backtop


Print 82 comment(s) - last by ElJefe69.. on Feb 15 at 11:24 PM

AMD's native quad-core design picks up more steam

In a series of memos forwarded to DailyTech, industry insiders discussed the upcoming launch frequencies regarding AMD's next-generation architecture, previously dubbed K10 by AMD Executive Vice President Henri Richard.  Since then, AMD has generally referred to the next-generation chips as the Barcelona family, although Barcelona specifically denotes the high-performance quad-core server processor codename.

Some details of next-generation AMD desktop processors, the Stars family, were revealed late last year.

The desktop equivalent of Barcelona, codenamed Agena, is the 65nm flagship of AMD's next-generation desktop processors.  Launch frequencies were quoted at "2.4 - 2.6GHz." Previous roadmaps had indicated Agena would debut at 2.7 to 2.9 GHz.  Agena will have a 2MB L2 and 2MB L3 cache per CPU.  AMD's internal guidance denotes this as a 125W TDP processor. As the flagship, Agena will be the first next-generation desktop launch and is scheduled for Q3'07.

Kuma, the dual-core mainstream next-generation desktop processor was quoted as having launch frequencies of "2.0 - 2.9GHz."  Unlike the quad-core Agena processors, Kuma will feature 1MB of L2 and 2MB of shared L3 cache. Kuma will launch with both 89W and 65W TDP variants, but Energy Efficient models scheduled for 35W TDP will follow shortly after.

Rana, the next-generation Sempron successor codename, will launch with frequencies in the 2.1 to 2.3 GHz range.  The dual-core CPUs will feature 1MB of total L2 cache, but no L3 cache.  Rana's TDP is rated at 65W.  Rana will not launch with the Agena flagship; AMD roadmaps have the processor launching at the same time as the Energy Efficient Kuma processors, or approximately Q4'07 if the launch schedule holds together.

As previously reported on DailyTech, Stars processors will use AM2+ motherboards.  These processors can plug into existing AM2 motherboards today given the proper BIOS updates, but without the AM2+ sockets Stars processors will drop down to the HyperTransport 1.0 bus speeds.

AMD's Agena FX codename also appears to still exist on the roadmap.  The only difference at this point between Agena and Agena FX is that Agena FX will use the Socket 1207+ interface.


Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Not so obsolete
By Omega215D on 1/26/2007 10:00:36 AM , Rating: 2
People were so quick to jump on the "early death of socket AM2" bandwagon and it looks like that won't be the case: As previously reported on DailyTech, Stars processors will use AM2+ motherboards. These processors can plug into existing AM2 motherboards today given the proper BIOS updates, but without the AM2+ sockets Stars processors will drop down to the HyperTransport 1.0 bus speeds.

Now when these are available I trust Anandtech to post a review on whether dropping down to HT 1.0 will have any noticeable effect on app usage. This will be most interesting in terms of backward /forward compatibility.




RE: Not so obsolete
By SacredFist on 1/26/2007 10:39:10 AM , Rating: 2
True ... but AM2 seems obsolete to get because you kinda bought a not ... okay, better said, its like buyers remorse


RE: Not so obsolete
By The Arete on 1/26/2007 10:41:19 AM , Rating: 2
I'm curious as to the major differences between the two platforms. I had assumed, like many, that with the new architecture, my AM2 motherboard would become completely obsolete.

I wonder as to the specs of the new AM2+ boards, and whether it might behoove me, and others, to upgrade regardless.

Anyone know any major upgrades, excluding BIOS, that are going to come about with the new Architecture?


RE: Not so obsolete
By KristopherKubicki (blog) on 1/26/2007 10:43:55 AM , Rating: 2
The only thing that's really been announced is HT3 support. Of course, if HT3 really doesn't make much difference over HT 1.0, there may be no reason to upgrade the motherboard.


