Slashdot Mirror
IBM Opts for AMD
ExE122 writes "Since the unveiling of the low-cost, low-energy AMD Operton in 2003, Intel has been struggling in the server-grade processor insdustry. Now, IBM has announced their decision to use the AMD Opteron processor in their new line of BladeCenter servers. System x3455, x3655 and x3755 rack-mount servers, two-way Bladecenter LS21, and four-way LS41 blade servers sporting the new AMD processors have already been announced. IBM will continue this transition over the next three months.
From the article:
"IBM's choice is by all means an important victory over rival Intel, which is struggling to sell the remaining deposit of server processors before the general acceptance of Woodcrest X5100 chips. Unfortunately for Intel, at the end of the second quarter, Advanced Micro had 26 per cent of the market for servers built on personal computer chips, more than double its share a year earlier, according to Mercury Research."
Could this be lights out for Intel?"
From the article:
"IBM's choice is by all means an important victory over rival Intel, which is struggling to sell the remaining deposit of server processors before the general acceptance of Woodcrest X5100 chips. Unfortunately for Intel, at the end of the second quarter, Advanced Micro had 26 per cent of the market for servers built on personal computer chips, more than double its share a year earlier, according to Mercury Research."
Could this be lights out for Intel?"
Sigs cause cancer.
You'd think that with AMD sponsoring Slashdot, they could at least spell 'Opteron' right once or twice in stories...
using namespace slashdot;
troll::post();
You just opened the floodgates, Mister. But yes, I love AMD - look at my sig for the sake of /.
Intel is going nowhere, however - there are far too many consumer-oriented PC corps out there that adore Intel. And sheesh, AMD has been on the short end for so long, it's hard to imagine that a corp like Intel couldn't wait it out, too.
--I gots 99 problems but a new machine ain't one!
AMD! Asus! Whoot! 6 years!
Could this be lights out for Intel?
No. No it could not.
If you're going to be elitist, it would help to be elite.
Intel are still ahead in market share, and have just released some very competitive chips.
I'm an AMD supporter, but the near future is them trying to hold the ground they've recently taken, not expanding further.
(And Intel probably the reserves to stuff up again, be uncompetitive for a few years, and still make a comeback with the next generation of chips.)
Quattuor res in hoc mundo sanctae sunt: libri, liberi, libertas et liberalitas.
It's kind of odd how everybody is jumping on the AMD train when Intel is finally having viable products with their new architecture (For instance, Dell finally jumping on board).
Paul Otellini : "AMD has 26% of the market? Well, screw that - if we can't have 75% of the market, we're outta here. Call up Slashdot, let 'em know we're closing the doors tomorrow."
Better question: Is this lights out for the Power line?
IBM seems to be giving up on their Power cores. That's what concerns me, because it looked like they had a big shot of gaining territory in the gaming-and-entertainment market.
No. Why?
Let's see... 100 minus 26... carry the 9... that leaves 74% share left for Intel, right? I'm going to have to go with "No, doesn't look like lights out."
Please, quit it with the retarded questions at the end of the article summaries.
--
Carnage Blender : Meet interesting people. Kill them.
Of course, as we all know, THIS the year that Linux takes over the desktop, and MS goes down the drain.
Gee, I love living in Slashdot Land!
You know this switch was coming sooner or later. AMD already does a lot of their serious R&D at IBM. They'll be the same company within 5 years.
Core 2 Duo is posed to dominate the desktop market unless AMD comes back with a strong chip ASAP.
It seems to me intel will gain back some lost market share with the Core 2 Duo.
It's ridiculous to add the "end of intel" comment to the end of the article.
All things are subject to interpretation, whichever interpretation prevails at a given time is a function of power and n
I think TFA misses an important point. It's not whether Intel or AMD captures the entire market, or what market share these two players have. With only two major players, I'd say the main problem is that we have too little competition, not too much!
