Slashdot Mirror
IBM Ships Fastest CPU on Earth
HockeyPuck writes "The 5-billion-instructions-per second Power6 processor from IBM would beat such rivals as the 3.73 gigahertz Pentium Extreme and the 2.4 gigahertz UltraSparc T2 from Sun. 'It's hard to make the average person understand just how fast this is,' said IBM Chief Technology Officer Bernard Meyerson, offering an example meant to explain his company's baby that still leaves the listener awed with the speediness of the two laggards. 'Hold your index finger out in front of your face,' Meyerson said in a telephone interview from IBM headquarters in New York. 'In less time than it would take a beam of light to travel from your knuckle to your fingertip, the new IBM chip would complete one task and start looking for the next, he said.'"
My blog
I'm glad they stopped measuring chip speed in Hertz and are now using the simpler metric fingertip-to-knuckle units.
steampunk web design
and all the water turned into steam!
If you keep throwing chairs, one day you'll break windows....
init 11 - for when you need that edge.
Nice chip. Now what OS and applications run on it?
Too bad Apple no longer uses IBM processors, this would've been a great marketing scheme for Steve Jobs.
'It's hard to make the average person understand just how fast this is,' said IBM Chief Technology Officer Bernard Meyerson, offering an example meant to explain his company's baby that still leaves the listener awed with the speediness of the two laggards.
Made me think of a National Lampoon Radio Hour (SNL before it was on TV) skit about the George Foreman-Muhammed Ali fight. Foreman (John Belushi IIRC) talking about Ali:
"He so fast he can turn off the light and be in bed before the room get dark!"
mcgrew's razor: Never attribute to stupidity that which can be explained by greedy self-interest
ob disc: I work at sun (but not ON those chips).
I write management software that lets admins turn on/off/standby (etc) the state of the various 'cpus' (threads, as sun calls them). there are 128 and 256 cpus in a regular 2u..4u style rackmount box. these are 'simple' air cooled systems with fans blowing over the whole U-style chassis and over the passive cpu heatsinks. nothing 'scary' at all, really.
it is pretty wild to be able to do the equiv of 'show cpu' and have an ascii output scroll 64, 128 and even 256 times; one for each 'cool thread' which is a real actual processor element.
the down side is that this threading stuff does not automatically get you faster speed on a SINGLE non-threaded traditional task. as I understand it, these T-series sun boxes are meant to process a lot of transactions (think webservers) and not so much number crunching.
how do you define 'fastest chip'? well, one thing is for sure, you do NOT simply go by 'gigahertz' alone. that's really an oversimplification.
--
"It is now safe to switch off your computer."
... it will be in a washing machine controller.
Does it have a Altivec unit ?
If it doesn't - it's just a useless overclocked G3 - Mac OS X Leopard will not run on it.
If it does have Altivec - anyone knows if it uses the same socket as the G5 ?
1% APY, No fees, Online Bank https://captl1.co/2uIErYq Don't let your $$$ sit in a no-interest acct.
The fact that they compare it to a last-gen CPU makes me suspicious. An e4300 probably runs at about a par with the Pentium Extreme, and the newest Core 2 Duo chips would surely wipe the floor with even the fastest Netburst chips. Show this chip running on a par with a QX9770 or even an e6850 and I'll be impressed.
...ecause we're still held back by the lethargic TSA officers and a single-file security line.
www.purevolume.com/martyd
Correct me if I'm wrong, but don't most of the algorithms for scientific simulations run just as well on parallel processors? If this is the case, it makes more sense to have cheaper processors (both to manufacture and run) so that the cap on the IPS is raised just as well (the cap being the ratio of the amount of funds an organization can allocate to these emulators versus the cost of the emulators). Though I'm no computer expert, it seems that making one sequential processor run faster isn't as efficient as making that same processor cost a fraction of the price to get more power from your dollar.
Help fight spam
Does it run Crysis?
If you haven't made a developer cry, you've wasted a day.
My knuckle to finger-tip is a big-ass distance. As big as my whole CPU!
:)
Come back when you can complete one 64-bit Floating Point Operation in the time it takes light to travel the width of one atom of silicon. Yeah, you'll need some quantum mechanics, so get on it. And no, I'm not telling you what the op is.
Does this mean somebody build a faster processor which is now in space ?
:)
Hmmmmmm.
(Bye,
Skybuck)
It mentions in the article that this processor would be the fastest "if a stopwatch were the only ranking system."
But what is a cycle from a Power 6 processor worth?
Does it get just one instruction per cycle?
What about the instruction set?
How efficient is the powerpc instruction set at running through logic?
