Intel Shows Off 80-core Processor 222
Posted
by
Zonk
from the next-up-a-skillion-core-system dept.
from the next-up-a-skillion-core-system dept.
thejakebrain writes "Intel has built its 80-core processor as part of a research project, but don't expect it on your desktop any time soon. The company's CTO, Justin Rattner, held a demonstration of the chip for a group of reports last week. Intel will be presenting a paper on the project at the International Solid State Circuits Conference in San Francisco this week. 'The chip is capable of producing 1 trillion floating-point operations per second, known as a teraflop. That's a level of performance that required 2,500 square feet of large computers a decade ago. Intel first disclosed it had built a prototype 80-core processor during last fall's Intel Developer Forum, when CEO Paul Otellini promised to deliver the chip within five years.'" Update: 06/01 14:37 GMT by Z : This article is about four months old. We discussed this briefly last year, but search didn't show that we discussed in February.
But... (Score:2, Funny)
Re:But... (Score:5, Funny)
Yeah. 80 different distributions at once.
Finally (Score:2)
Intel has found a useful argument they could give to illustrate their newest
" s/number of GHz/numer bor cores/ "
marketing propaganda.
Re:But... (Score:2)
Seriously, this thing doesn't even support the x86 instruction set (yet?), and is not form-factor compatible with today's processors.
When it finally comes out, hopefully they will have integrated the "3D" chip-stacking tech into it as well, and HOPEFULLY it will run both cooler and with less power requirements than today's offerings.
cue (Score:5, Funny)
Re:cue (Score:3, Funny)
Re:cue (Score:2)
Re:cue (Score:2)
Re:cue (Score:3, Funny)
Re:cue (Score:2)
Re:cue (Score:2)
Re:cue (Score:2)
No the Mac Pro isn't the only system with 8 cores (2 4 core cpus) it is a common system that has 8 Cores that most people know about. It is there to ilistrate that systems are already out today that support the optimal CPU usage for Vista for people with such configurations to prove me right or wrong. (Checking CPU Useage of vista running) while doing some mostly OS tasks.
Most programams are not very multi-threaded or paralized, which is the problem with Multi-Core systems if you have 2,4,8,16,32,64... CPUs your performance may not in good relation and for most cases will be the speed of one processor.
Re:cue (Score:2)
If you have info on current costs of 8-way x86-64 boxen, pls feel free to contact me offboard.
Older Story (Score:3, Informative)
Sure would be nice to have a play with it once they have worked out how to program it...
Re:Older Story (Score:3, Insightful)
But with single-thread performance growth at a virtual standstill, Moore's law is going to result in exponential growth in the number of cores, whether or not we're ready to write software for them.
I wonder if we won't move towards a more "biological" paradigm - massive parallelism, but with massive redundancy and therefore inefficiency from a computational standpoint, but also robustness to hardware and software bugs.
Re:Older Story (Score:2)
Re:Older Story (Score:2)
IA64 (Score:5, Insightful)
It didn't work out too well for Intel.
Re:IA64 (Score:5, Funny)
I guess we all know how that one turned out.
Re:IA64 (Score:2)
Yeah, but (Score:2)
Re:IA64 (Score:2)
Re:IA64 (Score:3, Informative)
Re:IA64 (Score:2)
It may be known as "a teraflop", but... (Score:5, Informative)
The measurement is "FLOPS". Floating Point Operations Per Second. It's an acronym. The 'S' is part of the acronym. Hence even if you only have oneof them, it's still a FLOPS. And it's capitalised.
Strictly speaking it should be "trillion FLOPS" as well since it's not an SI unit but my pedantry is limitted.
Re:It may be known as "a teraflop", but... (Score:2)
Re:It may be known as "a teraflop", but... (Score:5, Funny)
One flop is enough for me... (Score:2)
Re:It may be known as "a teraflop", but... (Score:5, Insightful)
If we're going to be speaking strictly, get it right:
FLoating point Operations Per Second
Re:It may be known as "a teraflop", but... (Score:2)
Re:It may be known as "a teraflop", but... (Score:2)
Re:It may be known as "a teraflop", but... (Score:2, Interesting)
One floating point operation (say, an add), is a FLOP, not a FLO. Just like a No OPeration is a NOP (alternatively, NOOP, but assembly mnemonic is almost always NOP). If you want to know the rate at which a processor executes FLOPs, you say that it computes at X FLOPS.
