bad boys hire their own software engineers and purchase specialized compilers for maximum optimization. All the compilers will be available for Amd hammer chips because they run on so many systems. Also more engineers know it inside and out and can write great optimized programs for it.
Re:glad cray ditched the itanium
by
Anonymous Coward
·
· Score: 0
And what languages do you code in?
Admit it, you haven't a clue what you are talking about. Do you even know what a RISC processor is?
Well excuse me...
by
Anonymous Coward
·
· Score: 0
CowboyNeal, What does non-free-as-in-beer mean? Isn't the term "free as in beer" ? If it's non(free as in beer), then we might be talking about some thing like the GPL right? clue me in...
-- Thilaa Amaa Fui
this might be interesting
by
asscroft
·
· Score: 1
Mostly I'm posting so that I can easily find this thread again and see what WSJ has to say about it. Maybe they'll politely ask/. to remove it, but what if google has already cached it?
Staying tuned..
-- because I have been enjoined by this Holy Office to abandon the false opinion which maintains that the Sun is the centre
Re:this might be interesting
by
handsomepete
·
· Score: 1
(Ditto on the place marker)
I wonder if they'll ask him to return his 4 points of Karma (i.e. the profit) he earned from reposting their material...
I wonder if they will read "Tom's Hardware Guide" when they get power supplies for this puppy ?
-- Semper ubi sub ubi
Re:Sandia's reliance on supercomputers make me ner
by
DeathPenguin
·
· Score: 1
We should begin testing nukes again, just not on Earth...
All of our research for the past many years has been strictly theoretical. I don't like that, but at the same time it's obvious that nuclear testing causes irreversable damage to the environment and contaminates the air. Would it be possible to launch nukes at Mars or something? We need to melt those ice caps anyway... It would take a while for them to reach their destination, but at least we'd kinda-sorta have an idea of how well they perform.
Besides, we need to protect ourselves from those damn aliens that keep vandalizing our crops!
Maybe Red Storm would be a computer for the Chinese government. After all, they already got Red Flag going.
Let's be scientific... was Re:Framerate
by
HuguesT
·
· Score: 1
OK, my monitor's refresh rate is 85Hz, but that's for the *whole* frame. If the game framerate is higher than that you still might be able to detect a difference because *parts* of the frame are redrawn quicker.
My theory is that people might be able to tell the difference between 80 and 100 Hz (say) when they move around a lot and tearing int the image is occuring.
If you think this is crazy because movies give the impression of smoothness at 25Hz, in fact each image at the movie is 4x oversampled, so what you see on screen at the theatre is 100Hz.
As for analog sound vs. digital I suspect you mean LP vs CD? I can play something digital for you that will sound absolutely horrid (8kHz sampling, mono, MP3 compressed to 16kbps).
Some people will *swear* that they think their precious LP sound better than the same record in CD format. I can't tell if this is baloney or not, I just don't know. I do know that a lot of people complain of even high-quality MP3 and I can't tell the difference between CD and MP3 above 128kbps, so I can't judge. Can you?
Re:Let's be scientific... was Re:Framerate
by
Fulcrum+of+Evil
·
· Score: 2
My theory is that people might be able to tell the difference between 80 and 100 Hz (say) when they move around a lot and tearing int the image is occuring.
Sorry, but double buffering eliminates tearing completely. The buffers are switched in the vertical blanking interval.
-- "We returned the General to El Salvador, or maybe Guatemala,
it's difficult to tell from 10,000 feet"
Re:Let's be scientific... was Re:Framerate
by
HuguesT
·
· Score: 1
If that were the case then the frame rate would be limited to the refresh rate. Re-read what you just wrote.
What you describe is correct only if you tick the little box: `sync display with refresh' in the driver's preferences, which does limit your framerate.
What's a TeraFLOP? A trillion (a million million) Floating OPerations... per second?... ever??
Re:TeraFLOP?
by
Anonymous Coward
·
· Score: 0
the correct term is teraFLOPS, (note 's' on the end), meaning 10^12 Floating Point Operations Per Second.
some kind of hypertransport switch??
by
Anonymous Coward
·
· Score: 0
according to this at the cray site http://www.cray.com/news/0210/sandia_redstor m.html
"...Opteron(TM) processors connected via a low-latency, high-bandwidth, three-dimensional mesh interconnect network based on HyperTransport(TM) technology..."
would this mean it would use hypertransport to connect all of the processors???? I thought the max for HyperTransport was 8 processors. The only way to go beyond this would be to use some kind of hypertransport switch? any ideas?
