The Dual-Core War - Is Intel in Trouble?

michaelMSFT writes "Cooltechzone has a column stating that Intel has already lost the dual-core war against AMD. From the article: 'From the performance numbers published on numerous online publications, Intel has lost the Dual-Core War. The only competing factor that Intel has right now is the possibility to keep their prices low enough to attract those with strict budget...I would like to forward a special note to Intel: Please make sure your next generation of processors aren't as atrocious as the Prescott, as AMD is making you look pretty silly right now.'"

I Disagree

Intel has not lost anything. They might be getting their asses handed to them by AMD -- but remember that it often takes huge losses before a company changes its approach to doing business. And that kind of change really is needed at Intel.

Re:And Furthermore...

And furthermore... since when is it whoever ships first wins? What about quality? Cost? Yes AMD beats Intel on both of those now, but that is what I mean when I say that changes at Intel could make them far more competative, far more agile.

Wars aren't won or lost on the first shot. Wars go on for a long time.

Do you remember Cyrix?

And the first Intel Celery's? I do. Cyrix and it's ilk won every perf/price war, and the early Celerons were a complete joke. Scroll forward a few years...Cyrix is gone, and Intel rules the low cost microprocessor roost. Remember Via and Transmeta, and the 1W processor war? I do. Via and Transmeta were handing Intel its head -- blades, this, that, the other; low power was king. Errr...oops. The Mobile Pentium chip (and the associated platform strategy Intel used) won the day. Again.

So now we've got AMD and the Athl/opteron. AMD has done better than any other competitor, so far, and has managed to maintain a narrow performance gap for several years. On a couple of occasions, they've opened a wide performance gap for a short time, but Intel has always closed it to a narrow one.

Neither great technical merit for a short time nor slight technical merit over a long time is enough to establish a market. The competition can always work around that by starting a skunk-works project, as Intel did to Cyrix, and Transmeta, and, more recently, with their implementation of x86-64. To catch up with AMD now, all Intel needs to do is build a dual-core x86-64 chip with a smaller power envelope.

Didn't I hear that x86-64 and dual cores were coming to the Mobile Pentium class soon? Hmm...I wonder why?

Re:And Furthermore...

Intel actually shipped dual core first (by a couple of days) to pick up bragging rights, but as you said, it was rushed and we're seeing the results of that now.

Intel's problem is that they're tied to the inefficient Netburst architechture for the time being, and it's really just hit a complete brick wall. The move back the the P6 architechture of the Pentium M is in the works, but that's going to take time.

In the meantime, AMD has the K8 which absolutely brutalizes Netburst in performance per clockcycle, and which, at this point, also seems to have plenty of headroom left.

The biggest problem Intel has is that, in the processor business, you can only be so agile. Yes, they've hacked in x86-64 and dual core support to Netburst, but their implementations are just that - hacks and bandaids. The result is what we're seeing now - AMD with a highly efficient and what's looking to be scalable (at this point) architechture, while Intel is limping along with Netburst that's really been on its last legs since Prescott was introduced 14 months ago.

The Pentium M shows some promise, and is probably Intel's best processor design to date, but has problems of its own (ie: its relatively weak FPU performance compared to K8 and Netburst both). It's entirely possible that Intel will come up with something in the long run that will straighten them out, but as it stands now, I wouldn't expect it in any time frame less than 18 months from now. The first dual core Xeons will likely be Prescott-based, and suffer all the same fundamental flaws Intel has been fighting for the past few years. By that time, AMD may have already had a chance to entrench itself in data centers, which would be a huge loss for Intel. That's why this is big news.

Re:I Disagree

As for difficult to test... not really, or at least I don't see why it would be worse than clocked chips. Have a test signal, put a stream of data into the chip, then see what comes out. Async should work the same as sync in that case...

The problem has to due with the variations in timing due to temperature and voltage changes. In general, synchronous logic is simulated at the speced max temperature and lowest voltage (this is typically the worst case (*)). With asynchronous logic you have to test over a range because you have to worry about things happening too soon rather than just too late, and as you add more parallel and sequential circuits things get really complicated.

(*) I'm not positive this is the worst case combination...

Better Review Over At...

The best comparison of the dual core reviews I've read is over at the great anandtech [anandtech.com] site.

AMD's push with dual core into the server markets half a year before Intel's dual core Xeon arrives is going to tempt a lot of IT departments out there.

On the desktop side, we are extremely excited about the Athlon 64 X2. The 4400+ that we compared here today had no problem competing with and outperforming Intel's fastest dual core CPUs in most cases

The real problem is that AMD has nothing cheaper than $530 that is available in dual core, and this is where Intel wins out. With dual core Pentium D CPUs starting at $241, Intel will be able to bring extremely solid multitasking performance to much lower price points than AMD will. And from what we've seen, it looks like that price advantage will continue for quite some time. It all boils down to economics, and in the sense of manufacturing capacity, Intel has AMD beat - thus allowing for much more aggressively priced volume dual core solutions.