Kind of like the chip itself. (Score:2)
my pedantry is limitted.
The average user won't notice. There's a threshold for everything, beyond which limits are meaningless.
beowulf cluster? (Score:4, Funny)
Re:beowulf cluster? (Score:5, Funny)
can you imagine.
Dude.... (Score:2)
Re:Dude.... (Score:3, Interesting)
Deja vu all over again (Score:4, Insightful)
Stop me if you've heard this one before...
OK: "Stop, for the love of God!" (Score:2)
Re:Deja vu all over again (Score:4, Funny)
Great now, I have to wait five years before I buy my next computer because nothing else will compare unless it's got 80 cores. My duel core looks soooo small now. =(
Re:Deja vu all over again (Score:2)
core 2 duo has a higher transistor density? (Score:2, Insightful)
Re:core 2 duo has a higher transistor density? (Score:3, Insightful)
transistor density? (Score:3, Informative)
80 cores means there are probably quite a lot of on-chip interconnects between the cores.
There has to be a typo hiding in there, but the whole thing is an empty set. It's hard to believe they can make 80 cores with 100E6 transistors when it take 261E6 transistors to make two. Each core would have less than a million transistors in the 80 core model. You have to go all the way back to the 486 [answers.com] to see that kind of count from Intel. It's possible because the cores are not x86, there's no "ability to use memory" and ... it's vapor ware. For the practical significance, they might as well have photographed a box of Pentiums and called it useful because Open Mosix does auto clustering and there are live CD versions. You've got a better chance of computing something with the box of Pentiums.
Bus space is not likely to be an issue either. It does not show up in this image of the cell processor. [anandtech.com]
Re:transistor density? (Score:2)
Yeah, when I read it I thought it must be a typo (especially given that the die area is bigger). Although, the article says it's VLIW (it isn't specific, but I guess it'll be IA64 based), which means you can throw away a shed-load of transistors from the scheduler that would have to be present on an OOE superscalar device.
Re:transistor density? (Score:2)
Re:core 2 duo has a higher transistor density? (Score:4, Informative)
(Btw, i fucking HATE the "millimeters squared" expression. Its square millimeter. 275 mm squared would be more than 65 cm^2.)
Re:core 2 duo has a higher transistor density? (Score:2)
Re:core 2 duo has a higher transistor density? (Score:2)
It's possible... (Score:4, Informative)
Another explanation might be that they didn't want to waste the time/expense to come up with an optimized layout, or that they intentionally spaced things out to make testing easier.
Frak everything, we're doing 80 blades (Score:5, Funny)
http://www.theonion.com/content/node/33930 [theonion.com]
Fuck Everything, We're Doing Five Blades
By James M. Kilts
CEO and President,
The Gillette Company
Would someone tell me how this happened? We were the fucking vanguard of shaving in this country. The Gillette Mach3 was the razor to own. Then the other guy came out with a three-blade razor. Were we scared? Hell, no. Because we hit back with a little thing called the Mach3Turbo. That's three blades and an aloe strip. For moisture. But you know what happened next? Shut up, I'm telling you what happened--the bastards went to four blades. Now we're standing around with our cocks in our hands, selling three blades and a strip. Moisture or no, suddenly we're the chumps. Well, fuck it. We're going to five blades.
Sure, we could go to four blades next, like the competition. That seems like the logical thing to do. After all, three worked out pretty well, and four is the next number after three. So let's play it safe. Let's make a thicker aloe strip and call it the Mach3SuperTurbo. Why innovate when we can follow? Oh, I know why: Because we're a business, that's why!
You think it's crazy? It is crazy. But I don't give a shit. From now on, we're the ones who have the edge in the multi-blade game. Are they the best a man can get? Fuck, no. Gillette is the best a man can get.
