Slashdot Mirror
Intel Pledges 80 Core Processor in 5 Years
ZonkerWilliam writes "Intel has developed an 80 core processor with claims 'that can perform a trillion floating point operations per second.'" From the article: "CEO Paul Otellini held up a silicon wafer with the prototype chips before several thousand attendees at the Intel Developer Forum here on Tuesday. The chips are capable of exchanging data at a terabyte a second, Otellini said during a keynote speech. The company hopes to have these chips ready for commercial production within a five-year window."
...Imagine a Beowolf cluster of those!
(Runs in shame.)
'Sensible' is a curse word.
But it still can't tell me 1/0.
This will finally run Vista, right??? Maybe? Hopefully?
Unto the upright there arises light in the darkness...
Unfortunately, they'll all choke on a shared memory bus :-)
promised us 8-10Ghz Pentium4 CPUs when they started with the P4 "Willamette"? Or how they promised us 5GHz Prescotts?
I'll rather wait and see what I can actually buy in 5 years. No need to trust a vendor so far in the future what they can do.
Why oh why won't Intel spend their research dollars on something useful, like a bus architecture that can actually keep up with present performance levels?
Faster processors are great, but when will we see massive improvements in data storage...
Imagine the pain of having to write a functional applications with so many cores. I hope the interconnect will be very very fast. Otherwise writing massively scalable parallel algorithms will be masssively painful. And with so many cores, one will need multiple independants memory banks with some kind of NUMA. And writing apps for those things isn't fun. You have to spend so much time caring about the parallel stuff instead of caring about the problem.
That's hot!
(/ducks)
I seriously hope that power consumption and heat disipation are really attacked before these things come out. Can you imagine needing a 200-amp service and liquid nitrogen cooling for something like that right now?
This is hilarious, because if this goes out on the market there's not going to be many operating systems capable of scheduling on that many chips usefully. OS X can't do it, Windows can't do it, and nor can BSD. But Linux has been scheduling on systems with up to 1,024 processors already :)
With the heavily threaded nature of BeOS, even demanding apps would really fly on the quad+ core cpus that are preparing to take over the world.
Not that you couldn't do threading right in Windows, OS X, or Linux. But BeOS made it practically mandatory: each window was a new thread, as well as an application-level thread. Plus any others you wanted to create. So to make a crappy application that locks up when it is trying to do something (like update the state of 500+ nodes in a list; ARD3 I'm looking at you) actually took skill and dedication. The default state tended to be applications that wouldn't lockup while they worked, which is really nice.
Slashdot Patriotism: We Support our Dupes!
In other news, Gillette pledges a razor with 81 micro blades. 80 blades are individually controlled via Intel's new 80 core processor. The 81st blade is available just because..
Just in time for Vista!
Just as Gates couldn't imagine what anyone would want with more memory than 640KB, we can't imagine what people will do with 80 cores. I'm confident in predicting that they'll find ways to use every bit of that capacity and demand more.
Wow, good point. I bet Intel never once stopped to think about THAT.
I sincerely doubt this will make it anywhere near Fry's or CompUSA, assuming it launches in +5 years. Most likely academic, corporate (think of the old days and mainframe number crunchers on wallstreet), and scientific.
Simply cheap teraflops for custom applications.
Of course, everyone thought it was a great idea when Cell announced they could do 64 or more cores. But since this is
https://www.accountkiller.com/removal-requested
The costs to make use of 80 cores (you're going to need hugely complex chips and hugely complex memory buses) mean that these chips will be severe overkill for PCs and will be outside any typical user's price range. They're only going to be useful for a a few servers in very niche applications. If there's only demand for, say, 10,000 of these chips in the world then they're going to be extremely expensive.
I smell marketing horseshit. I think they're just saying this to get people to start thinking of multi-core options. Most people don't see the need for multi-core (even 2 core) systems. By saying you'll get 80 cores in 5 years makes people start thinking that they should start using 2 or 4 cores now.
Engineering is the art of compromise.
Today, a 2 CPU x 2core computer can actually be slower than a 2x1 or 1x2 core for certain "cherry picked to be hard" operations due to the OS making incorrect assumptions about things like shared/unshared cache - 2 cores on the same chip may share cache, two on different chips may not - and other issues related to the fact that not all cores are equal as seen by the other cores.
In an 80-core environment, there will likely be inequalities due to chip-real-estate issues and other considerations. The question is, will these impacts be felt at the code level, or will the chips be designed to make these differences invisible? If the former, will the OSes be designed to use the cores efficiently, or will they simply see "80 cores" and, out of ignorance, make poor decisions when allocating tasks to various cores? If the latter, what performance penalty will be incurred?