RE: Not so obsolete
By Tsuwamono on 1/26/07, Rating: -1
RE: Not so obsolete
By KristopherKubicki (blog) on 1/26/2007 12:38:40 PM , Rating: 3
You cannot upgrade your motherboard from AM2 to AM2+ with just a BIOS upgrade. The BIOS upgrade only adds support for the processor, you will not get features like HT3 from BIOS.


RE: Not so obsolete
By Tsuwamono on 1/26/2007 6:00:29 PM , Rating: 1
oh, i miss read the post. sorry


RE: Not so obsolete
By dandres87 on 1/26/2007 12:52:29 PM , Rating: 2
Remember how Intel released the Pentium D, it was double core, not true dual core. And remember how much the Athlon X2s stomped them? I think its going to happen all over. Now the strength of the C2D will compensate a bit in the QX's for being a double core(Quad core platforms that is) but when AMD's true quad core comes out it will assuredly give Intel a run if not annihilate. Then you have the price wars (as with the dual cores ) all over again.


RE: Not so obsolete
By Adonlude on 1/26/07, Rating: 0
RE: Not so obsolete
By fxyefx on 1/26/2007 1:28:30 PM , Rating: 2
You are right, but I would not be so quick to say that AMD will fall back to business mediocrity so soon, especially in light of these recent news articles. : D It is very difficult to predict what the future will hold, especially with the switch to "fusion" between GPUs and CPUs, and the possibility of nvidia becoming competitive in the same market as AMD and Intel.


RE: Not so obsolete
By The Max on 1/26/2007 3:22:32 PM , Rating: 1
I Was under the impression Motorola invented the microprocessor , 8088a , the size of a house brick but never the less the 8086 process of logic gates progressed from a calculator chip to cpu which IBM and others developed , i guess AMD and INTEL must have invented the wheel too , similar in fact to Xerox and Apple were not the ones to blame for initiating icons , xerox in 72 and apple later , even my memory seems to be on the blink . Weird !


RE: Not so obsolete
By Hoser McMoose on 1/26/2007 4:05:35 PM , Rating: 3
The Intel 4004 processor is generally recognized as the first single-chip microprocessor. Before that there were other processor designs, but they were based on several independent chips working together. Bringing all the processor functionality into a single chip is really what allowed for the development of microcomputers and eventually the desktops and laptops we know today.

It perhaps should be noted though that Intel was really just the first to reach a fairly obvious finish line. TI and Motorola (and maybe others?) were also working towards the same goal and had their own microprocessors out shortly after Intel released the 4004.


RE: Not so obsolete
By Tom Tom on 1/26/2007 6:07:41 PM , Rating: 2
I believe Fairchild Semi invented the first microprocessor.


RE: Not so obsolete
By Viditor on 1/26/2007 8:10:04 PM , Rating: 4
quote:
I believe Fairchild Semi invented the first microprocessor

Correct...
Also, both Intel and AMD were founded by people from Fairchild Semi...


RE: Not so obsolete
By Phynaz on 1/26/2007 1:45:28 PM , Rating: 2
Ponder this:

Since current AMD chips use seperate caches, they would fall into the double-core catagory as you have defined it, and not into the true dual-core catagory.


RE: Not so obsolete
By fxyefx on 1/26/2007 2:30:24 PM , Rating: 2
I don't think dandres87 mentioned anything about cache sharing, nor did he define his idea of double-core versus dual -core...


RE: Not so obsolete
By ADDAvenger on 1/26/2007 7:58:04 PM , Rating: 2
Yup, and on top of that everything but the Sempron replacements will have shared L3.


RE: Not so obsolete
By coldpower27 on 1/31/2007 9:02:39 PM , Rating: 2
There will be low end K8L Dual Core which lack the 2MB of LV3 Shared cache. They will probably go against the Pentium E2xxx Series to the Core 2 Duo E4xxx sequence.