Regards,
Art
Is this the Apple Curse coming to fruition? Any chip supplier Apple picks seems to constantly have problems. It happened with Motorola, it happened with IBM, now Intel for goodness sake! Let's just hope Intel doesn't start having major production issues out of the blue.
Because, you know, the ADM Operton kicks the Inlet Xeno's butt.
Although I agree that less than +4 funny shouldn't be worth mod points, your post shows how much fairer it would be if funny mods countered down-mods, IMO.
I'm amazed at what's now called troll or flamebait, but it's actually a general rising tide of political correctness; we're "legislating" for people's feelings. Scary stuff.
Wikileaks, no DNS
Except I was thinking that even if Intel chucked their whole x86 line, they still make a boatload of other chips. Like XScale, for instance. Their previous line of ARM processors (the SA-1100 family) are freaking *everywhere*.
Weaselmancer
rediculous.
Even with their new Woodcrest CPU finally making it out of vapourware status, Intel has no real answer to Opteron in the 4P and above server space. Itanium is a failed venture that is on life support. It often performs worse than Opteron systems much much cheaper, so no hope there.
Intel might regain a little marketshare in the 1P/2P server space with Woodcrest, but they're still in full retreat in the Enterprise market as more and more companies move to 4P+ servers.
Seeing as AMD are releasing Socket F and Rev G chips soon, much of the performance advantage Woodcrest might have had will vanish quickly.
I recently Googled "AMD Quad Core" and clicked on the first or second result and read the article. If this article holds true, AMD could very well blow Intel off the map and into orbit. The article said that the recent buyout of chip maker ATI is part of a grander strategy by AMD to take a bite out of Intel. The article said that the current CPU dye made by any manufactuer contains 18 individual components, minus the cores, to create the CPU dye. AMD's 4x4 quad core slated to launch in early '07 is being rebuilt from the ground up. AMD is going to attempt to modularize a CPU dye to allow for quicker, cheaper, and easier manufactuering. By that they mean that each individual component will be interchangeable and have an on dye socket to be plugged into. A good visual image of this is building blocks. Identically shaped and sized units rearranged to create a new structure. The article said that the only difference hardware wise between an Opteron and an 64 X2 is 3 components. If AMD is successful in modularizing the CPU dye, this article estimates that AMD will have "entry level" 4x4 CPU's in 8000-9000+ range avaliable to CONSUMERS for around $400-$600 and industry quality models at around $1000 on the low end. Only time will tell if this is true, but for me, I hope it comes true as I'm being asked at school by the teachers as to who will have the better CPU in the future and my answer is "AMD of course.".
IBM already has bladecenters with opterons... why is this news? How is this a defeat for Intel? IBM is basically announcing a refresh of their current lineup... you can bet your ass they'll be doing the same thing when intel rolls out the new Xeon in full force as well.
Remember kids, just because you like to pull for the underdog, doesn't mean it's OK to make false statements about the king.
Intel dead? Have you people been living in a cave for the past few months?
Look at some benchmarks. The new 5100 series Xeons with the Woodcrest core have been out since June, and a dual Woodcrests crush dual Opterons in almost every test. AMD's only hope at the moment is HyperTransport, with which they rule the market for 4-socket servers (Intel's old-fashioned FSB doesn't really scale to 4 sockets). But thanks to Core2 (Conroe and Woodcrest), Intel has taken over the 1-socket and 2-socket market. Prepare to see AMD's market share take a nosedive.
I am generally an AMD fanboy, but my next system will use Intel chips. Now that Core2 is here, I am simply not interested in an antiquated AMD chip which can only complete an SSE2 operation once every two cycles. Until the K8L comes out, it's Intel Inside for me.
It's definitely lights out for this writer...
I hope so
Actually I have never owned a computer with a intel processor so I dunno.So,pretty much my post is useless
Wow, AMD Really are spnning evrything they can get their hands on in the last few weeks - could it be that they are trying to divert attention away from something?
Let me think.. what was AMDs last real news?? When is the next major milestone in their processor lineup?
