Intel Unveils New Chips to Battle AMD

An anonymous reader writes "Reuters is reporting that chip giant Intel hopes to get back on track in their continued market share war with AMD when they unveil a new line of chips at their upcoming twice-annual developers forum. From the article: 'AMD, once content to mimic Intel's advances, has set the technological pace in recent years with innovations such as putting two processing cores in a single chip -- moves that have helped it gobble market share from its much-larger rival.'"

Comments

Which innovation?

[hyc]

Of course, IBM had multicores years ago, so AMD wasn't really the innovator on that front.

Re:Which innovation?

[cfx666]

>Of course, IBM had multicores years ago,
So did SUN with their UltraSparc platform. But for the consumer market this really was something new.

Cfx

Re:Which innovation?

[dunstan]

No, IBM were first with the dual core Power. Sun have now leapfrogged ahead with Niagara, which not only has 8 cores but has four threads per core, so the OS sees a single processor as a 32 way system.

Re:Which innovation?

[Bert64]

But that's multiple physical processors on a single board, that's no more sophisticated than a dual processor motherboard.

Re:Which innovation?

[samkass]

There was large amounts of photographic evidence of the Wright brothers' accomplishments, some of which was lost in the Ohio floods early in the 1900's, and some of which survives today. Needless to say, NO ONE is documented to have flown out of ground effect, nor make a coordinated turn, until the Wright brothers demonstrated their plane publicly in France. By 1906 when Santos-Dumont made his little hop, the Wright brothers were flying for 20-30 minutes at a time at heights of 100 feet before spectators from the US Army as well as others in his town.

The Wright brothers didn't demonstrate publicly because they were in it for more than a hobby. Not being an independently wealthy tinkerer, they wanted to make their living making airplanes, and realized that they had the only viable design anyone had come up with, so not trusting the patent system, held out until they could secure agreements with various military organizations. They were engineers more than scientists.

Much of the "evidence" of earlier flight, including claims that Ader flew in the late 1800's, was concocted to try to overturn the Wright brothers' patents on their system of differing the angle of attack of the two wings in order to bank the plane. (Almost no one had banked planes before, either... most others were still thinking of planes like ships that would use the rudder to steer, which at those speeds every pilot now knows would lead to a stall.) Newspaper reports from before the patent battle clearly admit the Wright brothers unique invention, while those after the patent battle try to find almost anyone else to assign the invention to. As most know, though, the Wright brothers won every patent battle they faced and the only "evidence" of earlier flight lies in retellings of myths on sites like wikipedia.

Re:Which innovation?

[Sique]

The main accomplishment of the Wright Brothers was the steering. The principle of flying 'heavier than air' was shown to be sound before (Lilienthal et.al.), and the idea to have the plane being self propelled was obvious. It was just a matter of time until the gas engines were light and powerful enough.
But it was the Wright's analysis of the bird flight, and the realisation that you have to have bendable wings and tail/front flaps to get to a controlled flight, that was really new. Ironically it was this idea that was published in the patent application of 1904, which enabled the other flight pioneers to get their planes ready until 1910.

Re:Which innovation?

[default luser]

There is nothing "innovative" about Cell. Cell is basically multiple vector processors on a chip, and is a very predictable path for Sony after the release of the Playstation 2.

Each Cell SPE is simply a highly-optimized vector unit with 128-bit registers. It is capable of operating on 4 32-bit operands per cycle, just like SSE2/3 and Altivec. The difference is, eache SPE runs an independent task, while the Altivec / SSE units execute vector instructions in parellel with normal operations. However, the SPE is cut down: it has no branch prediction hardware or out-of-order execution, and depends on the main processor filling and emptying its Load /Store memory.

If you think the speed makes it innovative, think again: neither the Cell SPE nor its predecessor, the Emotion Engine, are IEEE754 compliant for 32-bit floating-point operations (for speed reasons). Cell can do IEEE754 compliant 64-bit floating point, but at an estimated speed hit of 10x, which makes it just "competitive" with existing solutions.