I was trying to do 100 trillion flops on my cluster, and it was like beep beep bleep bleep, and then, like, half of my data was gone, and I was like, "What?" It couldn't do it. It was totally good data. It was a bummmer.
As another poster pointed out, Athlon competes well with P4 right now. They may not be faster, but they are as fast. That was not the case with k6 vs. Pentium 2. k6 was beaten very badly.
Athlon was a revolutionary cpu, which got everybody excited long before it was released. It would obviously be much faster than what Intel offered unless it was too late. Fortunately, it was not too late. OTOH who gets too excited about Opterons are probably lost in "64 bit" hype. All hammer cpus are basically Athlons with 64bit, SSE2 ISA extensions and an integrated memory controller. Even though the performance increase associated with additional registers, double precision vector processing and lower memory latency is quite a bit, the Athlon to Opteron step is definitely evolutionary, comparable to evolution of Athlon classic to Athlon XP. Athlon is AMD's Pentium pro, they will milk it until they can't.
One may argue that since Athlons can compete with P4s, the fact that hammers are evolved Athlons is not really bad news. That is also my position. One may also argue that since Athlons are not scaling as good as P4s and Opterons offer one-time performance increase (after all, you can't add new instructions every full moon), the future is not good for AMD. That is also a valid argument. What IMHO is an invalid argument is saying that the situation is like back when amd was designing Athlons; the difference is too much to ignore.
--
Gentlemen, you can't fight in here, this is the War Room!
Re:Sandia's reliance on supercomputers make me ner
by
Anonymous Coward
·
· Score: 0
Not really. True, much of the emphasis of research at the national labs has changed from new designs to management of the stockpile, the driving force behind the ASCI program is to replace nuclear testing with computer modelling, both for new and existing designs. The new designs are surely aimed at a 'tactical' use, small and as 'clean' as possible. I think even GW realizes there isn't much use for multi-megaton devices nowdays, as much as I'm sure he'd like to try one out.
Slashdot Mirror
There's something comic-bookish about that name... maybe it's just cos it sounds like a Transformer.
Hokey statistics and ancient misconceptions are no match for a good thought in your head, kid!
Dang! Just imagine a Beow--
Ah, never mind.
(Shuffle, shuffle.)
I watched C-beams glitter in the dark near the Tannhauser gate.
Ok I'm not going to use any words that even resemble beowulf or cluster in this post anywhere!
Hopefully AMD will be able to produce a quality product that far down the road. They've been bleeding money at an alarming rate and some say they're focusing too much on marketing instead of engineering- much like 3dfx did back in the day. Let's hope this pans out- my framerate could definitely be more killer.
Bad joke. Sorry.
I'm the Devil the Windows users warned you about.
This sort of thing must just be braking all the classic Crayons hearts. I mean, people were getting upset when Cray started building the T3 series Alpha based stuff, nowadays they are cooperating with Dell and making AMD based clusters. At least they have a new vector machine coming out soon.
I wonder if it will be as fast as the computer William Shatner and Priceline.com use?
...building a supercomputer as a cluster of commodity off-the-shelf personal computers, interconnected with a local area network technology like Ethernet, and running programs written for parallel processing out of those!
I wonder if those are PR ratings that they're throwing out!
If the Opteron's anything like my T-bird, that supercomputer's going to melt a hole to the center of the earth a la Chernobyl.
I wonder what kind of framerates UT would get on this thing. Of course, they would have to use software rendering, but it'd still be dang fast.
Cray and Sandia say it is a 40 tera*OP* system, not a 100 teraflop one. See what Cray says here and what Sandia says here The really interesting thing is not the processor, but rather the interconnect which seems to be very similar to the torus used in the T3E.
In other supercomputing news, check out what NERSC is proposing for their Earth Simulator Response Proposal. It's a 160 teraflop machine...
Do you know why the road less traveled by is littered with the bones of the unwary?
http://www.ccs.ornl.gov/PR/craytest.html
I've been wondering if it was time to buy AMD's stock yet. I was going to wait until they were closer to the release of their new 64 bit chips, but I might have to jump on the hypewagon a little earlier.
I really need to know, what's the difference between free and "free as in beer" ?