What part of this don't you understand? If two blades is good, and three blades is better, obviously five blades would make us the best fucking razor that ever existed. Comprende? We didn't claw our way to the top of the razor game by clinging to the two-blade industry standard. We got here by taking chances. Well, five blades is the biggest chance of all.
Here's the report from Engineering. Someone put it in the bathroom: I want to wipe my ass with it. They don't tell me what to invent--I tell them. And I'm telling them to stick two more blades in there. I don't care how. Make the blades so thin they're invisible. Put some on the handle. I don't care if they have to cram the fifth blade in perpendicular to the other four, just do it!
You're taking the "safety" part of "safety razor" too literally, grandma. Cut the strings and soar. Let's hit it. Let's roll. This is our chance to make razor history. Let's dream big. All you have to do is say that five blades can happen, and it will happen. If you aren't on board, then fuck you. And if you're on the board, then fuck you and your father. Hey, if I'm the only one who'll take risks, I'm sure as hell happy to hog all the glory when the five-blade razor becomes the shaving tool for the U.S. of "this is how we shave now" A.
People said we couldn't go to three. It'll cost a fortune to manufacture, they said. Well, we did it. Now some egghead in a lab is screaming "Five's crazy?" Well, perhaps he'd be more comfortable in the labs at Norelco, working on fucking electrics. Rotary blades, my white ass!
Maybe I'm wrong. Maybe we should just ride in Bic's wake and make pens. Ha! Not on your fucking life! The day I shadow a penny-ante outfit like Bic is the day I leave the razor game for good, and that won't happen until the day I die!
The market? Listen, we make the market. All we have to do is put her out there with a little jingle. It's as easy as, "Hey, shaving with anything less than five blades is like scraping your beard off with a dull hatchet." Or "You'll be so smooth, I could snort lines off of your chin." Try "Your neck is going to be so friggin' soft, someone's gonna walk up and tie a goddamn Cub Scout kerchief under it."
I know what you're thinking now: What'll people say? Mew mew mew. Oh, no, what will people say?! Grow the fuck up. When you're on top, people talk. That's the price you pay for being on top. Which Gillette is, always has been, and forever shall be, Amen, five blades, sweet Jesus in heaven.
Re:Frak everything, we're doing 80 blades (Score:2)
You'll love this then.
octaginator 8 blades [youtube.com]
-Reece
Re:Frak everything, we're doing 80 blades (Score:2)
http://www.washingtonpost.com/ac2/wp-dyn/A61952-20 03Jul15 [washingtonpost.com]
Blade Inflation
By Dave Barry
What's next from the razor-sharp minds of the shaving industry?
Attention, consumers with bodily hair: The razor industry has news for you! You will never in a million years guess what this news is, unless your IQ is higher than zero, in which case you're already thinking: "Not another blade! Don't tell me they're adding ANOTHER BLADE!!"
Shut up! Don't spoil the surprise for everybody else!
Before I tell you the news, let's put it in historical context by reviewing:
THE HISTORY OF SHAVING
Human beings are one of only two species of animals that shave themselves (the other one is salamanders). The Internet tells us that humans have been shaving since the Stone Age. Of course, the Internet also tells us that hot naked women want to befriend us, so we can't be 100 percent sure about everything we read there.
But assuming that www.quikshave.com/ timeline.htm is telling the truth, Neanderthal Man used to pluck his facial hairs "using two seashells as tweezers." No doubt Neanderthal Woman found this very attractive. "You smell like a clam," were probably her exact words. It was during this era that the headache was invented.
By 30,000 B.C., primitive man was shaving with blades made from flint, which is a rock, so you had a lot of guys whose faces were basically big oozing scabs. The next shaving breakthrough came when the ancient Egyptians figured out how to make razors from sharp metal, which meant that, for the first time, the man who wanted to be well-groomed could, without any assistance or special training, cut an ear completely off.
This was pretty much the situation until the late 19th century, at about 2:30 p.m., when the safety razor was invented. This introduced a wonderful era known to historians as "The Golden Age of Not Having Razor Companies Introduce Some Ludicrously Unnecessary New Shaving Technology Every 10 Damn Minutes."