What goes around comes around.. Intel have been busy beavers for a while, relying on their rapidly aging netburst architecure, and hurting for it, while they got their next generation in order - perhaps AMD should have done a bit more work to have an answer ready... oh well.
IBM of course will do anything to divert server attention away from Intel, due to the fact that they have their own large-systems architecture to support - Power, and Intels ia64 is a competitor, whereas AMD have nothing even close to that market - of course IBM want people to like AMD and avoid Intel...
I think the correct technical term is 'Duh!'
Could this be lights out for Intel?
Do the editors really think that adding stupid little phrases like this to every article enhance the experience? Why do we need to read this every time that something happens that gives an avatage to Intel or AMD? Sheesh, give us a break.
Minnesota twins batters hit 2 home runs in yesterdays game. Could this spell apocalptic doom and drawn out painful deaths for the New York Yankees?
AMD needs competition just as Intel does. More competition = good.
Actually, Woodcrest is trash at anything more than 1p (well, not exactly trash, but not noticeably better than any of AMD's offerings, and loses on price/performance). Much of the major improvements in Woodcrest, such as some of the more significant microcode and cache optimizations, only work well in a 1p setup, and are outright disabled in 2p+.
So yeah, right now they'll beat anything AMD has in a 1p setup, so it looks like Intel is coming out swinging hard in this round of the desktop wars, but it seems like they're set to take a licking in not only the 4p+ server arena, but also the workstation market yet again.
Yes, Hypertransport is better than the FSB, but for up to 4 processor systems, I don't think Intel is too disadvantaged. And Intel is working on CSI which they claim will beat Hypertransport and will be out in 2008. Will it live up to their claims? Who knows, but when Intel is pointed in the right direction (the management isn't throwing around their money like a spoiled trust fund kid) they seem to put out some pretty nice stuff.
Everyone keeps talking about K8L. If they can deliver it in the short term (hah!), great, but what they really need is K8 Rev. G.
A current AM2 K8 dual core with 512k+512k cache is similar in size to Conroe with a 15% (roughly) disadvantage in performance (at common price brackets). Moving to 65nm will drastically reduce die sizes, allowing AMD to squeeze more chips on each silicon wafer, even compared to Conroe.
Consequently, AMD will be able to sell their chips to us at really cheap prices while still making a good profit, building a war-chest for when K8L faces off against CxQ. Then we repeat the whole process again when Intel moves to 45nm.
I think having a 4m L2 cache might be a bit of a boat anchor around Intel's neck when it comes to manufacturing. Time will tell.
check this out:+ luis+obispo,+ca&daddr=lee+vining,+ca&ip=10.10.10.3 &ie=UTF8&om=1
http://maps.google.com/maps?saddr=86+johe+ln,+san
it's where the new fab facility is.
... . Chrome E Eye Purr Si Eye Opertron Eight.
Planet
Kip! Grab the nail clippers. I have an invasive infection on toe unit 16.
Slashdot, aka Pimp My Clicks
Yep, click-whoring by adding a troll sentence to the end of each front-page summary.
You're forgetting two very important factors:
1. More cores does not necessarily a faster CPU make. Simply adding more cores isn't going to be enough for AMD. Every core you add tags on overhead... in many cases I would venture to guess 8 cores would be SLOWER than 1.
2. Exactly how many applications out there right now are able to take advantage of 8 cpu's? Most everyday business applications people use aren't even SMP aware.
Yes, of course. Absolutely, indubitably this is the end for Intel. Anyone who argues otherwise is clearly a fool, an Intel phanboi, or someone with a lot of Intel stock.
Flying is easy, just throw yourself at the ground and miss. -Douglas Adams
This isn't just about more cores, think bigger.
:)
How about on chip physics processing?
On board graphics processing? Say pre-processing before its sent to the GFX card
Soundcard on a chip perhaps?
Heck you could have dedicated modules added for ANY type of application in big business, best part is because its using the same base components you have the advantage of mass production with the bonus of specialization for much less of a price!