RE: Not so obsolete
By Viditor on 1/26/2007 8:18:37 PM , Rating: 4
quote:
Since current AMD chips use seperate caches, they would fall into the double-core catagory as you have defined it, and not into the true dual-core catagory

Not quite...the AMD caches are directly connected to each other and don't require a FSB to negotiate through.

In other words, the MCM (Intel's dual-dual) is operationally identical to having 2 dual core sockets, though it has the advantage of being only one which keeps software licensing cheaper.
AMD's quad core will be 4 cores directly connected to one another on the die...this should improve trace speeds to be equal with Intel's shared cache.


RE: Not so obsolete
By coldpower27 on 1/31/2007 9:36:01 PM , Rating: 2
I simply disagree Viditor, I thought you learned your lesson before, the current definition for a "true" Dual Core processor is that all the cores are on a Single Socket. This has been established already by the Server environment. The monolithic vs MCM nature has no effect of whether it is true or not. It is either a MCM or a Native implementation but it remains all the others words you have described are only used by AMD fanboys to incite flamebait in others, and are completely redundant not to mention waste alot of other peoples time.

AMD is touting their Native Dual Core advantage, however we will actually have to see if it's an advantage or not and see how the scaling is from AMD's Altair Quad Core to the Antares Dual Core at the same clock frequency in multi-threaded applications. This would have to be compared to a Intel Kentsfield and Conroe each of the same clock frequency.

There is currently no data to indicate that scaling is any worse on Intel's current implementation with MCM vs AMD's implementation with Native. At this point this is only a theoretical advantage, nothing more.

The other issue at this point is that there is a rumor belief, incited mainly by AMD fans, that Kentsfield is bandwidth limited, AND that this factor is affecting performance in a significant detrimental way. If that was the case there should be significant performance gain when the FSB is increased holding the clock speed constant as that provides more bandwidth assuming the Memory used is Dual Channel DDR2-800 in this case, to the processors but the data available shows that the processors don't gain very much if any due to the increase bandwidth which makes sense, as Core Architecture is designed so that bandwidth is not a significant requirement for performance.


RE: Not so obsolete
By Super XP on 2/4/2007 6:38:41 AM , Rating: 2
I simply will have to agree w/ Viditor.

One would expect a large company (Intel) to release a True Quad-Core, and the smaller company (AMD) to desperately release 2 of there 2x Dual Core’s glued together onto a package. RIGHT?

Who cares who or what defined what a Quad-Core? What? You expect them to tell you what a Quad-Core is? LOL,

We all should be smart enough to see what is really going on here. AMD taking longer with there “Native” Quad-Core, because it’s been built completely from the ground up. Intel’s approach is not built from the ground up, though there Core 2 was.

Just look at the CPU micro-architecture, AMD has 4 completely separate chips onto one single die, where as Intel’s approach is taking 2 Dual Core’s & gluing them together onto a package. I call them Double Dual Core, because that is what it is if you are looking for a definition.

We all know what these Quad-Core’s are all about. AMD’s approach takes longer because it is the right way to do it. IMO




RE: Not so obsolete
By SexyK on 1/26/07, Rating: 0
RE: Not so obsolete
By IntelUser2000 on 1/26/2007 7:59:20 PM , Rating: 2
quote:
I'm sorta getting tired of this "native" vs. "glued" talk. Until someone brings some facts to the table showing that "glued" solutions perform worse than "native" solutions in the real world, it would be nice if people stopped touting AMD's "native" quad cores as inherently superior for some reason. Just more FUD in my opinion, since several Kentsfield reviews showed that the "glued" nature of the chip had no impact on performance whatsoever.


Now I am not saying you are wrong, but the difference is that people in the forums are really PC users and PC enthusiast sites like Anandtech and DT benchmark for PC. On the workstation/server environments, going "native" dual core has a benefit. It presents as a single load to the FSB rather than two load. I am not exactly sure of the benefits but things become complicated like cache coherency issues and such.