How efficiently does this processor handle those instructions?
The article doesn't answer any of the really interesting questions about the new processor. It is neat though that IBM gets them to go at 5 ghz with water cooling, I haven't heard of any intel processors going that high without using liquid nitrogen.
But does it run linux?
can I render 3D images of naked women? When are they going to start using a relevant metric to measure speed? Geez.
"Know but never fear the consequences of your actions."
For multi-core do you have to hold as many fingers up as there are cores to understand the speed over multiple cores?
I has Altivec, but IBM calls it VMX
And no, it is not socket compatible to my understanding with the G5 (PowerPC 970).
There are indeed many algorithms that run well in a parallelized environment. IBM even makes the world's fastest supercomputers that take advantage of this fact.
However, there are many other tasks fit for computers that do not parallelize well. In addition, writing massively parallelized software is often quite HARD. It is far easier to design software for a single CPU running very quickly, than a whole boatload of CPU's running slower. There have in fact been quite a few articles in CS journals lately wondering how on earth software is going to be written for all these new bunch-o-cores CPUs. While it can be done, it is tedious, expensive, and error-prone for all but the most trivial tasks.
SirWired
You see Arthur, no wonder the thumb is used for space travel.
It DOES run Linux!
...>ahem and other OS's too... like Windows NT...
(mod me down if you must - but I just HAD to...)
Seven Days with Ubuntu Unity
...imagined a beowulf cluster of these?
"I think it would be a good idea" Gandhi, on Western Civilisation
That would only be true if the CPU is able to retire a sustained average of one instruction per clock cycle. SFGate's article makes a raw comparison between chips with different number of cores, threads and other factors, considering only GHz...
This is one of the main reasons I decided to stick with a threaded model for concurrency in FreeSWITCH. There are many alternatives cropping up to obtain concurrency in your applications but if you stick to a threading model you will gain the benefits of more cores on faster CPUs and better schedulers in the kernels. The number of concurrent instructions are going up much faster than the actual speed of a single CPU these days.
5 billion THEORICAL instructions per second just mean nothing.
Anyway, the DSP I'm working on, the TI C6416 (1GHz), claims up to 8 billion instructions/s (5 to 6 can be realistically obtained).
1 (short ton / firkin) = 89.1432354 slugs / keg
I see this all the time when I upgrade. I often wait so long that several hardware generations have passed and I think the upgrade will blow me away but it just doesn't happen. Yeah, it's faster but I have never been blown away by how much faster it is because it's not that much more.
your soul.
Muahahahah
-D
How many Libraries of Congress can it index per fortnight?
Knowledge is how to play a game, intelligence is how to win, wisdom is knowing what game to play.
or just Vista ready?????
Donde Ser Geek No Duele
Lyle: I'm gonna get a NAD T-770 digital decoder with a seventy-watt amp and and Burr Brown D.A.C.'s.
Hansome Rob: [at a loss] Yeah...
Lyle: It's a big stereo. Speakers so loud, they blow women's clothes off.
Handsome Rob: Now you're talking!
Set your phasers on "funky"!
“If the earth is this basket ball and the moon, a golf ball, the space craft would have to enter through a window no thicker than this piece of paper.”
Power6 is a big change from Power4 & Power5 series. The key factor is: it gains clock and SMT at the expense of OoOE. In-order execution means its performance is deeply dependent on perfectly tuned compilers.
Other than the lack of out-of-order, on paper it looks pretty strong. Dual core, lots of bandwidth, up to 7 IPC (5 in one thread, 2 in the other), big GHz, voltage & frequency slewing, and yes it has AltiVec.
p.s. No, it would not be good for Macs. POWER chips are all made for big iron.
...because, for a start, a 3.73GHz Pentium Extreme is as powerful as a 1.87GHz Core 2 Solo. Not that it's not good news that these guys are pushing the speed envelope without burning the chip, but the underlying microarchitecture can make a huge difference.
You have no idea how difficult multicore Operating Systems are.
Your post clearly demonstrates:
1. The new powerpc6 you were typing on is too fast for your feeble fingers to keep pace.
2. An IBM FDIV prime number calculation bug?
3. Your ideas intrigue me and I shall order your subscription.
4. A well hidden "poo on you" subliminal to all of slashdot.
5. ???
6. Profit!
A pico-second is how long it takes a slashdotter to think of something stupid?
From the article, "Then why don't Intel and Sun just crank up the speed? Well, just as is the case with cars, the faster chips run, the hotter they get, and IBM has created water-cooling systems akin to the radiators in cars to keep its processors from overheating. Not doing so, Meyerson quipped, "results in setting fire to the user, which is bad.""