Sony / IBM actually inflate the performance numbers of the SPE, advertising it as 25.6 GFLOPS. But this doesn't take into account that the two pipelines of the SPE are NOT flexible, and can only perform certain types of instructions. The "Even" pipe can do arithmetic, and the "Odd" pipe can do Load / Store / Permute / Branch. Thus, the maximum arithmetic thoroughput per SPE is cut in half, to 12.8 billion arithmetic operations per second, and the double-percision performance is just 1.28 billion arithmetic instructions per second.

It's a nice idea for a media processor, but the complexity guarantees it will have a hard time finding buyers, and programmers for the Playstation 3 will be slow on optimization.

I mean, really, can you really break down a game into more than a few concurrent tasks without going crazy trying to synchronize it all? In addition, how many of those independent tasks can be designed with few or no branches? In your average code, branches make up about 20% of instructions. With an 18-cycle penalty per-branch, you'd have to keep those SPE branches under 1% of all instructions to avoid a serious performance loss.

Innovative dick comparison

[Opportunist]

Where's the innovation? And I'm not talking about AMD, Intel is just as guilty for equaling innovation with "make that damn thing run faster". Instead of shifting gear, they just basically upped the engine speed. 100 MHz, 600 MHz, 1 GHz, 4 GHz... now that the ceiling is more or less reached and enough waste heat is generated to heat a medium sized home, they change the measurement. Instead of length, we compare circumference. One core, 2 cores, 4 cores, 8...

Where is that innovation?

Re:Innovative dick comparison

[cfx666]

The innovation is that the good old software you are running does not automatically profit from cpu upgrades any more. So you need some new which is a good thing for me, cause Im a software developer.

Cfx

Re:Innovative dick comparison

[10Ghz]

And I'm not talking about AMD, Intel is just as guilty for equaling innovation with "make that damn thing run faster".

If they made their processors slower, then they would be "innovating"? What do you want processors to do, really? EVERYBODY wants their CPU to be as fast as possible. If you could choose between two identical CPU's, but one of them were twice as fast as the other, which one would you choose? the slower one? I doubt it. So why are you then whining as if making CPU's faster is a bad thing, since everybody wants faster CPU's? What benefit would there be in having slow processors?

And they have been doing pretty interesting things in order to make it faster. Pentium Pro with the on-die cache, SIMD, multithreading etc. etc.. Hell, even Cell with it's SPU's was designed the way it is, so it would be as fast as possible. But according to you, that's not innovcation?

ow that the ceiling is more or less reached and enough waste heat is generated to heat a medium sized home, they change the measurement. Instead of length, we compare circumference. One core, 2 cores, 4 cores, 8.a

Uh, they are still comparing performance, the means to get performance has just been changed that's all. They are NOT adding cores for the sake of adding cores. They are adding cores in order to increase performance.

But since you apparently think that making CPU's faster is not the way to go, why not share ith us what YOU want processors to do?

Re:Innovative dick comparison

[gormanly]

Well, there's been tons of innovaton at Intel. Even just looking at the CPU side, between the speeds you list:

100 MHz (1994): DX4 (P24C), Pentium (P54 version) - both, AFAICR were 0.6 um processes, and the DX4 had a 33 MHz bus and the P100 had a 50 MHz bus. I can't remember which was released first though. 600 MHz (Summer 1999): Pentium III (Katmai), the first rev of Pentium III, which was a new revision of the P6 core used in the PPro and PII chips. It had a new instruction set, SSE, and 512MB (external) L2 cache and a 100 MHz bus. Like the Pentium II, it also had Intel's MMX instructions for 64-bit SIMD integer operations. 1 GHz (Spring 2000): Still a Pentium III, though now with 133 MHz FSB and smaller (256MB), on-die L2 cache. No real changes from the 600 MHz version, but then it's only 2/3 faster again - and Intel were working on the Netburst architecture for the Pentium 4 and had somewhat taken their eye off the ball at this point. 4 GHz does not exist. Currently P4EE is at 3.73 GHz, but the clock speed race is over.