Please explain.
"Cray Chairman and CEO Jim Rottsolk said Red Storm reflects Cray's strategy to deliver high-efficiency, high bandwidth supercomputer systems. "Red Storm embodies the same design philosophy as our new Cray X1(TM) vector-based product in a highly cost-effective superscalar architecture and will be a key initiative for Cray."
Quoted from the Cray Press Release.
Ah, I remember my days on the venerable Cray Y-MP, optimizing my programs for vector processing. I am unsure how Cray has managed to make a combined parallel-vector machine like the Y-MP out of PC chips provided by AMD, but I do not envy the programmers who must now begin the task of vector-optimizing their code to take advantage of this beast.
I had hoped that this idea died with Cray. Apparently not.
Dr. Joseph Hairston
Superintendent, CCBC
Now, I know that my little ol' Athlon runs hot as a mother, so I can't imagine the cooling necessary to keep this baby running at an optimal temperature. Last I heard (and I could be mistaken), Crays were cooled by bring submerged in liquid nitrogen, and more recently with some sort of liquid plasma cooling (don't ask me, I have NO idea how something like that would work). Does anyone have any information on how they're going to keep this thing from incinerating itself the moment it's turned on?
Just so we can nuke 11 million chinese with one warhead instead of only 10 million?
Cry me a river.
The machine, code-named Red Storm, will require over 16,000 microprocessors to achieve that performance level, according to a researcher quoted in the Journal report.
This can be rephrased easily into
The machine, named code-red Storm, will require over 16,000 microprocessors' performance level to archive that, according to a researcher quoted in the Journal report.
oh well.
Once again, AMD missed their chance. They should have named the K8 after the male lead on the robot soap-opera that Bender watches on Futurama:
"Calculon! We thought you were dead!"
We reserve the right to serve refuse to anyone. -management
Maybe now AMD won't go the way of the buffalos... or was that buffalo? No, not buffalo: Apple.
resume nuclear testing b/c we really don't know what they do...
Nuclear testing is sorta pointless. Nuclear weapons cause long-term irreversable destruction and human death.
I really don't see a need to restart testing.
These countries that have nuclear weapons programs currently, aren't going to be deterred by us testing them. They are going to be deterred by us using them (which is really not going to happen in the forseeable future).
Mod parent down please.
This will be really cool because I would imagine this will be water cooled like Cray's other computers. I can see the online adds now.. "Personal water cooling setup. As seen on the Cray Red Storm"
Indeed, things do not look very good for AMD. Is there any AMD believers to explain how they will survive through next year? A $254 million loss in a quarter is not very convincing. Apparently, the have had to take a huge risk with putting the money in the design of this "new generation". Is it good enough?
computer on earth!!
1. cluster lots of opterons
2. place popcorn on top
3. sell popcorn and cycles
4. profit
I'd like a beowulf cluster of those...
does it play quake?
yeah, but how much longer before it is DRM-enabled to run only MS....
etc..
because I have been enjoined by this Holy Office to abandon the false opinion which maintains that the Sun is the centre
Have they ported Linux, or one of the BSDs to it yet, though?
:-)
A lot of people don't like coding for Unicos, and think how fast Tux Racer would run on it
AMD has been suffering recently because they are focusing their efforts on Opteron, and meanwhile Athlon isn't getting any younger. This is a temporary setback; if Opteron is any good then AMD will be performance competitive again, allowing them to sell at higher price points and get a better margin.
AMD was in the same funk back when Intel released the Pentium II and AMD was still working on Athlon. Once AMD got Athlon out the door they started doing a lot better.
Since the expression "100 teraflop" contains no time related unit, it probably refers to the total number of flop that will be delivered by the system running at 40 teraflop/s before its predicted meltdown after 2.5 s.
I have seen the expression "Free-as-in-Beer" a few times on Slashdot, yet I have no idea what this means. I have scoured my head to think of what it could mean, but can't figure it out.
What does "Free-as-in-Beer" mean?
Free beer, to me, would seem to be the greatest thing in the history of the world!
I know, you need that 100 frames/sec when playing doom 3 even though the human eye can only see up to 50 frames/sec.
and come over to my house, and I will demonstrate.