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
I'm not sure this is 80 general-purpose processing cores: the article claims that there are "80 floating point cores". Clearly, the big selling points of the chip are, in Intel's view, its data transfer at 1 TB/sec, and its floating point speed at 1 TFLOP.
An 80-core chip with RAM attached directly to the processor chip, as TFA discusses, is going to have an advantage in transferring data between cores, and plus it'll probably be a lot smaller. Than 40 dual core (or 20 quad core) chips.
Why oh why won't Intel spend their research dollars on something useful, like a bus architecture that can actually keep up with present performance levels?
Yes, because if Intel is working on one thing, that means they can't work on anything else at all anymore...
I believe posters are recognized by their sig. So I made one.
Otellini meant both flops and memory xfer rate.
Clarifiation from TFA:
"But the ultimate goal, as envisioned by Intel's terascale research prototype, is to enable a trillion floating-point operations per second--a teraflop--on a single chip."
Further clarification from TFA:
"Connecting chips directly to each other through tiny wires is called Through Silicon Vias, which Intel discussed in 2005. TSV will give the chip an aggregate memory bandwidth of 1 terabyte per second."
Your wish has been granted.
Next!
Education is a better safeguard of liberty than a standing army.
Edward Everett (1794 - 1865)
Moore's law states that transistor density doubles every 24 months, it says nothing about speed or number of cores.
I believe posters are recognized by their sig. So I made one.
Moore's law says nothing about speed, that is a common error. IIRC he made a general statement about the number of transistors that could be in a defined area doubling every 2 years and that was later changed to 18 months. It also had to do with cost of transistors I believe.
640 cores ought to be enough for everybody
Slashdot, fix your code or at least hire someone who is competent at it to do it for you.
Scheduling isn't a one size fits all process. What works at 4 cores doesn't work at 40 and so on. As for other operating systems, FreeBSD has been working quite actively on getting Niagras working well with their sparc64 port. I've been saying it didn't make sense until this announcement. I figured we'd have no more than 8 cores in 5 years. We'll see what really happens.
The BSD projects, Apple and Microsoft have five years. Microsoft announced awhile back they want to work on supercomputing versions of windows. Perhaps they will have something by then. Apple and Intel are bed partners now. I'm sure intel will help them.
What this announcement really means is that computer programmers must learn how to break up problems more effectively to take advantage of threading. Computer science programs need to start teaching this shit. A quick you can do it, go get a master's degree to learn more isn't going to cut it anymore. There's no going back now.
MidnightBSD: The BSD for Everyone
How many processes are running on your machine?
A basic strategy would be for the OS to devote each process to its own processor.
This would reduce the need for TLB/cache flushes or eliminate context switches entirely. The whole machine would be really snappy.
That said, for a desktop machine, this is a huge amount of overkill, but with economies of scale being what they are, we'll probably have this power available soon.
What I'd like to see more though, is extra functionality in hardware rather than more of it. Wouldn't it be great if hardware was able to handle some of the things an OS is now used for, like memory (de)allocation? Or if we could tag memory according to type? Or if there were finer-grained controls than page-level?
If moderation could change anything, it would be illegal.
Wicked... hmmm.... a castle full of nubile virgins all asking me to spank them?
like a bus architecture that can actually keep up with present performance levels?
There's nothing wrong with bus architecture in my opinion.
I stand at the stop, the digital sign says the bus will be along in 4 minutes.
4 minutes later the bus turns up.
I don't see what the problem is.
Summation 2
not 80 general purpose integer cores. They're essentially copying the Cell design with large numbers of DSPs each of which has a local store RAM burned onto the main chip. Is this a good idea? Guess we'll find out with the Cell. What interests me most about this announcement is not the computing potential from such a strategy, but that it's an obvious response to IBM and Sony technology.
You fools! Do you have any clue how much Oracle licenses will cost for this thing?
A couple of things to mention here. Many years ago I read an Intel road map for the x86 processors. It was more than 10 years ago, less than 20 I think. In it they said they would have massively multicore processors coming along around now. They may have forgotten that and reinvented the goal along the way, companies do that. But, they really have been predicting this for a very long time.
:-) The counter example everyone came up with was, "well, if that is true why would AMD buy ATI?" The answer to that is simple, they want their patent portfolio and their name. In the short term it even makes sense to put a GPU and some shaders on a chip along with a few cores. At the point you can put 16 or so cores on a chip you won't have much use for a GPU.