I thought IBM were chucking the new Uber-chip the Cell into blades and this was going to revolutionise the world [/hype], did it not work? Or haven't they actually put these on sale yet?
I would have though that this would at least muddy the waters a bit with the whole Intel vs AMD war on IBM servers.
If this were really happening, what would you think?
It is Opteron not Operton!
--
Dyslexics of the World, UNTIE!
there are a couple of architectures to count in for those 74%
One man, one word.
Another reason why these tricks don't really scale to more than 2 sockets is that the two Intel CPUs still have to share one memory controller (dual channel of course but still a handicap), while AMD gains a memory controller for each socket.
So for applications that need plenty of memory bandwidth, Intel still has a bottleneck in the larger servers. The 4MB cache cannot hide that completely. I suspect that AMD will start to feel the pinch too when they go to 4-core CPUs.
For Intel, I guess that two sockets with four cores each will be too much for one memory controller. At that point, Intel might need to adopt HyperTransport.
C - the footgun of programming languages
What you say is right in a narrow sense, but why would we have become so much more sensitive?
It is the rise of political correctness that is "legislating" by altering acceptable behaviour. Increasingly criticism is unacceptable. This is far wider than Slashdot.
Wikileaks, no DNS
when a slashdot editor writes "could this be lights out for companyX" when companyY is a sponsering vendor.. I have a feeling you guys are heading twards shark-jumping waters fast.
this is a design win for AMD, yes. and an important one at that. but this is a far cry from lights out for intel. jeezz.
I think IBM had some of their 486 clones made by AMD, so their working relationship goes back at least to the early 90s. IBM's sold Opterons in smaller servers for quite a while.
Seriously, I've been looking for Woodcrest benchmarks, and have found zilch. I've suspected that because Core 2 only sees major gains from 1P systems, they've concentrated on the 1P solution marketing blitz, to get the most bang for their buck. After all, if benchmarks come out that 2P systems perform only 50% better than a 1P system, and AMD's is almsot 90% (or whatever), that's going to remove a lot of zing from their marketing blitz.
So, where are these benchmarks?
The cesspool just got a check and balance.
Intel's own tech documents state that their famous "Micro-Ops Fusion" does not work in 64-bit mode. This feature is one of the main features that make the Core2 architecture as fast as it is. Losing this knocks performance down around 5 to 6 percent, which puts Opteron just that much closer.
xbitlabs has some more details.
No one is trying to say that Woodcrest isn't a good 64-bit chip. We are just telling it like it is: woodcrest does not gain much from 32 -> 64-bit code (in some cases it is a bit slower), while the Opteron gains decently almost across the board.
I am a viral sig. Please help me spread.
Grubman. Need we say more?
BTW, that's a classic reverse of the pump and dump scheme, berate and buy. Take a look at IBM's choice for server hardware, followed by Sun, and HP. Do you think that these guys are all wrong? Or could it possibly, just possibly, be that analysts' primary job is to cause churn in the market, as most are paid by trading houses? Personally, if an analyst is screaming "buy buy buy" it's time to "sell sell sell" and possibly sell short. NOTE: this is personal opinon only, for financial advice pick your poison.
The cesspool just got a check and balance.
For 4-way MP, one of the reasons that IBM Xeons (like x366) have been excellent performers is the sophisticated memory controller hub chipset ("X architecture", the 'Hurricane' chipset) which has an advanced L4 cache and snoop filter integrated. What IBM now calls the x3850 (basically same as x366) still has this advantage - and combined with the new Intel 'Tulsa' Xeon MP chips with huge L3 caches, IBM should probably outperform their generic design x3655 Opteron (and other Opteron 4-way boxes). The real reason, I suspect, why IBM is doing this is not so much technical (they have felt their Xeon MPs were competative against Opteron 4-ways) but that there is now an AMD market mindshare which had created a disadvantage in some corporate accounts that had adopted the other AMD 4-ways (generally the HP 585) and were happy with it - IBM felt shutout wherever AMD fanboys became dominiant in IT thinking.
