RSSIntel Seeks to Upstage AMD 'Barcelona' with Quad-Core 'Tigerton'
By Scott M. Fulton, III | Published September 5, 2007, 5:47 PM
The fact that AMD plans to unveil its quad-core Opteron server processor on Monday is the worst kept secret since rumors yesterday that Apple CEO Steve Jobs would unveil something having to do with iPods today. What has come as a surprise is that Intel decided not to wait to let AMD have its day before releasing some initial specifications for its forthcoming "Tigerton" class Xeon MP quad-core server processors.
If you believe the performance test numbers professed by CPUs' own manufacturers, then Intel definitely plans to make a wrestling match out of Tigerton vs. x4 Opteron. This afternoon, Intel stated that an IBM System x3850 M2 4-way server produced a SPEC_int_rate_base2006 benchmark rating of 184. To give you an idea of relative standing, the best 4-way Intel quad-core servers today tend to just break the 90 barrier.
An AMD Opteron 8222 SE dual-core system today, delivering the same number of total cores but just in more chips, built by AMD itself, was tested last month and produced a 108 on the same benchmark.
On Monday, AMD will promise that Barcelona will produce 48% better performance results for 4-way systems over its existing dual-core product line, citing this figure for reference. It takes the simplest of calculators to determine that, if AMD keeps its promise, that quad-core Barcelona system will only score a 160.
If there's a difference to be made up, it will be in terms of frequency and power. As we already know, AMD plans to introduce its lower frequency parts first. The IBM system in Intel's test uses four quad-core Xeon X7350 processors clocked at 2.93 GHz, according to the company, which implies that the first servers based on these models - IBM's included - will include the nearly 3 GHz speed. Quad-core Opterons at that speed won't be introduced until sometime next year.
However, if AMD's estimates are based on the slower parts it plans to introduce first, then it could yet make up that performance ground and leap-frog over Intel...even if it's late to the table in doing so.
There will be six models in the 7300 series, including the 2.93 GHz performance model at the top of the line. Intel states its power rate is 130 watts, presumably TDP (a relative rating of how much power it should take to keep the chip cool). On the low-power end of the scale is a 1.86 GHz processor at around 50 watts. In-between will be four models that hover around 80 W TDP.
For AMD to stay competitive, it will need to make a much stronger case for its lower-power value proposition. Last year, the company argued that Intel was using either unfair or unrealistic (or both) metrics in determining power consumption numbers for its CPUs. Now that Intel's numbers appear to be lower than AMD's have been, AMD will need to avoid the appearance of "parsing" as it argues it still has the power/performance/price lead.
Intel's choice of price range might be of some help to AMD, as its announcement of $856 on the low end to $2,301 on the high end, doesn't exactly break through any pricing floors.
In today's announcement, Intel painted a picture of IT shops having completely transitioned to Core Microarchitecture-based CPUs, now having the ability to pool all that processor power together into next-generation virtualized environments.
"With the introduction of the Xeon 7300, users will now be able to pool all of their Intel Core microarchitecture based server resources, whether they are single-, dual- or multi-processor based," the company stated, "into a dynamic virtual server infrastructure that allows live virtual machine migration that can improve usage models like failover, load balancing, disaster recovery, or server maintenance."
Perhaps caught just a little bit off-guard today, AMD found itself partly lifting the curtain on its Barcelona launch as it attempted to argue away Intel's new assertions like a political campaigner on the defensive. "'Tigerton' has the unfortunate distinction of being near last in a line of a dying architecture based on a Front Side Bus bottleneck," reads the first part of an AMD press release sent to BetaNews just on the heels of Intel's announcement.
The challenger's case is starting to sound familiar: Intel has an extra part, AMD points out: a memory controller that resides off of the die and thus consumes more power. What's more, AMD continues, that extra part is relied upon to tie together one dual-core component and another dual-core component, pairing them together like a double-decker sandwich, but rendering those parts captive to a separate device that becomes a bottleneck in high-performance conditions.
"Tigerton is still a dual-core processor design, just as 'Penryn' will be," the AMD statement goes on. "Intel won't offer a quad-core processor design until late 2008, more than a year after AMD. To achieve full performance scaling on real world multi-threaded workloads, real design work is needed. Packaging dual-cores together into quad-cores is insufficient, as clearly Intel itself understands. Why else transition to native quad-core in late 2008?"
By showing its hand earlier than AMD, Intel has forced its competitor to make its arguments in the alternative five days earlier than it had anticipated. In so doing, it puts the onus on AMD, come Monday, to come forth with less talk and more proof, rather than merely repeat its arguments of the previous Wednesday afternoon.
If AMD continues now as it had planned, it will disappoint both prospective system builders and investors; now that company will need to make an extremely convincing demonstration, reassuring everyone that Opteron is still in the ball game.
I hope nobody thought my comment was "Pro-Intel" wasn't trying to come off that postive towards Intel. I just have to question did AMD really offer a product that was that much better, or was the market just ready for additional chip manufactor at the time. Going back to the days before AMD can anyone really say there was anyone but IBM and Intel that was able to stay in the market at that time?
I think the changes to the computer industry during that time, where speed and performance was starting to come become what customers wanted, instead of just having a computer.
At least when I remember AMD being on the market, I wanted a "better" computer at the time that would have been my second computer. The first family computer was our desire to learn and have this new device ( to many households it was exactly this ) and I was just happy to have one ( I did learn use computers while at school ) but never for prolong periods of time.
While I wish AMD the best to continue to be a leader in the market, I have my fears that they perhaps are another 3dfx who just couldn't make it after some very good products they released.