LETS DECOMPOSE & ENJOY ASSEMBLING
AMD's New Opteron Chips
Are Tapped for Red Storm
By DON CLARK
Staff Reporter of THE WALL STREET JOURNAL
Endorsing the technology of one of Intel Corp.'s key rivals, Sandia National Laboratories and Cray Inc. plan to build a massive supercomputer using a soon-to-be-introduced line of microprocessor chips from Advanced Micro Devices Inc.
The development project, estimated in June to cost $90 million, is a high-profile vote of confidence for AMD's new Opteron chip, in a small but prestigious market long dominated by other chip suppliers. It represents a missed opportunity for Intel, which has been targeting its new Itanium line at high-performance computing applications.
Red Storm, Sandia's name for the new supercomputer, also marks a step forward for the U.S. effort at leadership in supercomputers, which suffered a blow this year with the completion of a huge machine called the Earth Simulator by Japanese government agencies and NEC Corp. Where recent U.S. machines have largely been constructed out of components used in commercial computers, Cray is expected to develop special technology for connecting the AMD chips that should make Red Storm suited for more-complex scientific problems.
"This is a move away from commodity components," said Horst Simon, division director of the National Energy Research Scientific Computing Center, a supercomputer facility affiliated with Lawrence Berkeley National Laboratory. "It's very exciting."
Sandia, which does research for the U.S. Department of Energy in Albuquerque, N.M., and Livermore, Calif., has a performance goal of 100 trillion operations per second for Red Storm. It hasn't disclosed most technical details, including the chip selection. But Mr. Simon estimated that the machine will require 16,000 or more microprocessors to hit its speed target, which would appear to surpass the Earth Simulator's current performance.
Sandia said in June that it had selected Cray, a longtime supercomputer maker based in Seattle, to negotiate a development contract. Cray and Sandia officials didn't return calls seeking comment Friday. AMD and Intel officials declined to comment.
AMD could use some good news. The company's Athlon chip line, mainly used in personal computers, has been falling behind the performance of comparable Intel chips. The company reported last week a third-quarter loss of $254 million on sales of $508.2 million, off 34% from the year-earlier period.
Opteron is a high-end member of the new line, code-named Hammer, that is due out next year and viewed by analysts as AMD's best hope for recovery. Like the Itanium, Hammer chips are designed to process 64 bits of information at a time, instead of 32 bits, a capability that helps run huge databases and solve scientific problems.
Intel's Itanium line, developed over eight years with help from Hewlett-Packard Co., is based on an entirely new architecture and achieves its best performance on new 64-bit programs. AMD, by contrast, made 64-bit additions to the original Intel technology used in the past by both companies.
The difference, AMD says, allows Hammer-based computers to run both 32-bit and 64-bit software at high speed. AMD released preliminary test results last week for Opteron -- so far not validated by outside researchers -- that show the chip exceeding Intel's latest Itanium 2 model on one of two widely-used speed measures, AMD said.
Itanium 2, introduced last summer, has already been selected for at least a half-dozen high-performance installations. The Pacific Northwest National Laboratory, another Department of Energy facility, is building a $24.5 million system based on 1,400 Itanium 2 chips. Based on past Sandia announcements, the Red Storm project's stated performance goal is more than 10 times that of the Pacific Northwest project.
Write to Don Clark at don.clark@wsj.com
man what is up with you fucking slashbots today
The "(US Dept. of Energy) has just built a 100 Teraflop super computer employing AMD's Opteron (Hammer) processors", causing rolling blackouts. Rummor has it that it isn't the power consumption by all of the procesors that is causing the blackouts, rather it is all of the power needed to keep all of the industrial strength fans and nitrogen cooling systems running. Citizens of the area are asked to turn off all their power consuming items (please use candles) so that we can see how this moster machine performs.
... you can't afford power. Watch TV at a buddy's house ... it's cheaper!
Besides, you're all unemployeed
It seems like Linux is the most likely choice for an operating system for this, but they don't mention anything. Does anybody know?
We have gotten quite accustomed to the very cool and efficient operation of our high density VLSI microprocessors. How soon we forget the days when computers were constructed not of VLSI devices, nor even discrete component semiconductors, but good old tubes! Those babies got HOT! I can't imagine cooling on the scale of a supercomputer is any more difficult than cooling the old discrete component and tube computers had been. I would guess the primary challenge is finding a cost efficient cooling technique that efficiently absorbs the heat from all of those CPU chips.
s s i a
why is everyone so fucking retarded today
This is exactly the same as the first post. And yet this is "funny" and the first post is "redundant". Are the /. moderators so fast that they move backwards through time?