Intel gambled on Netburst, which was designed to get faster rapidly, and scale all the way from the 1.4 GHz at launch to 6 or 7 by now. Yes, they lost, but that doesn't mean that they weren't innovative - it's just that their process teechnology couldn't keep up, and failed to meet predictions. That's not the CPU designers' fault.

The earlier processors did scale fantastically well (486 16->120 MHz; P6 150->1400 MHz) but they hit an unexpected brick wall this time, so they've gone around it with clever scheduling and power management, and doing dual core versions of what is essentially a new rev of the P6. There's plenty of innovation in that chip too...

Also, remember that during the same timeframe, they've invented and developed the PCI, PCI Express and Universal Serial Bus(es). Pretty innovative, really, IMHO.

And yes, I'm typing this on an Athlon 64 and all 3 of my home PCs are AMD-powered.

Re:Innovative dick comparison

[khellendros1984]

I'm guessing that the gp is referring to picking a single aspect of processor performance, and equating that single aspect directly to performance of the machine, i.e. "1200Mhz will always perform faster than 1000Mhz". No manufacturer says this (that I know of), but so many consumers take it as a given, when the only number quoted to them concerning a processor is clock speed.

That isn't to say that I agree with the grandparent, though. Intel's Pentium-m processors are pretty nifty...lower power usage, high performance (compared to a P4).

So, while it may not be some amazing quantum leap, I'd say that Intel is showing plenty of innovation, at least from the standpoint of the consumer market.

Energy efficiency

[Lisandro]

"Multicore gives us the ability to get back on traditional performance growth lines," Intel Chief Technology Officer Justin Rattner told reporters on Monday. "We have become fanatical about energy efficiency. We have to continue to make progress in terms of energy efficiency."

    Does this means these new multicores will fry eggs even faster? I hate it when my meal isn't done in time!

Re:Energy efficiency

[Anne Thwacks]

Does this means these new multicores will fry eggs even faster? I hate it when my meal isn't done in time!

No - Sun manage to get four multithreading cores in their Niagra, and only run at 72 watts with 32 threads. see this

However, with Intel's cores, I expect be able to have a hot dinner faster than you can say Microwave".

Advantages and disadvantages of multicore

[replicant108]

Advantages

        * Proximity of multiple CPU cores on the same die have the advantage that the cache coherency circuitry can operate at a much higher clock rate than is possible if the signals have to travel off-chip, so combining equivalent CPUs on a single die significantly improves the performance of cache snoop operations.
        * Assuming that the die can fit into the package, physically, the multi-core CPU designs require much less Printed Circuit Board (PCB) space than multi-chip SMP designs.
        * A dual-core processor uses slightly less power than two coupled single-core processors, principally because of the increased power required to drive signals external to the chip and because the smaller silicon process geometry allows the cores to operate at lower voltages.
        * In terms of competing technologies for the available silicon die area, multi-core design can make use of proven CPU core library designs and produce a product with lower risk of design error than devising a new wider core design. Also, adding more cache suffers from diminishing returns.

Disadvantages

        * Multi-core processors require operating system (OS) support to make optimal use of the second computing resource.[1] Also, making optimal use of multiprocessing in a desktop context requires application software support.
        * The higher integration of the multi-core chip drives the production yields down and are more difficult to manage thermally than lower density single-chip designs.
        * From an architectural point of view, ultimately, single CPU designs may make better use of the silicon surface area than multiprocessing cores, so a development commitment to this architecture may carry the risk of obsolescence.
        * Scaling efficiency is largely dependent on the application or problem set. For example, applications that require processing large amounts of data with low computer-overhead algorithms may find this architecture has an I/O bottleneck, underutilizing the device.

http://en.wikipedia.org/wiki/Dual-core

Pretty light reading, but...

[porkThreeWays]

It's a fluff piece, but there was nothing mentioned in there to make me believe Intel is really doing anything new. All I saw was mention of 4 cores. Are cores the new mhz race? 2 cores is all 99% of people will see benefit from right now. The 4 core race is moot because it's like a race for automakers to produce the first production 16 cylinder family sedan. It's not going to really benefit anyone. Really only a marketing gimmick. I'd rather see Intel clean up their current 2 core chips.