The other thing is that with that many cores and all the SIMD and graphics instructions that are built into current processors it looks to me like the obvious reason to have 80 cores is to get rid of graphics coprocessors. You do no need a GPU and a bunch of shaders if you can throw 60 processors at the job. You do need a really good bus, but hey, not much of a problem compared to getting 80 cores working on one chip.
With that kind of computer power you can throw a core at any thing you currently use a special chip for. You can get rid of sound cards, network cards, graphics cards... all you need is lots of cores, lots of RAM, a fast interconnect, and some interface logic. Everything else is just a waste of silicon.
History has shown that general purpose processing always wins in the end.
I was talking to some folks about this just last Saturday. They didn't beleive me. I don't expect y'all to believe me either.
Stonewolf
This is the last 3 years of Intel, all over again. Only now the megahertz race is replaced with the multi-core race.
Intel will create the "CoreScale" technology and make 4, then 8, then 16 cores and up while their competitors are increasing operations per clock cycle per watt per core. Consumers won't know any better, so they will buy the Intel 64-core processor that runs hotter and slower than the cheaper clone chip that has only 8 cores. Then when Intel starts runs up against a wall and gets their butt-kicked they will revert to the original Core 2 Duo design and start competing again.
Oh, and I predict that AMD will release a new rating called the "core plus rating" so their CPUs will be an Athlon Core 50+ meaning it has the equivalent of 50 cores. Queue n00bs who realize they have only 8 cores and complain.
And to think I didn't like history in school. Maybe I just hadn't seen enough of it to understand.
Software hasn't really improved for maaany years now, Spreadsheets and Word Processors are more colourful, higher resolution. But are these products smarter, better at all? Would a postgraduate write a better doctoral thesis with Office 2007 than with - say - Word 6.0? Is image manipulation thaat much better with the latest photoshop than with PS 5.5? With some minor exceptions the answer is clearly no.
- We were promised Virtual Reality with VR Helmets more than 10 years ago - is this _just_ a matter of hardware?
- Smart voice recognition? Anyone tried it lately? Anyone tried to write pretty standard letters with it? Desastrous.
- Intelligent assistents, understanding the user's needs? Operating system/application wizards that improve it's capabilties while you're working with 'em?
The applications are missing, they're faster, more colourful, higher resolution, antialiased... but still DUMB.
Computers are already pretty powerful, please start and make the software smarter, not faster.
CPU power is not that important anymore.
WASHINGTON (Reuters) In lights of Intel's 80 Core Processor pledge in 5 Years, scientists are worried that Richard Branson's pledge is now too little too late.
[alk]
The big question is how these processors interconnect. Cached shared memory probably won't scale up that high. An SGI study years ago indicated that 20 CPUs was roughly the upper limit before the cache synchronization load became the bottleneck. That number changes somewhat with the hardware technology, but a workable 80-way shared-memory machine seems unlikely.
There are many alternatives to shared memory, and most of them, historically, are duds. The usual idea is to provide some kind of memory copy function between processors. The IBM Cell is the latest incarnation of this idea, but it has a long and disappointing history, going back to the nCube, the BBN Butterfly, and even the ILLIAC IV from the 1960s. Most of these, including the Cell, suffered from not having enough memory per processor.
Historically, shared-memory multiprocessors work, and loosely coupled network based clusters work. But nothing in between has ever been notably successful.
One big problem has typically been that the protection hardware in non-shared-memory multiprocessors hasn't been well worked out. The Infiniband people are starting to think about this. They have a system for setting up one way queues between machines in such a way that appliations can queue data for another machine without going through the OS, yet while retaining memory protection. That's a good idea. It's not well integrated into the CPU architecture, because it's an add-on as an I/O device. But it's a start.
You need two basic facilities in a non-shared memory multiprocessor - the ability to make a synchronous call (like a subroutine call) to another processor, and the ability to queue bulk data in a one-way fashion. (Yes, you can build one from the other, but there's a major performance hit if you do. You need good support for both.) These are the same facilities one has for interprocess communication in operating systems that support it well. (QNX probably leads in this; Minix 3 could get there. If you have to implement this, look at how QNX does it, and learn why it was so slow in Mach.)