I, personally, grew up during this era. I got my first razor when I was 15, and I used it to shave my "beard," which consisted of a lone chin hair approximately one electron in diameter. (I was a "late bloomer" who did not fully experience
puberty until many of my classmates, including females, were bald.) My beard would poke its wispy head out of its follicle every week or so, and I, feeling manly, would smother it under 14 cubic feet of shaving cream and lop it off with my razor. Then I would stand in front of the bathroom mirror, waiting for it to grow again. Mine was a lonely adolescence.
The razors of that era had one blade, and they worked fine; ask any older person who is not actively drooling. But then, in 1971, a very bad thing happened: Gillette, looking for a way to enhance the shaving experience (by which I mean "charge more") came out with a razor that had TWO blades. This touched off a nuclear arms race among razor companies, vying to outdo one another by adding "high-tech" features that made the product more expensive, but not necessarily better. This tactic is called "sneakerization," in honor of the sneaker industry, which now has people paying upwards of $200 a pair for increasingly weird-looking footwear boasting the durability of thinly sliced Velveeta.
Soon everybody was selling two-blade razors. So the marketing people put on their thinking caps, and, in an astounding burst of creativity, came up with the breakthrough concept of: THREE BLADES. Gillette, which is on the cutting edge (har!) of razor sneakerization, currently has a top-of-the-line three-blade razor -- excuse me, I mean "shaving system" -- called the "Mach3Turbo," which, according to the Gillette Web site (www.gillette.com) has more technology than a nuclear submarine, including "open cartridge architecture" and an "ergonomic handle" featuring "knurled ela
Exponential (Score:2)
Add 2 cores, you get 2x TFlops, add 4 -- get 16x, add 8 you get 256x TFlops. Why stop there, add 80 you get 2^80 = 1208925819614629174706176x more TFlops. Heck, add 1000 cores and you can simulate the universe! Well, I am putting my life's savings into Intel stock!
Re:Frak everything, we're doing 80 blades (Score:2)
It was an interesting parody. Perhaps not entirely on topic, but related.
And it was freakin' hilarious.
Re:Frak everything, we're doing 80 blades (Score:2, Insightful)
http://money.cnn.com/2005/09/14/news/fortune500/g
Re:Gillette Fusion (Score:2)
Fusion Turbo
Fusion Power
Fusion Power Turbo
Then we start all over again with 10 blades or something, ~doubling the price at each iteration.
Re:Gillette Fusion (Score:2)
Not only a dupe... but of an old story (Score:5, Insightful)
Not to mention that Slashdot (even Zonk) Covered this LAST YEAR [slashdot.org].
But that's OK, I'm sure Slashdot gave insightful and cogent coverage of real events that actually matter to geeks on this site, you know, like the Release of a new major version of GCC [gnu.org]
Oh wait.... that (like a bunch of other actually interesting stories) would be in the aptly-named, sir not appearing on this website category due to it not making enough banner revenue.
That's why we need 80 cores. (Score:3, Funny)
Imagine a... (Score:5, Funny)
AMD's response (Score:4, Insightful)
Besides, with most software being single-threaded I don't know if a consumer will immediately need more than 4 cores for a while. I can still see software companies trying to come up with ways to keep all 80 cores busy..."Well, they need at least 20 anti-virus processes, 10 genuine advantage monitors, and we'll install 100 shareware application with cute little icons in the task bar by default. There, that should keep all the cores nice and warm and busy -- our job is done!".
But in all seriousness, I would expect some extremely realistic environmental physical simulations (realtime large n-body interactions and perhaps realtime computational fluid dynamics)...now that's something to look forward to!
Re:AMD's response (Score:2)
I think one of the possible uses for an 80-core CPU that nobody's really talked about is multiple redundancy. If one core should somehow get fried, you have 79 left.
Re:AMD's response (Score:2)
Or... you could just have two and order a CPU to be delivered when one burns up. In fact that's what happens on some mainframes. If a part fries, the machine calls "home" and the company will send a replacement immediately. Sometimes the administrator will find out something went bad only when the replacement part already arrived at the door.