Conclusion: AMD have better chips. But they don't have the manufacturing capacity to bring them out in volume. So they focus on their higher margin chips. Meanwhile Intel keeps from losing face by selling at the volume, lower-priced end of the market. At least until AMD get some new fab plants up and running.

--
VPS Hosting on Dual Xeon Hardware - but just for the time being :) [rimuhosting.com]

Re:Better Review Over At...

I find this all very amusing actually, since until recently AMD ruled the desktop/home PC market based on price/performance, whereas Intel maintained its market share because they were so datacentre friendly. When I'm buying racks of enterprise 2, 4, and 8-way x86/x64 processors, a price difference for $100 per CPU is a small component of the overall cost, and frequently is worth the up front capital cost for a 10-20% performance boost.

What I'm seeing is AMD is going to begin kicking ass in the enterprise space for enterprise rack servers and blade configurations, a traditional domain where Intel has ruled. And as for dual-core on the desktop, I don't think the market is really there for that level of performance yet... not many desktop apps can take advantage of those features, just like x64 is just future-proofing your destop for the time being.

So the immediate price difference between AMD and Intel offering doesn't tell the whole story. Intel is going to get hit where it hurts the most -- enterprise markets.

Hack! And not the good kind...

I had a root cannal done yesterday, so it might be the Percocet talking:

This article is total hackery. Any two comments will have better background and more insight than TFA.

It's just a "AMD is better!" article that mentions dual core CPUs for some reason. No context, no information.

Re:Hack! And not the good kind...

It ain't your medication - I felt the same way.

The guy says "I think" and "I feel as though" too many times for me to take it any more serious then a forum or Slashdot post.

It's some dude that thinks he can be an online "journalist" by posting garbage like that on a web site.

Silly?

as AMD is making you look pretty silly right now.

Yeah - but the thing is - is the performance worth twice the price? Being that's the only way you can go with AMD right now - paying twice as much (if not more) for their dual core chips compared to Intels.

Funny that.

Ah the usual /. overstatement

Massive, multi billion dollar corporation suffers small setback and so doom is imminent.

Intel will only be in trouble when Dell, HP, Fujitsu and every other major manufacturer drop them in favour of AMD. Until that happens its business as normal.

Re:Ah the usual /. overstatement

Exactly. I'm a huge AMD supporter but all you have to do is look at last quarter's financial reports to see the bottom line here. The Athlon 64s have pretty much been kicking Intel's ass all over the place in the applications that most people use for the past couple of years and they still make a small fraction of what Intel makes.

I would agree with other posters that right now AMDs largest issues are capacity and marketing (people need to ask for AMD) as the technology is there.

Intel 0-2

It lost out in 64-bit acceptance and dual-core, so it seems. But apart from tech-aficionados, the world is 32-bit x86, which is Intel's domain.

What does that mean for the future? Absolutely nothing. Until and unless the world switches to 64-bit or dual-core computing in droves, Intel still has time to catch up where it matters.

IBM is playing it smart, however. It's investing in consumer electronics with the Cell. That is growing faster than the desktop or server market.

Even if AMD is beating Intel, it has nothing in the consumer electronics domain.

Intel hasn't lost until Dell sells AMD

I seriously consider that one of the biggest factors in this so called CPU "war".

The other thing is volume. I believe one of the reasons AMD is able to create superior processors is because they don't crank out nearly the volume that intel does.

If they did it would take them longer and cost more to upgrade the fabs each generation.

Top Down vs. Bottom Up (no it's not dirty)

Intel and AMD's approach to the dual Core (and to chip production in general since the A64) producion has one fundamental difference that will shape their incomes and profit margins.

AMD produces their chips Top down, they introduce dual core in the professional space first. In this case the 8xx series, then 2xx then finally 1xx and the desktop space. This allows them to take advantage of the increased profit margins and lower volume of the professional space. While their chip producion is ramping up they don't have to worry about demand outstripping supply. Thus they maximize their profit on their smaller fabrication abilities. The con to this is market penetration is smaller and validation "should" take longer.

Intel OTOH goes bottom up (and don't get me started on whether the pentium D is really dual core) producing first their desktop chips, and then when they're properly validated and their market presence and fab capabilities have been fully leveraged, then they move the chips into the professional realm, this allows the to maximize their profits as a fab that caters to volume.

Declaring a winner at this point is silly, as neither one has actually completed their cycle let alone vanquished the other.

The power of marketing

The only competing factor that Intel has right now is the possibility to keep their prices low enough to attract those with strict budget...

I disagree, and I say that as someone who hasn't built an Intel machine in years. There are a lot of regular Joe consumers out there who are barely even familiar with AMD, primarily due to Intel's aggressive marketing (ding-ding-ding-DONG) and their partner relationship with major manufacturers (cf. Dell).

Good marketing will keep inferior products afloat for quite some time.