No they don't. Right now I'm building a Linux kernel and it is only using approx 35% of the CPU. Why? Because my memory and disk are not fast enough. If I swapped out the CPU and kept everything else the same, it would not go much faster. Sure, with a faster motherboard etc I could get better speed, but that is very difficult to scale to 80 cores
As I said before.... to get 80 cores working properly is going to require huge amounts of memory as well as hugely wide buses out of the chips (say 512 bit-wide buses), huge increases in disk rw speecd etc.
Nobody is going to design 80 core systems unless someone is prepared to buy them and nobody is going to design 80-core chips if nobody can show how to design effective systems with them.
For people wanting to crank SETI etc, it is going to be way cheaper to build a cluster with 20 4-core systems.
Engineering is the art of compromise.
So think more like Cell with 80 SPEs. Great for lots of vector processing.
No folly is more costly than the folly of intolerant idealism. - Winston Churchill
In the 80s and early 90s, most of the bus speed limitations were due to capacitance issues (ie. how fast can we switch a transistor and discharge the capcaitance). We can make things faster by reducing capacitance through various measures. Now memory buses and speed are now getting so fast that they're starting to get constrained by the speed of light etc so it is getting harder to find large multiplier improvements.
I think there is still a lot of room for new stuff, maybe twice or four times what we have now. The biggest impovements that can be made, however, are in power reduction etc.
Engineering is the art of compromise.
If I was writing a game and knew that the majority of gamers had at least 80-core CPUs I would:
- Dedicate 45 cores to the opponent AI (which would run on simple neural nets)
- Dedicate 20 cores to physics (because physics is the next-big-thing)
- Dedicate 8 cores to keeping the former fed with usable data (like game logic, asset management, etc)
- Dedicate 4 cores to 3d sound (because with so many cores it's cheaper for me to develop the sound myself than license the latest EAX from Creative, or whatever's hip at the moment.)
- Dedicate 1 to networking and voice-chat (because the better the compression, the better the experience)
- Dedicate 1 to coordinating the rest.
- Leave 1 for the OS and any parallel tasks.
Oh and not having to make my code terribly efficient would cut my development costs a lot.
So that's that for using 80 cores. Sure could use more in the AI department.
And the advantages of an 80-core chip over 40 2 core chips? A hell of a lot of physical space.
May the source be with you.
Er, I wasn't pointing to any particular multiplier. I was pointing out that, even if you are right that, when released, these would be prohibitively expensive for most purchasers, that history suggests that processors go from "prohibitively expensive for most user", to "common", to "you really need to upgrade that old piece of crap" pretty quickly.
Is image manipulation thaat much better with the latest photoshop than with PS 5.5? With some minor exceptions the answer is clearly no.
Hah! I am forced to disagree in the strongest possible terms..
Speaking as a former production artist and current art director, the last couple of generations of graphics software have introduced powerful tools that streamline my workflow in ways I find it hard to even fathom. Ok, let's talk about Illustrator, for example. From 10 -> CS Adobe added in-application 3D rendering of any 2D artwork onto geometric primitives. This is something I used to either have to fake, or take out of the application and into a 3D renderer in order to render simple bottle/can/box packaging proofs. Marketing wants to make a copy change? Make the change to the 2D art and the 3D rendering is updated in real time. Oh, and the new version of InDesign recognizes that the art has been updated and reloads it into the brochure layout. Automatically.
This is just one feature out of literally hundreds. This one alone saves me an hour or two a day. Seriously, there are projects I can take on today that would have been unthinkable 5 years ago. Pre-press for a 700 page illustrated book project has gone from a week of painful, tedious work down to 30 minutes, of which 20 is letting the PDF render. Seriously.
Here's the thing, unless you use a piece of software all day, every day, you're really not in any position to comment on how much it has or hasn't changed.
Photoshop (et. al.) are software for professionals, despite the number of dilettantes out there using them for sprucing up their MySpace page.
m-
You catch enchiladas by picking them up behind the head and holding them underwater until they don't kick anymore -VeGas
When will Intel fix their floating point issue?
Is 1994 soon enough for you?
-jcr
The only title of honor that a tyrant can grant is "Enemy of the State."
In other words, get out your functional languages like Haskell and OCaml and use the side-effect free feature set to develop multi-threaded programs. Or do it the hard way with an OO language.
Fascism trolls keeping me up every night. When I starts a preachin', he HITS ME WITH HIS REICH!
ust as Gates couldn't imagine what anyone would want with more memory than 640KB, we can't imagine what people will do with 80 cores. I'm confident in predicting that they'll find ways to use every bit of that capacity and demand more.