Here's what most consumers need in a computer...
A low latency desktop that can handle about 2-3 running applications with no slowdown that runs cool and doesn't use a lot of power.

Here's what we are getting...
A high latency desktop with fat pipes that run hot, optimized for running 7-8 cpu intensive applications at once, and idles at 200 watts. Because it should take 10+ seconds to open a basic program on an out of box pc.

Re:Intel giving in to the pressure

[Nuffsaid]

No doubt, the heat is on their processors!

Naming Conventions

[Freaky Spook]

They also need to name their chips better to actually differentiate more simply between their lines.

Telling a customer the difference between a Pentium D, Pentium 4, Pentium 4 EE, Celeron D is hard enough without actually having to know what chips are out and what is offering the best performance for price. It feels a lot like market saturation sometimes.

AMD at least is a little bit simpler to follow.

Re:wanna compare cpu speeds?

[80 85 83 83 89 33]

sorry for the bad formatting, but the lamness filter is killing the proper layout.

factorial times for "100,000!"

look at the two athlons running at 2.0GHZ (3200+ and 2400+) and notice how it is frequency dependant

P4 3.2GHz 81 seconds

athlon XP 3200+ (2.2GHz socket A, barton)81 seconds

Pentium 930 dualcore (3.0GHz) 82 seconds

P4 3.0GHz (laptop) 90 seconds

Pentium 920 dualcore (2.8GHz) 90 seconds

athlon 64 3200+ (2.0GHz socket 939, venice) 91 seconds

athlon XP 2400+ (2.0GHz) 93 seconds

athlon XP 2100+ 106 seconds

athlon XP 2000+ (1.67GHz) 121 seconds

athlon mobile XP 1800+ (1.52GHz) 122 seconds

celeron 2.7 GHz (northwood core) 130 seconds

celeron 1.4GHz (tualatin) 205 seconds

athlon 900 (thunderbird) 228 seconds
(used msconfig to disable everything)

celeron 1.1GHz 253 seconds

celeron 800MHz (win98) 333 seconds (5min 33sec)

celeron 800MHz (XP pro) 373 seconds

PIII 800 (XP pro) 378 seconds (used msconfig to kill all crap running)
474 seconds (lots of junk running)

PIII 450MHz (underclocked coppermine) 490 seconds

PII 333MHz 686 seconds

PII 300MHz 760 SECONDS

P 166MHz 2417 seconds

P 100MHz ~4000 seconds (66 minutes)

P 75MHz 5330 seconds (1:28:50)

Re:Too Little, Too late?

[hyc]

What a sobering thought. "We've invented a space heater that produces computations as an operational byproduct."

IBM Power 6 @ 6Ghz

[CypherOz]

IBM have Power6 chips running at 6Ghz. IBM have been able to do 4 cores with this new technology.

Refer here

Re:IBM Power 6 @ 6Ghz

[Somegeek]

Unfortunately your link (or the article that it links to) doesn't say anything about 6 GHz. A little googling found some that did however, but they still talk about this as 'in the Lab'. I bet Intel and AMD can get stuff running at high speed 'in the lab' too. All of the stories that I have seen say that the chip will come out at 4-5 GHz, and not for another year.

It's also important to remember that one of the reasons that Intel is walking away from the clock speed race is that AMD showed that it wasn't necessarily the best way to higher performance. My point is that just because the new IBM chip may have four cores and a high clock speed doesn't mean it will be any faster than a chip with AMD's architecture. No one will really know until it's released and compared against whatever else is available at the time.

Link to an article that does mention the 6Ghz Power 6:

http://www.theregister.co.uk/2006/02/07/ibm_power6 _show/

Re:wanna compare cpu speeds?

[80 85 83 83 89 33]

sorry for the formatting, here it is easier to understand:

a very simple perfomance check i love to run on every computer i come across:

put windows calculator in scientific mode (yes, mathmatica or maple will do factorials in a fraction of the time, but try to post windows scores for comparison purposes....)

type in 100,000

hit the n! button

ignore the warnings that it will take a long time, don't even bother clicking on "Continue", because the calculation is still going.

and report how long it takes to complete a factorial of 100,000

please report what CPU you have

**64 bit XP is twice as fast

celeron 800MHz (coppermine): 333 seconds (5min 33sec)

1.4GHz celeron (tualatin) does it in 205 seconds

P4 3.2Ghz and Athlon 3200+ both do it in about 80 seconds....