Remember PAE?? Page Address Extensions for 32-bit Xeons so they can address more than 4GB of memory?? Well for the same capacity, it slowed down, but in servers, capacity is also extremely important. Servers have different issues than desktops.

Conclusion: K8L will excel for workstation/servers. For desktops it might as well not be faster per clock than Conroe.


RE: Not so obsolete
By PlasmaBomb on 1/28/2007 10:10:10 AM , Rating: 2
You are incorrect, both Smithfield and Presler are two discrete pieces of silicon in a single package. Therefore they are both "glued" together. The reason Presler chips are better than Smithfield ones is because Smithfield is a 90nm part and Presler is a more advanced 65nm part which runs cooler and has more cache.


RE: Not so obsolete
By coldpower27 on 1/31/2007 8:59:33 PM , Rating: 2
Your the one who's in error, if you actually seen a Smithfield die shot, you would see that it's a single slab of silicon vs the MCM nature of Presler.


RE: Not so obsolete
By coldpower27 on 1/31/2007 9:43:43 PM , Rating: 2
A monolithic implementation is only worthwhile if the cores can communicate on the die itself without having to leave it. The Pentium D Smithfield while being a monolithic design did not have this ability.

There is also no point on discussing Pentium D at this point in time we don't know what affect a Native Dual Core Pentium D would perform over a Pentium D Dual Core without Intercore communication such as Smithfield and Presler. The only reason the Smithfield was monolithic was to get it out ASAP, if they had more time Smithfield would be MCM as without intercore communication on die the monolithic implementation is worthless.

The same thing will happen with Core Architecture, we will never find out if by how much faster (if any) a Native implementation would be in comparison to the MCM utilized with Kentsfield.


RE: Not so obsolete
By Thorburn on 1/26/2007 9:40:14 PM , Rating: 1
Pentium D got stomped by Athlon X2 because Athlon 64 stomped on Pentium 4.

If you look at Kentsfield scaling in some applications such as 3D rendering (Cinebench, 3D Studio) it can approach the 80-90% range or even better.

The two dice approach for the majority of applications isn't a major stumbling point, a single die brings benefits definately but its not the be all and end all.

Will be interesting how it pairs up against 3GHz+ Penryns and Yorkfields.


RE: Not so obsolete
By PlasmaBomb on 1/28/2007 10:17:38 AM , Rating: 2
It certainly will since Intels' current flagship quad cores (clovertown and kentsfield) run at 120-130W, and the move to 45nm should improve this situation whilst AMD's new quad core will run at 125W, although AMD will hopefully improve this as they get to grips with 65nm (and then there is the difference in how TDP is defined).


RE: Not so obsolete
By Thorburn on 1/28/2007 11:03:04 AM , Rating: 2
I believe the top quad core 45nm parts are supposed to be 95w, with the dual cores being 50w max.


RE: Not so obsolete
By coldpower27 on 1/31/2007 9:09:47 PM , Rating: 2
There are too many variables with that comparison to determine that the MCM nature of the Pentium D was what did them in with regards to Dual Core.

The Pentium D was a true Dual Core, thank you AMD for spawning this subjective crap about "true" Dual Core.

The only definition required for a "true" Dual Core is that all the Cores fit on a Single Socket.

There is currently not enough information to determine if AMD's K8L Quad Core's will win decisively across the board, as we have no access to published benchmarks of any sort, only subjective words that give vary vague clues as to performance.

Core Architecture is vastly different from NetBurst, so the effects of the MCM issue aren't going to produce the same effects. Were also talking about Quad Cores, so we don't have any data at the moment as to how MCM or Native affects scaling for the time being. As we don't have a Native Quad Core Core 2 Quad to make the comparison with.