In the 2-way DP space the new IBM x3650 with Xeon 5100 series Woodcrest (and any other Woodcrest based DP) will be as good or better than anything AMD throws at the DP space in 2006 -incl the new 2.8GHz F socket stuff.
We are talking about servers, so we are not talking "most everyday business applications people use". That means 8 cores will be faster when the server is doing many thing in parallel, which unless you are a bad SA is what a server should be doing. The AMD HyperTransport is a decendent of the Alpha chip, which did scale nicely to 32 chips.
That said, there are some limitations. SUN nifty 32 core processor is a POS. About all it does well is web serving. They provide a profiler to determine the suitablity of a given program to run under Niagra. We have yet to find one other than Apache and we think it is because while it has 32 cores, it has one math co-processor.
Never answer an anonymous letter. - Yogi Berra
That sounds like an extremely BAD thing to me, ESPECIALLY for big business. Now, instead of being able to hot swap out your dedicated SSL accelerator, you suddenly have to replace an entire CPU?
Same goes for home users. Instead of being able to replace that 50$ soundcard, now you have to buy a new 800$ cpu or go without sound... Integration isn't always a good thing.
You would think IBM would make their PPC chip their main choice or force them to improve PPC to match their competitors.
Looking forward maybe one year down the road, Intel looks to have the best performance/watt server solution, not AMD. IBM, however, will be stuck with Opterons.
I understand; you're saying "don't stand for political correctness by pretending that it's not there", that is: do the right thing, and refuse to inherit ridiculous standards. Act deliberately "blindly", but reasonably in your own terms.
I agree that doing the right thing is an important part of the fix, but it is not in itself enough. In brief, you need a network to fight a network. The PC network works through "building awareness", and this is therefore an important part of the response. If you don't do this, you will find yourself "showing insensitivity", and it will hurt your career. This "insensitivity" has nothing to do with real compassion, but is rooted in self-justification for minding one's own back. Weakness is promoted into "morality" of the weakest kind, that has nothing to do with courage.
Here's an example of poor humour on Digg. PC nonsense and ridiculous thin-skinnedness needs to be pointed out, and this will occur at some (slight) cost to oneself. Ironically my own post was modded as Troll, as if it was intended to get a rise out of people and generate replies as sport, rather than state an opinion and make people think.
When I say legislation, I agree that social pressure is not in itself legislation, but it becomes legislation when one is considering "hate speech", for example. Hate speech used to mean speech that has the intent behind it of causing harm. Now increasingly, it need only cause offense. The creeping erasure of robust speech needs to be brought to light repeatedly and courageously, or we will find ourselves breaking the law for reasonable criticism in a few years. In fact, it is already happening.
Wikileaks, no DNS
Could this be lights out for Intel?
OH GEESH YEAH THE END IS NEAR INTEL WILL DIE!!!
what the hell is this guy thinking? Intel is making a healthy (marginal, but still) profit and actively investing in new technologies. It's not going to go belly up just like that. AMD has a little bit of advantage that it can "dump" cheap chips into the server market as Intel needs time to ramp up production on Conroe and Woodcrest, but to call it "lights out" is just stupid.
Intel has 70% of the market for servers built on personal computer chips.
What a stupid article.
http://lkml.org/lkml/2005/8/20/95
"AMD has some very interesting products at the high end, but the new Intel chips ar more in-line with the kind od product we build."