He wasn't asking what 80 cores are good for. He's asking why we need 80 cores on one chip. As opposed to 40 dual-core processors, for example. And it's a good question. I imagine that these 80 cores can communicate extremely fast between their nearest neighbors. This could be amenable to implementing mesh-based parallel algorithms. Maybe toss in a couple hundred K of local RAM for each core, or shared blocks of RAM for every 4 cores, or whatever. Or maybe the core interconnections are arranged in a hypercube... Or imagine a super-hypercube configuration where each node of the hypercube is an 80-core mesh, to develop hybrid mesh/cube algorithms. Cool stuff.
Power requirements for a single, 80 core CPU are probably going to be much less than 40 dual-cores, as well.
Title and summary have it (slightly) wrong.
Intel's prototype uses 80 floating-point cores.
Very interesting in itself, but not the same as 80 CPU cores, which is hinted at by summary.
Carbon based humanoid in training.
Smart voice recognition? Anyone tried it lately? Anyone tried to write pretty standard letters with it? Desastrous.
lol
They are claiming a terabyte per second interconnect. I think it is safe to assume it will be something like an Infiniband, myrinet or similar (NEC's IXS, IBM's HPS) high performance application networking technology.
What you're asking for is pretty standard stuff in the high end, where hundreds of processors is quite common. Cache coherency is a killer, and so they have died out long ago in the high end. when you think about it, CC basically requires a crossbar switch style memory archictecture which expands with the square of the number of processors, and much higher speed logic to resolve conflicts. So eventually, it doesn't scale. Instead multiple applications with large numbers of processors tend to only have small groupings (say 8 or so, but can go upto 30 odd) using shared memory/cc access) and then MPI for anything bigger.
Clusters have been using MPI for years for this sort of
thing. all the custom interconnects for supercomputing
have customized implementations in their MPI libraries to
take advantage of 1-sided communications. Most use a facility which can loosely be termed RDMA - remote direct memory access, another word sometimes used is OS-bypass. The idea is that for this sort of communications, you want to skip the TCP/IP stack and other OS buffering overhead, and just have straight memory to memory copies going on (under userland library control.)
folks generally don't do the direct invoking of things on other processors, but instead fire off jobs on blocks of processors, and have them communicate with 1-sided primitives. This is the sort of thing done on hundreds or thousands of processors today. It will just gradually percolate down to normal applications.
The condensated core will also be subject to the laws of quantum mechanics in that, before a program has finished running, there will be no way to know if it will crash or not. Microsoft plans to leverage this to further stablise their latest version of Windows. Security experts worried about the onboard "Quantum-Threading" technology redirecting portions of thread output randomly to other threads, were dismissed as not being "forward looking".
Meanwhile, AMDs new 1W, 128 core, 4098bit chip with 1GB L2 cache retails for almost 50% higher than Intel's Bose-Wintel chips, and has seen sluggish sales since the arrival of the new technology, despite its lower running cost that the 5MW Intel chip. When asked for comment, AMD's spokesman added; "Ch@#&t!! What the f**k is wrong with you people!??! Our chips save you money!! F@#*&^g cheapskates!!!"
Upon hearing the news, Linux founder and lead developer Linus Torvalds(51) said: "We're not rewriting the kernel for that monstrosity." Intel representative declared that the company was "dissapointed" in Torvald's remarks. Apple cofounder Steve Jobs(65), when asked whether Apple intended to release a the new Mac based on the chipset, declined to comment as he went about his daily 5km morning run. Apple pundits widely believe that the new Mac will run on a quad core Bose-Wintel Condensate, and to complement this will sport a blazing white, ultra smooth case made out of Bose-Einstien condensate, the fifth phase of matter.
In a related story, Microsoft cofounder Bill Gates(65), assaulted a technology reporter at a company press conference disccusing the new chip. Details are sketchy, but reports mention that one of Mr Gates older quotes about appropriate amounts of computer memory was brought up. Remond police have declined to comment on the case.
May the Maths Be with you!
um... I guess it ain't clear then... the parent post was saying that you need OS support for accelerated remote procedure calls and one-sided communications. However, 1 sided communications already is in standard use by folks using hundreds of processors through an already
standardized library: MPI - Message Passing interface. Rather than the OS needing to define a new API, the folks creating high speed interconnects just create optimized libraries (in order to sell their hardware). Folks writing codes for hundreds of processors tend to want to treat them as array elements, so the chaotic calling of procedures just is not that useful. so the RPC support he is asking for is not really important.
In other words, the software stack for using large numbers of processors is already well-known. No need for any new OS features.