Early Death Syndrom
By Misty Dingos on 1/26/2007 10:31:37 AM , Rating: 2
Or EDS is a symptom of a larger problem for a company. It happens when the press or the gray press (blogs and others) tout the benefits of products before they are in the market place. This leads to over amplification of the predicted products impact in the market. In this case AMD is trying to shore up support for a product that is only in the preproduction stage. Unless the product, when released, is better than the hype then the product is great danger of EDS.




RE: Early Death Syndrom
By rippleyaliens on 1/26/2007 10:59:28 AM , Rating: 1
I guess it looks good, "on paper" so far. The specs are decent. BUT, talk about a marketing push, WELL before there are even demos' avaliable. If i recal, and someone correct me if i am wrong, because i am old, lol. But i didnt really hear much about the Core 2, until it hit at a trade show. Amd is leaking specs 9 months early, i guess in hopes of making customers hold off on purchasing intels, to wait for new Amd products.

Here is the scary part. I predict that once AMD rolls out their new quad cores, and dual cores, that Intel will release in FORCE their next level of Core 2 products. I just have that feeling, that Intel is waiting for amd to release their next versinos, just to rain on AMD's parade.


RE: Early Death Syndrom
By Griswold on 1/26/2007 12:03:04 PM , Rating: 2
quote:
But i didnt really hear much about the Core 2, until it hit at a trade show. Amd is leaking specs 9 months early, i guess in hopes of making customers hold off on purchasing intels, to wait for new Amd products.


Intel was touting it some 6 months in advance with benchmarks comparing it to AMDs products. That was one year ago.


RE: Early Death Syndrom
By hubajube on 1/26/2007 12:39:29 PM , Rating: 2
quote:
Intel was touting it some 6 months in advance with benchmarks comparing it to AMDs products. That was one year ago.
This proves that people's memories are indeed short term. There was a TON of hype well before launch and there was a TON of posts right here on Dailytech about C2D.


RE: Early Death Syndrom
By deeznuts on 1/26/2007 2:27:04 PM , Rating: 3
You just proved your memory is indeed short-term. There was no hype about conroe until Intel demoed actual processors vs. AMD ones, with review sites doing the demos (Intel provided the machines). What he said was partially correct. AMD is not demo-ing (is that a word lol) any processors, they are just talking.


RE: Early Death Syndrom
By Viditor on 1/26/2007 10:44:53 PM , Rating: 2
quote:
There was no hype about conroe until Intel demoed actual processors vs. AMD ones

Huh???
The very first showing of Conroe was at IDF in March of 06, but independent benchmarks weren't released until May 06. If you look at the Feb article here (pulled at random)
http://news.com.com/Intel+strikes+back+with+next-g...
Note that Intel is projecting that C2D will be 20% faster than AMD's comparable products...5 months before launch.

Barcelona is expected near the end of next quarter (Q2), which would make this...what do you know, 5 months before launch!


RE: Early Death Syndrom
By Viditor on 1/26/2007 10:54:20 PM , Rating: 2
BTW...if you think about it, it would be a very typical (and smart) move for AMD to give the first Barcelona demo across the street from the Spring IDF this year...
If the numbers are there, they will certainly steal a good deal of Intel's thunder that way.


RE: Early Death Syndrom
By coldpower27 on 1/31/2007 9:55:34 PM , Rating: 2
The nice thing with Intel is that they published benchmarks, they were done in house, however but still were benchmarks at least. In AMD's case all we have is AMD's word, and were still blind as to where the performance gains are to be had.

Well there isn't very much Intel can do to stop Barcelona in Q2, as their 45nm process is not ready yet, but since Barcelona is Quad Core and AMD is capacity constrained, lucky for Intel it won't have much if any financial impact till Q4 according to AMD. This makes sense as Barcelona die size was quoted at around 288mm2 from what I recall.

Basically for 2007 Intel will be concentrating it's efforts on being the economical solution, and using the price hammer against AMD rather then concentrating on pure performance, though as it stands Intel will be no slouch in performance.

There is more then one way to compete, having the performance crown is only 1 method.