This part of the article made me laugh. Sick sense of humor? But, I would formally like to thank IBM for caring about setting me on fire or not.
Bearded Dragon
Apple's development tools continue to generate binaries that support both Intel and PowerPC, and they continue to support the operating system on both architectures, so why not? Yes, I know, there's a difference between POWER and PowerPC, but it's not a big difference -- after all, you run the same version of Linux on an IBM pSeries that you do on a Power Mac.
Apple would have buying leverage against both IBM and Intel by being able to shift portions of their manufacturing from one architecture to another with each model. And they'd have access to some of the fastest processors on earth. Can you imagine one of these things powering Photoshop, or even rendering the next Pixar movie?
Tired of FB/Google censorship? Visit UNCENSORED!
The "speed" of a single CPU is a commercially useless measure of performance. We care about performance per dollar, not performance per CPU. And that dollar includes the cost of of energy required to operate the CPU associated cooling.
Ceci n'est pas une signature.
"The baseball player stepped out of the box and spit like a fountain statue of a Greek god that scratches itself a lot and spits brown, rusty tobacco water and refuses to sign autographs for all the little Greek kids unless they pay him lots of drachmas."
a Beowulf cluster of these... ;)
I always said rating CPU's in Hertz is like rating engines by cubic inches. Bigger *can* get more performance, but it's no guarantee. The compression, carbeuration, transmission, fuel flow, exhaust, all add up to final performance, same as cycles per instruction, the amount of work each instruction can do, the memory bandwidth and the IO system all add up to system performance in a computer.
Karma Whoring for Fun and Profit.
Imagine a beowulf cluster of these!
Imagine a Beowulf cluster of ... awww forget it. I'm getting too old for this.
and the software monkeys will toss in 12 different abstraction layers and indirect ways of doing things that Firefox will still take 30 seconds to show up... That anything these days is not instant is mind boggling to me. What is the CPU *doing* in all that time?
...but it's not my index finger....
It's hard to make the average person understand just how useless this is when you hit the memory wall.
However, it is fair to say that IBM still build machines whose memory subsystems scale waaay better that anything in PC-land.
He's never had a help desk job....
"We are all geniuses when we dream"
- E.M. Cioran
This processor should be used with the fastest internet connection available to surf pornographic web sites! Note all that smut would break a mortal man but this would be a mission to catalog ALL XXX rated web sites around the world. A feat never undertaken! A feat that could be done! A feat worthy of a world record!
But anyone caught running Windows on it would be declared a terrorist and an enemy of America...to be sent to Gitmo for torture and forced into Linux group think.
and I thought he was going to finish that with "and it goes THIS fast!!!", as he waves his finger across his face as fast as he can.
That's how my brother and I used to measure seconds when we were 5 years old. Accurate to within 500% (your mileage may vary).
Maybe I'm misunderstanding something, but isn't the speed of light supposed to be an absolute for things in the realm of "real world applications"? How can a single "task" (whatever that means) get processed faster than the speed of light? Even with multicore technology, I wouldn't think that a single task could get split up between the cores because that task is, well, singular.
*slight crashing sound*
...comparing a high end IBM and Sun server processors with a desktop Intel processor. Something like a Xeon X5482 (3.2GHz, 12MB cache 1600 FSB) would have been a much better comparison and, on some benchmarks , probably would beat the P6.
Yes, I spoke to him on the phone. I asked him when I'd be able to benefit from this cpu to build better performing software, on a stable and widely used platform, and more important than that, when would I earn enough money to buy a machine to build this environment I'm talking about.
Well, he said, imagine me holding my middle finger in front of your face...
I did not say anything else.
..the next version of this chip will be..
Has there been a leap forward in compiler technology to make the trade off of going from out of order to in order execution? (ignoring the more obvious gain of not having to implement out of order in silicon) I just think back to VLIW and EPIC machines as a shining example of how overhyped compiler technology has been in the past. Somehow I don't see how moving to an in-order processor somehow requires something new of a compiler that hasn't been around for a long time now.
someone in the know care to comment?
Hold your index finger out in front of your face, now poke yourself in the eye. That's how much it will cost you.
alias possession='chmod 666 satan && ls
It's very useful to remember that the speed of light is about a billion feet per second, or a foot in a billionth of a second. He was just looking for a measure that is 1/5 of a foot long.
But how many BogoMIPS is that? After all, that is most computers do, most of the time.
Think about it: This new chip can do nothing up to 2 to 3 times faster than any other chip on the market! We are talking about incredible productivity gains during idle times!
(why yes, I do work in marketing...)
Tequila: It's not just for breakfast anymore!