Only an opinion piece - don't bother reading

What a waste of 5 minutes of my life and a few hundered K of downloads :-\

Turns out this was nothing more than an extended opinon piece (yes, yes, I know that's perfectly fine) but I was looking for something with some hard-core comparisons, especially since they started out saying "we'll just compare the desktop scenario to keep matters simpler" --- *click to next page* Aw what!? nothing!

Sounds like someone was just after some slashdot publicity.

Even worse -- AMD can't buy Intel's dual cores

The Inquirer points to an Info Week blog via Silicon Investor that basically says that AMD can't even purchase dual core Intel CPUs to benchmark them against its own offerings -- they're not available for sale anywhere:

http://www.theinquirer.net/?article=23018 [theinquirer.net]

I agree that the costs of the AMDs are exorbitant right now as they migrate their production, but if and when they get their fabs worked out, prices could really drop and even things up on that level.

I guess the real concern, though, and some have already noted it -- so what? Until I see an AMD dual core CPU option on Dell.com's various stores, Intel isn't going to be hurtin'.

IronChefMorimoto

It is all about the Memory!

Dual Xeon systems (even with DDR2 RAM) cannot supply enough memory bandwidth to keep both CPUs running. We are lucky to get 80% of the single CPU performance when running two instances of a scientific app on two CPUs. Adding more cores just makes the situation far worse as now four CPUs will be competing for that same memory path.

Opterons are way ahead here with their built-in memory controller and dedicated memory banks for each CPU. Intel's SMP folks really need to pull a rabbit out of their hat and right quick. The last cluster (256 CPUs) I built used dual Xeons because they were still slightly faster on our applications over similarly priced Opterons in spite of the degraded SMP performance. Next time around, I doubt that will be the case.

Article Summary

From TFA:

Personally, I think Intel has pretty much lost control of the enthusiast segment.

This nicely sums up the entire article...a two page personal opinion, berift of any real facts, statistics, or figures.

This 'story' was pointless.

Prescott

Please make sure your next generation of processors aren't as atrocious as the Prescott, as AMD is making you look pretty silly right now.'

I still haven't figured out why anyone would want to name a processor after John Prescott [google.co.uk], British Deputy PM and Eater of Pies.

What's next? The Intel Widdecombe [google.co.uk]? The mind boggles.

Why are we reading somebody's rant?

There is no news in this article, just a splashy title. Intel will eventual build something based on the Pentium M. The Pentium M is much more efficient and will be more so in the future. The next generation of dual core Pentium Ms will be an interesting challenge for AMD. I still think AMD will win, but the war is far from over.

Bored with hardware?

I remember when was I was in my teens (15 years ago) I used to love reading about the latest hardware advances.

These days you'd find me hard pressed to get excited about anything hardware related. I have a fast system with lots of ram and a decent GPU, and thats all folks.

Much prefer to pay a raging teen money to build the damn thing for me.

Do people find themselves as they get older more interested in software design, algorithms, and licensing debates?

Hardware just doesn't do it for me anymore.

Intel Ok, AMD Ok, Everybody Can Win

I don't think it's time to bury Intel yet, or AMD for that matter. If I've learned anything over the last many years, the better technical product doesn't always go on to dominate the market. But I would say that market domination is not necessarily needed to make a nice profit. AMD is making better performing chips than Intel right now, and with the dual core chips they should be making a nice profit with premium pricing. The market for such chips is currently limited to tech enthusiasts (such as PC gamers) and those with specialized computational needs. That's okay because with the right pricing that could be a very profitable niche. Apple is doing pretty well right now without being anywhere near dominant in the personal computer market due to a fiercely loyal user base, premium pricing, and some genuinely cool products. Apple is not the biggest, but they are profitable. AMD can do likewise.

Intel of course has nothing to worry about. They have the volume customers through deals with Dell, HP, and others. And they are picking up the low end of the market for dual core chips. These are not as good performance-wise as the AMD chips, but the lower cost is appealing to a more budget minded market. And with economies of scale, Intel certainly makes a nice profit. So it is unlikely they'll be shutting their doors anytime soon.

Everybody can win here. AMD has the harder job for now because their survival depends on continuing to produce better chips and growing their niche of customers. But as long as they make a profit, they can stay in the game.

Re:Any dual core vs dual cpu benchmarks?

In the case of the AMD Opteron dual cores?

Pretty simple: Slightly better than two single cores, mainly due to the fact that the Opterons are close coupled, and teh CPU cores commun icate directly in the package over a Hypertransport channel.

In the case of the Intel dual cores?

They have no direct communication, no HT bus, and all messages and RAM access has go to the northbridge chip and face a FIFO queue to memory.

In the case of thermal/power performance?
At similar performance the AMDs are using about 60% the power, and making corresdpondingly less heat than the Intels.

Despite what the "Intel fanboys" are saying, Intel HAS lost this war at the first battle.

And Dell will keep selling Intel, for a while.
Until it start to hurt the bottom line.

In the meantime, the changes are already happening.
SuperMicro, who NEVER build boards for anything but Intel CPUs, are now selling a dual Opteron board.

The Emperor really does have no clothes!