Hopefully AMD will be able to produce a quality product that far down the road. They've been bleeding money at an alarming rate and some say they're focusing too much on marketing instead of engineering- much like 3dfx did back in the day.
I donno where you've been, but, "back in the day", 3DFX was the only 3d card manufacturer, and the products they produced were light-years ahead of their competition (who WAS their competition? S3? Nvidia? Remember how hard Nvidia sucked before they bought 3DFX?)
What's even more damaging to your argument is that, back in the day, the only CPU manufacturer worth a damn was Intel. Unless you've been living under a rock for the past few years, you know AMD's current line of CPUs is as competitive and high-quality as anything else in the PC arena.
In conclusion, stop being retarded.
- A.P.
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
I hope they are not putting that supercomputer anywhere near Antarctica.. 16,000 Athlons can get pretty hot resulting in some serious polar melting forcing me to evacuate from New Orleans (we are already underwater).
I'm a subscriber. Here's the article. I hope it convinces you,too, to subscribe.
Thanks for the article, but sorry, your valiant effort does not convince me.
Tat Tvam Asi
Why don't you just do as the rest of the world and use the word 'gratis' for something that doesn't cost any money? It fromes from the latin 'gra', which means 'free' and 'tis', which means 'as in beer'.
Sould it be:
Please Hammer, don't hurt 'em
or
Imagine a Beowolf Cluster of these?
Nah, how about:
Please Hammer, don't imagine a Beowolf cluster of these.
(I gave up mod points for this comment? Sheesh, I must be smoking crac....errr...oops)
Have you read the moderator guidelines? Well, have you, PUNK? (and I want a Karma: Gnarly option)
I was going to subscribe, actually. But now you posted it here, violating WSJ's copyright, I don't need to anymore. Thanks.
According to a tour guide, the powers that be at the national labs thought that "teraflop" sounded like a really, really bad mistake, so they decided to call a teraflop a teraop instead. In other words, the phrases refer to exactly the same thing.
I don't know what's funnier, that post or the moderations for it. Heh.
here on Cray's website.
Escher was the first MC and Giger invented the HR department.
I don't think you understand the point of nuclear weapons testing.. mainly it's to ensure that current weapons are still operational and not, say *malfunction*. It takes a vast amount of effort to keep weapons operational, especially those that contain decades old technology. Without going into the politics of whether they should exist or not, it is in everyone's best interest to make sure they don't introduce problems of their own.
... a Beowulf cluster of these!
I'm not even sure what that means.
How can you melt a ball of burning gass?
autopr0n is like, down and stuff.
Its going to take four years for the machine to reach full speed, and you like to use several years. Only the most established names can qualify- IBM & Intel. SGI, HP(alpha), and AMD are contenders, but I dont know if they have the financial stability.
Copyright is a figment of a diseased and avaricious imagination.
Eh, the Utahians actually own most of that desert, in preparation for the day that California slides into the ocean.
Parent is troll, or else trying to be funny but isn't. I'll bite anyway.
Basically, Sandia is betting the safety and sanctity of the Free World on computer simulations... Basically, you're saying the safety of the free world rests on our ability to blow the rest of the world off the face of the earth with nukes. But you know what? We can do that already. We have enough ICBMs to kill the "godless commies" a hundred times over, and i have no idea why we are doing more testing unless it's for some sort of intercepter. Now those have to be modeled first, and you can't test them anyway because that would require an atmospheric blast (bad).
Please, contact your local senator and representative, explain the dire need for the US to resume nuclear testing to prove that we have a valid, proven deterrent. See above, for why this is unnecessary. Also recall the so-called "Nuclear Test Ban Treaty." I wonder what that does....
In conclusion, the parent is a troll or a very stupid bigot. Let's hope it's the former.
I hereby place the above post in the public domain.
"Just so we can nuke 11 million chinese with one warhead instead of only 10 million?"
The US isn't developing new warheads or ICBMs. The stock we have we will live with. By the middle of this century, some of these warheads will be close to 100 years old. Sandia is busy simulating warhead decay.
Despite popular perception of vast quantities of cheap devices surrounded by evil Generals slobbering for a chance to use them, nuclear warheads are very complex and require intense maintenance by sober professionals. Making sure these weapons actually function is a big part of the freedom you enjoy, whether or not you actually live in Sandia's country.