FIY
By mlittl3 on 1/26/2007 12:03:18 PM , Rating: 1
For those commenting about the 125W TDP, remember that Intel's 2.66 GHz quad-core xeon has a TDP of 120W. This is just typical power dissipation and might actually be higher. Also, throw in the fact that you need extra power for the memory controller in the northbridge and FB-DIMMs (this is a Xeon platform where regular DDR2 is not supported) and the power will be even higher.

If the Barcelona 125W TDP is for the 2.7-2.9 GHz range and AMD is correct about the 40% performance number, then expect AMD to blow Intel out of the water with regards to total system performance/watt. If the Barcelona 125W TDP is for the 2.4-2.6 range, then both companies will have about equal total system performance/watt.

One more thing, Intel annouced the Core 2 Duo architecture, almost 6 months before launch with performance numbers. AMD will release K8L processors from Q2 '07 to Q3 '07 which is extactly the same time frame as the Intel lead on its next generation products. Let's not forget that.

And I will end with a question? I was doing some research and it seems to me that Intel has nothing new architecture wise coming out all year except a move to 45 nm near the end of '07 and maybe a native quad core design. Some people mentioned Tigerton and Peryrn but these are just MP processors and laptop 45 nm cheaps, respectively. Intel's roadmaps just have an increase in bus speed from 1066 MHz to 1333 MHz for desktop chips. What are these new architecture chips that every things Intel is releasing at the last minute? I can't find any info on them. Last time I checked Intel had the following strategy:

New Architecture (65 nm, Core 2 Duo/Quad July 06 to July 07)
Die shrink of this architecture (45 nm Core 2 Duo/Quad July 07 to July 08)
New Architecture (45 nm, Core ??? July 08 to July 09)
Die shrink of this architecture (32 nm, Core ??? July 09 to July 10)
etc.

No new architecture changes until July 08 or maybe a little sooner. Please, please correct me if I'm wrong with Intel codenames for future chips and wikipedia or other links if you're really nice.




RE: FIY
By cheburashka on 1/26/2007 3:14:17 PM , Rating: 2
People keep forgetting that Penryn is not just a shrink to 45nm but also adds SSE4 and HyperThreading.


RE: FIY
By JackPack on 1/26/2007 3:25:43 PM , Rating: 2
Plus high-k dielectrics and metal gates.

VT2

HT2 may allow for more than 2 theards per core, i.e. 16 threads for Yorkfield.


RE: FIY
By Viditor on 1/26/2007 9:15:24 PM , Rating: 2
quote:
Plus high-k dielectrics and metal gates

Yes...but most likely NOT in the first release of Penryn.
Intel (very wisely) doesn't perform 2 major process changes in a single release. This will probably be for the pre-Nehalem refresh in mid 08.


RE: FIY
By Viditor on 1/27/2007 12:25:15 AM , Rating: 2
I stand corrected!

Intel WILL be releasing both process changes simultaneously.
This is great news for Intel...HK+MG should allow them 3-4 speed bumps over AMD instead of 1-2.
It should also do very well for them in the server market for power/performance.


RE: FIY
By Viditor on 1/27/2007 7:00:31 AM , Rating: 2
Just to add to the confusion, AMD and IBM will ALSO be using High-K metal gates in mid 08 on their 45nm process...

http://biz.yahoo.com/bw/070127/20070126005722.html...


RE: FIY
By Viditor on 1/26/2007 9:08:41 PM , Rating: 2
Barcelona will be adding the new extensions as well, but remember that they only help when written for. It will be a year after release before SSE4 really matters much for anything.


Lots of Cache
By Phynaz on 1/26/2007 12:07:42 PM , Rating: 4
So now that AMD is using more cache than Intel, what happens to all the people derided Intel for having large caches?





RE: Lots of Cache
By Griswold on 1/26/07, Rating: -1
RE: Lots of Cache
By JackPack on 1/26/2007 2:22:36 PM , Rating: 2
What are you talking about?