There is history with Intel and IBM and Apple that AMD just simply cannot go up against. They are all leaders for many reasons, but what works best for them, is they have been there from the start.
Unless AMD can change the computing industry like those companies did one way or another, I fear AMD just cannot continue being the one step behind Intel.
Score: 0
|To be fair, AMD did change the industry, quite a bit with the introduction of X86-64. That instruction set forced Intel to go back on it's word of never putting in 64bit intrucstions into their own x86 processors. Now Intel in trumpetting EM64T.
The K8, and the Opteron processors totaly changed things. The Itaniums were made to look like hot toys, and the power of X86 was proven again. They changed the game. People started to realised that Mhz wasn't the only important thing when it came to performance, and that an architecture that they knew could be super powerfull, without need a lot of energy(Relativly speaking of course).
AMD has also changed chipset, and connection methods on motherboards... Both with their work with the HypterTransport Consordium(As a matter of fact, IBM, and Apple are both founding members too), and with the intergrated memory controller. Two design aspects that Intel will be using very soon into their own systems.
I think AMD is going to be around for a while... With their ATI buy-out, they've been pushing more and more into the mobile market as well. They are going after Intel on all fronts. Intel is facing anti-trust, and monopoly charges in both the US, and in Europe.
AMD's current roadmap is ambitious, and if they can pull it off, Bulldozer may do the same thing to Intel as the K8 Hammer series chips did. But like all things, all we can do is wait and see.
I like AMD as a company, and I really don't like the idea of going back to the dark days of: "Well, I need to build another computer... What are my choice. Let's see, I can either by Computer A with the Intel chip, or Computer with the same Intel chip. Both will cost me a huge amount of money, for no good reason."
I'd also would like to point out that AMD's Server parts are keeping neck to neck with Intel's current line-up, staying out of the Quad-Cores. Granted, their desktop line-up needs an updated, but for the money, they are still super competative. K8 beat Intel's best for a solid three years, more depending on how you look at it. AMD is one year behind... Unlike Intel they don't have a HUGE cash pile to sit on, to figure out their next move. But let's cut AMD a little bit of slack, eh? Let's see how K10 does. I think Barcelona will perform decently, probebly about as well as Intel's current quads. K10 will really shine with Shanghai, and Bulldozer, and then we'll see a good race on our hands betwen intel's Penryl, and the new K10s, going off to the Octo-Core CPUs, and so on and so forth.
Score: 0
|*sigh* the codenames are getting dumber with each new iteration. What next? Feisty Fawn? Slobberlight? Can't they just go with 2007.01.whatever? We can remember version numbers as well as stupid names. Maybe better.
Score: 0
|Lol Feisty Fawn has already been used by Linux.
Score: 0
|I totally agree-- both AMD & Mozilla are total morons about that.
Ubuntu, well-- at least they also post version number along with name at times.
Score: 0
|*bangs head on table* Well don't ya just reckon that he might have known that lmao. Point of his post!
Score: 0
|Clearly Intel has some good leadership that understands how both the market will change and how to keep expanding Intel based computers ownership numbers slowly by providing new products on a good roating cycle.
Granted this process does cost Intel a good deal of money ( I have no doubt they have put in millions into research ) both in making these new products and research into what customers actually want.
This is of course why Intel has been around for all these years, granted there was a time where AMD was a risk but perhaps that was more because of the industry change overall then the product AMD actually offered.
I am still waiting for somebody to offer the next generation architecture that will both adapted very fast and withstand the test of time like the x86 has.
* When I speak of x86 I am speaking clearly of 32-bit cusomer processors. IBM's Power PC clearly passes the test of time in that its still a valuable architecture to work with.
** x64 is to young to pass the test of time, plus the fact x64 cpu(s) are also x86 processor(s) which this engineer considers a major flaw and one of the main reasons it will never be accepted by itself.
*** The next major thing must stand on its own, be supported by the largest market share, and be able to adapt to future changes in the market.
****X86 is a perfect example of this being
able support both 32-bit and 64-bit programs ( depending on the Operating System of course ).
Score: 0
|Intel only over the last few years has been doing such a good job, but it took them about 6 years to do just that. Amd was beating them hands down in every single area and it wasn't until they started to reach around 23 percent market share that Intel finally said we have to do something. The sad thing is Amd fought for every single percentage of market they received and as soon as c2d came out they won that market share back almost over night.
Amd is great, and they are pretty amazing considering they are 1/10th the size of Intel with total revenue equal to Intel's profit.
Let's hope Amd doesn't go anywhere, they gave Intel the reason to innovate like they have.
Now if we could just get a Amd type of operating systems we would be ok.
Score: 0
|Intel has held a lead in manufacturing technology for the past decade, add to that a massive dominance in the mobile market and they have been able to keep margins good for quite some time.
K8 was a great product, and AMD should be proud that for almost 3 years they could claim superiority hands down in server and desktop... but a 6 year window is exaggeration.
AMD has only recently found any success in mobile with the Turion X2's, and I think most people are crossing their fingers that the initial sentiment on Barcelona is false and that AMD will remain competitive for years to come.
Score: 0
|Not really. AMD has been in the lead pretty much up until the point that the Core 2 Duo was released. Ever since then they have not been quite able to catch up.
Score: 0
|I concur. It seemed like the whole 32/64 bit hybrid thing caught Intel by surprise while they were try to push their totally new 64 bit architecture. I know they still sell them, but is the Itanium dead?
Aside from the "on core" type of performance metrics that Intel likes and the I/O metrics that AMD likes, I'd like to know how they compare in terms of their virtualization technologies.
Score: 0
|