Maw! Fire up the karma burner!
I donno where you've been, but, "back in the day", 3DFX was the only 3d card manufacturer, and the products they produced were light-years ahead of their competition (who WAS their competition? S3? NVidia? Remember how hard nVidia sucked before they bought 3DFX?)
Yeah, NVidia sucked, that's why they had enough money to buy 3dfx. Yup.
3dfx was way ahead of the other 3d accelerators for the first few revisions, but their "who gives a shit about image quality" attitude, and complete lack of innovation killed them long before nVidia purchased them.
Basically 3dfx had a killer product, and spent all their time doing nothing but making it faster, while nVidia and others concentrated on technological advancements, such as T&L engines, and the like. 3dfx died because they couldn't compete technically.
Hrm... I have a feeling you're a troll, but whatever.
autopr0n is like, down and stuff.
Invoicing, Time Tracking, Reporting
yawn. troll.
I'm sure most of you who've programmed have felt that way, until you tried to compile and execute your program (real programmers, I'm not talking interpreted baby languages like Basic or Perl), and then you were amazed by your simple syntax errors, and astonished by your modeling misconceptions
Well yeah, but that's just because I didn't bother to prove them....
autopr0n is like, down and stuff.
What part of the definition of science makes you think manually managing memory allocation and deallocation is better than having a smart algorithm for handling it?
What school or university did you attend (if any). Could be usefull so tat I know where not to send my kids.
unfinished: (adj.)
Perhaps we should name it after something simple, yet powerful...
Perhaps a Norse god, or legendary Scandinavian warrior?
--
Internet Explorer (n): Another bug -- that is, a feature that can't be turned off -- in Windows.
Guess he was too busy sucking you cock to spell-check.
It will be real interesting to be at local chamber of commerce meeting where Sandia Labs management gets to meet with managment from another big employer in Albuquerque.
That's right boys and girls.
On the west side of the Rio Grande is Rio Rancho, home of Intel Fab 9. (the same one that got struck by lightning a while back).
"Provided by the management for your protection."
Here is a 256-CPU IA-64 Linux cluster, #53 on the top 500 list. And here is CPlant, #50. You can find more Linux boxes on the top 500 list
___
If you think big enough, you'll never have to do it.
> Nuclear testing is sorta pointless.
You'd quickly change your mind if a dud nuke
failed to deflect an asteroid.
> Nuclear weapons cause long-term irreversable
> destruction and human death.
Well, Doh! That's what they're *for*.
Really, if we're going to keep a stock pile, we
should test them. I suggest testing them by
building a big glass tunnel underneath Costa Rica
connecting the Pacific to the Atlantic, so that
China (via the PLA proxy, Li Ka-ching's Hutchinson
Wampoa), doesn't control the *only* canal.
-I like my women like I like my tea: green-
Whoever modded the parent post "insightful" hasn't got a clue.
___
If you think big enough, you'll never have to do it.
Remember that the customers who purchase these bad boys hire their own software engineers and purchase specialized compilers for maximum optimization. All the compilers will be available for Amd hammer chips because they run on so many systems. Also more engineers know it inside and out and can write great optimized programs for it.
http://saveie6.com/
CowboyNeal, What does non-free-as-in-beer mean? Isn't the term "free as in beer" ? If it's non(free as in beer), then we might be talking about some thing like the GPL right? clue me in...
--
Thilaa Amaa Fui
Mostly I'm posting so that I can easily find this thread again and see what WSJ has to say about it. Maybe they'll politely ask /. to remove it, but what if google has already cached it?
Staying tuned..
because I have been enjoined by this Holy Office to abandon the false opinion which maintains that the Sun is the centre
I wonder if they will read "Tom's Hardware Guide" when they get power supplies for this puppy ?
Semper ubi sub ubi
We should begin testing nukes again, just not on Earth...
All of our research for the past many years has been strictly theoretical. I don't like that, but at the same time it's obvious that nuclear testing causes irreversable damage to the environment and contaminates the air. Would it be possible to launch nukes at Mars or something? We need to melt those ice caps anyway... It would take a while for them to reach their destination, but at least we'd kinda-sorta have an idea of how well they perform.
Besides, we need to protect ourselves from those damn aliens that keep vandalizing our crops!
Why didn't they instead call it VOLTRON!