Re:Too Little, Too late? My Arse!

[zaguar]

...your head stuck so far up your front side bus...

Ironically, the AMD64 series CPU's have no front side bus. This includes the X2 series. They have a hypertransport bus, which is similar but different. This is one of the premier reasons that the X2/Opterons scale so much better than the Intel equivalents, they do not have a saturated FSB as they have direct HTT links CPU-CPU.

Re:wanna compare cpu speeds?

[Anonymous Coward]

I'll add this to the list:
Opteron 146 (2.0GHz) : 43 seconds

Now I know what is the purpose of 64bit desktop CPUs - extreme Calculator performance!

Re:Too Little, Too late?

[10Ghz]

The dual-cores of this generation have been bought already, and the upgrade cycle won't start again until 2007-2008.

Dude, WTF?!? Are you saying that everybody who is going to buy a dual-core processor has already bought one, and next such CPU's wont be sold until 2007-2008?? What if someone decides to upgrade his computer in the summer (for example) to a dual-core machine? By your logic, he does not exist and/or he should wait until 2007/2008 because "that's when the next upgrade-sycle is in, you can't upgrade before that"?

People are upgrading their computer all the time. People are buying new computers all the time.

Competition good... Sloth bad...

[ursabear]

I think the competition has been good for all of us. AMD's strength in the market has kept Intel on its toes, and advances from IBM's Power processors has kept many architectures running pretty well.

I (this is IMHO) believe that Intel has been doing some laurel-resting for a number of years now. I do believe that they will come to bear with better stuff on a gradual basis. My only fear is that Intel will allow itself to do like GM, Ford, AT&T... allow itself to be way too slow to be quick to adapt. I personally would like to see IBM, AMD, and Intel all have truly great, smokin' processors going way into the future - it seems that it would only be good for us in the long run.

Re:wanna compare cpu speeds?

[Nuffsaid]

Ha! My super-optimized Gentoo x86-64 system, running on a 2 GHz Turion64 CPU, all unnecessary services and processes killed, under optimal condition (downslope, wind from behind, air temp below 10C) uses exactly zero seconds to tell me "Error" in kcalc...

TRANSPUTER

[poptones]

I'm pretty sure transputer predates IBM's multicore POWER. Furthermore, transputer was inherently multi - up to four cores on a die and they could be interconnected easily via into larger arrays.

Keeping the options afloat

[the eric conspiracy]

Intel has a serious problem in that they are perceived, and rightly so that to be a technical laggard. They are bleeding market share and their stock price has dived.

As a result Intel is trying to revamp their product line to become more competitive - but to keep from losing customers they are trying to darken the sky with marketing. This will work for a while because Intel has some credibility amassed from its earlier successes.

But if they fail to deliver at least parity with the next round of designs they are going to lose market share as fast as AMD can build Fabs. And right now they are running the risk of the 'Osborne Effect' - promising new product so attractive that the company loses large sales volume on current sales.

So Intel is making some really big bets here. If we get into the same time frame in 2007 with AMD still having a clear technical lead we could see AMD and Intel all of a sudden having a 40/60 split in market share, and a duopoly where once there was a monopoly.

Pentium M differs quite a bit my friend....

[macavity1977]

You *REALLY* need to look at the Pentium M...

OS: Slackware Linux (Current)
Application: kcalc (Comes with KDE)

These are both ASUS laptops with PC3200 RAM:
2.8 GHz Celeron:     65 secs
1.6 GHz Celeron M: 18.5 secs

This kind of makes you wonder now, doesn't it? It appears that the Pentium M achieves *quite* a bit more per MHz then the Pentium 4.

Aside from that... the calculator in windows is obviously a joke, as the 1.6 GHz machine took 118 secs to do it in WinXP   >_<