I think what he means is that you you can have different goals and your goal determines which processor is best. If you are building a sub $1K notebook an AMD Opteron would be a pretty dumb choise. Apple seelscomputers to users who value computers that don't make fan noise and don't take up loads of space and are will ing to pay a slight preium for a quality product. right now, if that is the goal The Intel "Core 2" series is the best processor. Other things matter too, for example Intel was able to supply the enginerring labor to design the main login boards for some of Apples new systems. I don't knopw if AMD is evenin the mother board bussines. When you have a "we are betting the company on this" kind of new product you absolutly have to reduce any risk. Having Intel design the main board lets them get to market first with a new CPU and have a good chance of it working. So if you compare Intel to AMD look at the services each can supply. The company that bys the CPUs mayb e buying more then just chips, service is also a key product that end users see only indirectly
They have a very successful line of desktop and portable chips with the new core solo and duo lines.
AMD Operton - Doesn't that run on Nucular energy ?
When Intel first released the Pentium, they claimed that they where now providing a server grade processor -- including for application servers. According to Intel's own marketing, the first generation of Pentium was a sign of how good Intel would be at supporting use of their processors in application servers.
They also claimed a floating point module that was ten times faster than the 486DX floating point. The only problem was that the floating point module got the answer wrong when there was a series of 1 bits in the mantissa and IBM did a study of the problem which resulted in a report that the bug would come up frequently in business applications. Intel's response lied and claimed that IBM was talking about calculating sending satalites to Pluto and back. They went on to point out that the bug can be work-around in software, but Intel did not point out that when using such work-arounds the performance of the floating point was no where close to ten times faster than the 486DX. The IBM's POWER and Sun SPARC processors did not contain any such floating point bug and accomplished floating point calucations at the speed the processor was marketted at performing. The only problematic servers where the ones with Intel Inside and Intel was not taking responsiblity (until much later when they got around to it).
Later was the F00F bug. Intel claimed it was a non-issue since the code does not "normally exist" and even when it does come up it just requires restarting the computer. So, when someone tried this out on our Citrix server, we "just" had go over to the operations room and restart the server.
Intel has made the point that no processor manufactor is perfect. On this point I agree. But Intel has a history of making marketting claims they will not take responsiblity for honoring. The bottom line is that Intel does not provide the type of support and responsiblity that is expected of a company providing a critical item for reliability/performance of a server. Those that don't learn from history are forced to repeat it and companies have learned that trusting Intel for servers is a huge mistake.
Only trust Intel on machine your willing to "just" have to restart due to known problems.
AMD has ~26% of the server chip market. That means that everybody else (primarily Intel and some IBM/Sun) has ~74%. Let's say for sake of argument that Intel has 50% of the server chip market (didn't check the number, just as an illustration) - that means that for every AMD chip sold, two of Intel's are sold. This isn't a 'lights out for Intel' scenario - it's a scenario in which real competition is going on, and Intel can no longer just rest on its sizeable laurels.
antipaucity
I find it hard to believe Intel could fail in the long run. I have been using Intel processors since the 8048. I do think they have some problems though. I bought two Dell SC1420 systems containing Xeon 2.8GHz 800 MHz FSB processors and no one can tell me exactly what processor chip is in the machines. The CPUID utility shows the family and stepping of the chip, but the CPU Finder web page has no record of the stepping of my processors. Intel has so many different CPU chips called Xeon that they cannot keep up with them. I supplied Intel with the output of the CPUID utility and asked them to identify the chip. They could not, and responded that I should remove the heat sink and thermal tranfer material and read the sSpec number off the chip. This would void my Warranty with Dell and is not an acceptable answer. Intel has told me it's up to Dell whether they choose to "disclose" which chip is in their product. Dell has not chosen to disclose this information to me and I seem to be in limbo. I am very dissapointed in Intel's position on this and I find it non-professional to the extreme. I have grave doubts about Intel now. Also I downloaded the evaluation of their c++ compiler and the readme indicates that their compiler only operates if Microsoft's Visual studio is loaded underneath. This means you have to buy Microsoft's c++ first and then Intel's second to use Intel's compiler. After reading Intel's compiler READ.ME file, there is no way I would touch their compiler with a ten foot pole. The limitations are staggering. After the READ.ME, I went back to cygwin and Visual Studio for Windows based development. Douglas W. Goodall Systems Software Engineer