... was if the processor were an acid, and you were to put on a pair of goggles, the goggles would do nothing.
it's a good visualization. And thats all it is,it wasn't designed as a spec for engineers.
The Kruger Dunning explains most post on
Great. Does it play Crysis? :D
IBM beats Intel hands down on raw processor speed. Has the same gap in performance existed between high end desktop processors and high end workstation/server chips like these?
Thanks Steve ( jobs ), with your shortsightedness.
---- Booth was a patriot ----
... big numbers sound impressive, but they don't necessarily mean better performance.
I prefer rogues to imbeciles because they sometimes take a rest.
But wrt to the bad analogy... what is the medium between the eye and finger?
That'll be about 5.9GHz then...
They're overclocking the POWER6 chip, is that right? Sure, IBM, Cray, DEC et. al. used to do this routinely on big iron back in the day when computer technology was still a science. I still remember seeing a beautiful oil "waterfall" on the front of some mainframes. It wasn't called overclocking back then - it was just how things were done. Now, with computing being a commodity, most companies don't bother with this - too unreliable, too bulky, too power-hungry. Remember, the weakest part of any electronic device is the mechanical aspect and water cooling involves a lot of mechanical processes. You've really got to have a need for speed to bother with this (and, yes, some big environments have such a need - but not many).
For those few environments which need this much speed per processor, this is an important development. Just don't count on it ever impacting the average desktop (commodity) system - the technology won't "trickle down" (unlike the coolant?).
three knuckles. Which one is he talking about?
The software guys are already busy working on software upgrades that will counteract that processor speed so that the software is even slower than the current release.
Those who do study history are doomed to stand helplessly by while everyone else repeats it.
I would mod this up if I had the points.
DRM: Terminator crops for your mind!
I was under the impression that persistence-of-vision lasts a whole 60th of a second, and I can *definitely* see things go dark when I blink.
DRM: Terminator crops for your mind!
a Beowulf cluster of these
I'm still laughing at my previous assertion - IBM certainly understands the wisdom of "Field of Dreams" (*whispers* If you build it, they will come).
http://www.youtube.com/watch?v=2eSwf5LxGAM
http://forums.vr-zone.com/showthread.php?t=195427
Using the technology below it shouldn't be too long before we have devices in the 10's of GHz.
http://www.compoundsemi.com/documents/articles/news/8479.html
Toshiba Develops 60 GHz Receiver Made With CMOS Processes
June 18, 2007...Toshiba of Tokyo, Japan, reports that it has developed a new technology to manufacture integrated circuits for the millimeter-waveband. Toshiba says that its new fabrication uses low-cost CMOS processes to produce devices that can achieve high speed, wireless communication in the 60 GHz band. The company points out that at the 60 GHz frequency (which is ten times greater than wireless LAN), communication distances are limited to a few meters, but data can be transferred at a rate of more than a gigabit per second.
I am always doing that which I can not do, in order that I may learn how to do it. - Pablo Picasso
Additionally, in new large installations, IBM often oversees design and construction of the building (or floors) the computer is going to be in.
There are two types of people in the world: those who divide people into two types and those who don't.
Hah! I'm a system administrator, I can do nothing faster than any hardware out there.
home
5 Billion instructions per second.. Now; thinking back to my assembly language menomics, some instructions used less T states than did others.
So; is this 5 billion instructions per second an average?, or is this using one of the shortest instructions such as NOP, or LD???
(grin) Just curious.
(And yea; I expect to get the standard flaming about "who does assembly language anymore!!?")
All content in this message is copyright (c) 2008. All rights reserved. RIAA is prohibited here.
Very often, mathematical analysis will involve matrix algebra, systems of linear equations, systems of ordinary differential equations, or some other highly parallel task. Whether you use Freshmeat or Google, you'll find an extraordinary number of packages that are either highly multithreaded or use PVM or MPI to achieve parallelism. If the new POWER chips don't support out-of-order, maths users will suffer.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
However, previous generations will also need to be supported, so all of these programs will need to support both styles of code execution. Bloatbloatbloatbloat.
It's a small world and it smells funny; I'd buy another if it wasn't for the money; Take back what I paid (SoM)
inst core 2 duo doing like 24-36 billion of instructions per second?
(4-6 instructions per cycle *2 * 3 ghz)
i think he should speak about gigaflops or something where that really shows the cpu power,because 5 billion of instructions per second is not exactly impressive alone
i mean,even a single core 3 ghz pentium 4 cpu can do around 9 billion of instructions per second
From the Article:
"Sun spokesman Mark Richardson took umbrage at the focus on speed. "It's an easier marketing message to deliver to say that faster gigahertz means a faster processor," he said. His colleague, chip expert Fadi Azhari, explained how the Mountain View firm uses a different technical trick, called multithreading, to make a computer faster but not hotter.