Re:AMD's response (Score:2)
Re:AMD's response (Score:2)
GPUs already have "many cores", if you can really call them that; 16 ROPs, 80 texture units, 64 shader cores, etc. Intel's approach is a much simpler architecture (in fact, "too simple" right now, the cores are practically feature-less), but DAAMIT's makes more business-sense (re-use what we've already got vs. invent something new).
Re:AMD's response (Score:2)
The problem with Fusion is if they kill the add-on graphics and you just buy one Fusion processor that costs say $400 to plug into the CPU slot. In the meantime, NVIDIA releases their new generation board and Intel releases a new generation CPU. The consumer can choose to upgrade one or the other or both, but an AMD customer is stuck just one expensive part and would have to upgrade it as one piece.
Perhaps in the future the CPU, the graphics card and the memory will all be on one giant module. You get high performance but not the ability to customize individual components.
Re:AMD's response (Score:3, Funny)
81 cores? Fuck that! We're going to 100 cores and putting a goddamn window on the CPU so all of the fan boys can watch the electrons flow. Then we're going to put the ethernet connection DIRECTLY on the chip. Ya, you heard me right, a connector right on the damn chip. You disagree? Great, I need your soon to be empty cube to store my prototypes you pussy.
Wait! Brace yourself, I've got another one. A speaker slapped onto the CPU. You hear that? That is the sound of genius and of every command to the CPU being broadcast. The bling-bling crowd will eat that shit up like a cupcake at a fat camp. Let Intel suck on THAT for a while.
Re:AMD's response (Score:2)
You're ignoring the server market, where the applications tend to be highly multi-threaded, and it's not difficult at all to keep your CPUs busy.
Ob (Score:2, Funny)
Re: Ob (Score:2)
ah, to hell with all these "obs"!
For the love of god... (Score:3, Insightful)
This isn't a general purpose processor. Think "cell processor" on a larger scale. You wouldn't be running your firefox or text editor on this thing. You'd load it up and have it do things like graphics processing, ray tracing, DSP work, chemical analysis, etc...
So stop saying "we already don't have multi-core software now!!!" because this isn't meant for most software anyways.
Tom
Re:For the love of god... (Score:4, Insightful)
So no, this model you're not going to be running Firefox or your text editor on (in fact, I doubt you even _could_ do this, these cores currently are very, very stripped down in their capacity to do work, to where they're basically two MACs tied to a small SRAM and a "network adapter"), but never-say-never, this style of chip is right around the corner.
Not usefull yet.. (Score:4, Interesting)
80 cores is an absurd number, with the parallelism level that we have in today programs, most of the cores should be idle most of the time.
Re:Not usefull yet.. (Score:2)
To the point: right now, we typically schedule applications to run on time-slices, to virtually expand one processor to every single process we have running. With multi-core, we need to change the way we schedule to be more granular, and to assign better core affinity (to the point where we can address specific cores directly from the operating system, always running the same application on the same core). Every task gets its own core, and can then use threading (or spawning another process) to request/force more core-time if necessary.
Ordinary desktops have been parallel for a long, long time, we've just hidden it from the users and from the programmers because it's hideously complex when it comes to timing and scheduling. The whole idea of the Operating System was to hide this complexity from the users to begin with, and to put it instead on the shoulders of smarter, better software that has been combed over and refined. To the point: our OSes have emulated parallel machines because we didn't have parallel machines, real-time (or at least near-real-time) multitasking would be impossible without it. Now, we have parallel machines, and we can stop emulating it or at least minimize our need to.
Re:Not usefull yet.. (Score:2)
"He never really said that" posts in 5,4,3...
Re:Not usefull yet.. (Score:2)
We can throw away the whole idea of NOPs when it comes to this chip and the technologies we have today. If we're not using the core, simply turn the core off, it's just going to be wasting energy, and you can turn the core back on pretty damn quick (~100's of ns) if you need it. Since these cores are virtually cache-less already, there's no huge to-memory penalty we get with today's enormous cores when they shut down to lower energy states, and it makes sense to have an on-chip load balancer which will make sure no one core gets over taxed and is on more than any other core.