Voltron was first with 5 lions, and then with about 15 space vehicles.
"Form legs, and body!"... "ROWRRR!"
"Form arms, and I'll form the HEAD!"
CLANG! (as hands meet to form sword) ZZZRRRRR!
ROARRRRR!!!!!
Must be taking out a hell of a Insurence policy with that big of a fire hazard in their lab :-)
Imagine a Beowolf Cluster of THESE!!!
Every day? I'd really love the answer to that one. But alas it's just the way it is.
A BEOWULF CLUSTER OF THESE!!
MARIJUANA, SHROOMS, X: ONLINE?! - E
An x86-based system that can run Doom3 at 9999 fps!
will the red storm be the fastest once it's completed? 100Tflops is like 100000Glops? That exceeds #1 on top500.
"build a 100 Teraflop super computer employing AMD's Opteron (Hammer) processors"
In other news...scientists predict a 10 degree average temperature increase on the West coast this winter.
It's 10 PM. Do you know if you're un-American?
Maybe Red Storm would be a computer for the Chinese government. After all, they already got Red Flag going.
OK, my monitor's refresh rate is 85Hz, but that's for the *whole* frame. If the game framerate is higher than that you still might be able to detect a difference because *parts* of the frame are redrawn quicker.
My theory is that people might be able to tell the difference between 80 and 100 Hz (say) when they move around a lot and tearing int the image is occuring.
If you think this is crazy because movies give the impression of smoothness at 25Hz, in fact each image at the movie is 4x oversampled, so what you see on screen at the theatre is 100Hz.
As for analog sound vs. digital I suspect you mean LP vs CD? I can play something digital for you that will sound absolutely horrid (8kHz sampling, mono, MP3 compressed to 16kbps).
Some people will *swear* that they think their precious LP sound better than the same record in CD format. I can't tell if this is baloney or not, I just don't know. I do know that a lot of people complain of even high-quality MP3 and I can't tell the difference between CD and MP3 above 128kbps, so I can't judge. Can you?
You might as well quote benchmarks for AMD's new 8Ghz Unobtaineron; neither it nor any Athlon faster than the 2400+ are actually available.
What's a TeraFLOP? A trillion (a million million) Floating OPerations ... per second? ... ever??
according to this at the cray siter m.html
http://www.cray.com/news/0210/sandia_redsto
"...Opteron(TM) processors connected via a low-latency, high-bandwidth, three-dimensional mesh interconnect network based on HyperTransport(TM) technology..."
would this mean it would use hypertransport to connect all of the processors???? I thought the max for HyperTransport was 8 processors. The only way to go beyond this would be to use some kind of hypertransport switch? any ideas?
I was trying to do 100 trillion flops on my cluster, and it was like beep beep bleep bleep, and then, like, half of my data was gone, and I was like, "What?" It couldn't do it. It was totally good data. It was a bummmer.
My Cray's never let me down once.
www.cray.com/switch
Athlon was a revolutionary cpu, which got everybody excited long before it was released. It would obviously be much faster than what Intel offered unless it was too late. Fortunately, it was not too late. OTOH who gets too excited about Opterons are probably lost in "64 bit" hype. All hammer cpus are basically Athlons with 64bit, SSE2 ISA extensions and an integrated memory controller. Even though the performance increase associated with additional registers, double precision vector processing and lower memory latency is quite a bit, the Athlon to Opteron step is definitely evolutionary, comparable to evolution of Athlon classic to Athlon XP. Athlon is AMD's Pentium pro, they will milk it until they can't.
One may argue that since Athlons can compete with P4s, the fact that hammers are evolved Athlons is not really bad news. That is also my position. One may also argue that since Athlons are not scaling as good as P4s and Opterons offer one-time performance increase (after all, you can't add new instructions every full moon), the future is not good for AMD. That is also a valid argument. What IMHO is an invalid argument is saying that the situation is like back when amd was designing Athlons; the difference is too much to ignore.
Gentlemen, you can't fight in here, this is the War Room!
Not really. True, much of the emphasis of research at the national labs has changed from new designs to management of the stockpile, the driving force behind the ASCI program is to replace nuclear testing with computer modelling, both for new and existing designs. The new designs are surely aimed at a 'tactical' use, small and as 'clean' as possible. I think even GW realizes there isn't much use for multi-megaton devices nowdays, as much as I'm sure he'd like to try one out.