Imagine a long line of airport passengers waiting for the ticket agent to check them in, Azhari said. The IBM speed trick would have that ticket agent working faster and faster - with maybe a blower overhead to cool the agent down. But multithreading would be like putting two or more ticket agents on duty, which is another less-heat-intensive approach to processing, he said."
What he should have said is, imagine a long line of passengers waiting for the ticket agent. The Power chip is like a catapult flinging those passengers through the gate. Sun's approach is simply to use more ticket agents. Who would you travel with?
This sig contains a manual self-destruct. Kindly please put your foot through your monitor in 8 seconds.
IBM has also patented this "holding finger in front of nose" process for measuring CPU speed.
It must have been something you assimilated. . . .
The Power 6 CPU is not just fast, but according to this paper from IBM it's also already dual-core and includes an SMP switch to allow up to 64 cores (32 chips) to run coherently.
That is, IBM has really fast ticket agents, AND they can fit a lot of them at the ticket counter.
So this computer will be faster for people with small hands? Or do they plan on varying the speed of light?
Bullshit. My 2.5GHz Penryn is faster than this thing.
This POWER6 has two 2-issue in-order cores at 5GHz.
My Penryn has four 4-issue speculative out-of-order cores at 2.5GHz.
That means that my Core 2 can retire, theoretically, 40 billion instructions per second (4 cores, 4 issue, 2.5GHz).
This CPU can retire, theoretically, 20 billion instructions per second (2 cores, 2 issue, 5GHz).
In reality, neither CPU gets anywhere close to that because real code has data dependencies, branches, and other pipeline hazards.
Every few years, someone comes along and claims that they can make a faster CPU by "eliminating the complexities of out-of-order execution". Intel tried with Itanium. IBM is trying with POWER6. But it doesn't work. Some code, with some compilers, can run very fast on an in-order CPU. But once you throw code that is branch-happy or has lots of data dependencies at the compiler, it goes and pisses all over itself. That's why you don't usually see in-order designs that are more than 2-issue.
Intel claimed that compiler technology would make up for the in-order nature of the Itanium. Like POWER6, it has lots of transistors and lots of cache.
That stuff may make POWER6 a good CPU in some applications. And unlike Itanium, POWER6 has high clocks to make up for reduced IPC. But the fact remains that this is a dual-core, in-order CPU in a quad-core, out-of-order world.
If it were true that we could just parallelize any problem on commodity hardware then you probably wouldn't see the NSA propping up Cray with orders for the type of processors that can crunch through non-parallel problems quickly.
There will always be a need for both.
OS X and all Universal Binary apps support both platforms, x86 and PowerPC. The work to accomplish this is done, in the bag. To maintain support for both platforms going forward would be relatively easy. As non-Universal apps fall by the wayside, users care less and less about what kind of processor happens to be under the hood.
With another breakthrough like this, IBM would pull ahead of the x86 players -- which would be a tremendous advantage for Apple, if it maintains the ability to resume building PPC Macs. The assembly line could make the switch often: this week we're installing PPC processors, but the buyers have a batch of Intel processors on order, and next week we'll install those. Whichever happens to offer the best price / performance ratio at the moment. Always riding the horse that's out in front.
Now for an unlikely scenario: there once was a version of Windows NT that ran on PPC machines. If IBM's performance lead becomes large enough, it would be in M$'s interest to revive that project. (Granted, it would have to be a pretty darn large performance lead.)
That that is is that that that that is not is not.
Same pipeline as the Power5, same power consumption.
Itanium has been trailing Power for a few years now in those benchmarks, often by 2x per core. Why will this change?
The GHz in Toshiba's project refer to wavelength.
The GHz in IBM's project refer to the frequency of the CPU clock.
These are different things.
The saddest poem
Sigh, I guess this meme really is dead. Apologies.
Power architecture processors use up to half their instructions doing in two steps what CISC chips do in one.
And the "instructions per second" metric is not a winner. A 3-GHz Penryn will do 12 billion IPS vs. this single-core 5-GHz Power6.
Not to take away from IBM that they can get 5 billion cycles per second out of silicon, something Intel and AMD may never (need to) accomplish. But it's not the fastest CPU in the world by any means (until IBM puts a few dozen thousand of them in a Blue Gene rack and it starts thinking out loud; then we redefine "CPU" in HPC terms and go, well, yeah, I guess it is the fastest).
This story is ignorant, fluffy, and wrong.