Lastly, we don't need to change the apps as much as people think we do. For the applications that make sense to parallelize, great, do it. For the ones that don't, don't. It's not going to hurt anything if you don't. You still get N.mGHz speeds on your mono-core app, and to be truthful, it's highly likely that's more than your app will ever need, especially ones that spend most of their time I/O bound (word processors (keyboard), web browsers(network)).
Re:Not usefull yet.. (Score:2)
Likewise, I bet there are models that show more than 14 cores hurt performance in a super-plurality of applications. Could've you been more vague? What models, what types of applications, and most importantly why? You can use Amdahl's law to show the theoretical maximum speedup for different types of programs, and guess what, none of them go down after 16, you just flatten (essentially diminishing returns). It's true you do have communication and combination issues which can outweigh computation issues, so if you program like an idiot you can certainly hurt performance... but then just don't code like an idiot, use extra cores when you need the parallel computation, not just cause they're there.
What "most applications" are you referring to? Do multiple cores hurt word processing, or matrix multiplication, maybe prime factoring or photoshop?
Re:Not usefull yet.. (Score:2)
What twit is programming NOPs into modern programs? An idle core doesn't consume much power.
I would like to see a source for your 16 core claim.
Isn't this just (Score:2)
Hmm has it really taken 20 years of research or
Hey I've got a program it can run (Score:2)
perl -e 'fork while 1;'
There ya go. Think of it as a benchmark. How long can the 80-core processor run that without dying?
Re:Hey I've got a program it can run (Score:2)
Re:Hey I've got a program it can run (Score:2)
Wafer scale (Score:2)
We Did It in 1990 (Score:2)
Over a decade and a half ago, in 1990, I programmed parallel AT&T DSP32C boards (multiple DSPs per ISA board in an 80386 host). Up to 5 25GFLOPS chips on a 125GFLOPS board, up to 8 boards in a 1TFLOPS host PC. That PC, nearly double the "decade ago", had over 1TFLOPS (including its FPGA glue) in about 3 square feet.
And it actually ran applications (commercial image processing) in the field. This Intel chip might be smaller than 3'^2, but it still needs to run in the same size PC, and it doesn't run any commercial apps. Even 17 years later, on twice the cores we had in 1990.
Sure, the past 17 years hasn't seen our parallelism innovations become common (though finally digital cameras are catching up to our 16Mpxl, but not at 40bit color). Because the same problem we didn't solve in general, parallel programming semantics and debugging that reuse existing codebase and techniques, is still hard. But if we'd discarded the uniprocessor codebase then, or just ignored it as we built a new, parallel codebase, we'd have 17 years of code now that would be "legacy" which didn't largely lock us, and our thinking, out of simple multiprocessing development.
So I hope that Intel's multicore development is really just a platform for developing parallel coding systems. I'd love to see all that Intel money and brains put behind "executable UML", or some other flow "language", by inventing "lossless" lexical/graphical interconverters and expression standards. But they'll probably spend it all on marketing and emulating an 80386 to run Windows. Because that's the part of our 1990 platform that can be reinvented and resold as "brand new", just like it was back then, without taking much of a risk.
80 cores but is it still on the FSB? (Score:2)
Bandwidth (Score:2)
And I'm not impressed by the Flop rate. Not with the Cell Processor already out for a year.
I'm getting tired of the ancient computer comparis (Score:4, Funny)
How Well does it Scale? (Score:2)
how fast does it make toast? (Score:2)
Just I suspected... (Score:2)
Re:Just I suspected... (Score:2)
The issue is the fab and the die. How many bad die's did they go through to get to this prototype? I would wager they had many more bad dies then good ones.
640k cores (Score:2)
Seriously, why don't we put the cores into the memory DIMMs? It's been tried before but now it seems that a CPU core is just a little commodity thing, and memory bandwidth is where the bottleneck is.
How many AMPs? (Score:2)
Yikes, that's over 65 amps! Well, OK, it's less than 1 amp per core.
Here's the sad part... (Score:2)
Guess who's laughing now?
Re:Comparison? (Score:2)
Re:No x86 (Score:2)
And that is somehow supposed to be a bad thing?
Re:look closer (Score:2)
Re:look closer (Score:2)
Re:look closer (Score:2)