AMD reduced the L2 cache to 512 KB per core and is using a slow 2 MB L3.


RE: Lots of Cache
By fxyefx on 1/26/2007 2:33:27 PM , Rating: 2
One will be able to make the same criticism of Intel when they come out with an integrated memory controller.


RE: Lots of Cache
By Chillin1248 on 1/26/2007 10:10:16 PM , Rating: 3
Actually...

quote:
The Intel 386SL microprocessor integrates a fully static processor, memory controller , ISA bus controller, EMS 4.0 hardware and system control circuitry for battery-powered systems. It offers three times the integration of Intel 386SX CPUs.


It was released in the early 1990s, oh the irony.

--------
Chillin


RE: Lots of Cache
By bohhad on 1/27/2007 12:04:51 AM , Rating: 1
thanks for that extremely relevant piece of information


RE: Lots of Cache
By Hoser McMoose on 1/26/2007 4:17:51 PM , Rating: 2
Umm.. AMD's quad-core chips have 4 x 512KB of independent L2 cache and 2MB of shared L3 cache for a total of 4MB. Intel's current Core 2 Quad and quad-core Xeon chips have 8MB of L2 cache, shared 4MB per pair of cores.

So how exactly is AMD's 4MB of cache more than Intel's 8MB?

Personally though I see nothing wrong with a large cache, never derided Intel for that (though I may have derided them for having an outdated bus architecture). All that really matters to me is how a chip performs on various applications vs. how much it costs and how much power it consumes. Whether AMD or Intel achieve these goals through higher clock speeds, more cache, integrated memory controllers or whatever is interesting to geeks like me from a sort of academic point of view, but it's not going to influence my buying decisions much.


By Phynaz on 1/26/2007 12:43:18 PM , Rating: 2
I agree. These aren't really "leaks", but an attempt to create FUD by AMD. They are hoping to freeze the market until these chips are out.


By Viditor on 1/26/2007 9:05:31 PM , Rating: 3
quote:
It seems as if the "spec-leaking" of the next-gen AMD CPUs is just a marketing echo of the "spec-leaking" on the R600 series of AMD-ATi GPUs

You forgot the core2duo...Intel began that spec-leaking in Dec 05, and didn't show one until the Spring IDF (March 06).
They didn't submit them for benchmarking until April/May 06...

Barcelonas are currently in volume production for a release next quarter, so I don't really see this as a bid purely for "hype" (though that certainly is part of it).

quote:
I then sold AMD short and have made a large chunk more money

If you sold AMD short when the buyout happened, then you must have been sweating it out for some time now!
The stock went up from a mid-day low of ~$16.90 to a high of ~$27.90 in September.

quote:
Why AMD made a bid for ATi, (knowing at the time that Intel was coming out with a powerful reply to the X2 series) instead of totally focussing their management and financial concentration on both CPU design and process development is beyond my feeble mind to grasp.... It is not as if AMD was lacking first-tier chip-set and motherboard partners....


A very common question and one that takes many things into account.

1. AMD learned the hard way that in order to secure a launch of a CPU, they MUST develop their own chipset (remember when Via burned them at the original Athlon launch?). It costs AMD more to develop a chipset than it does to develop a CPU because they have to hire a contract team to do it...with ATI, they no longer need to do this and will keep the long-term costs down.

2. Since CPUs require ~5 years to develop, both AMD and Intel keep a majority of their focus on the future. As Jerry Sanders said, "The semiconductor industry is a wierd kind of russian roulette. You pull the trigger, and 3 years later find out if you've blown your brains out".
The ATI synergy is focused on 2009 products, not todays so much.

3. Notice that Intel has responded by ramping up their own graphics division, so that in 2009 they hope to be somewhat competitive...

quote:
in terms of market penetration that have now set themselves back a couple of years, maybe even longer

This is completely incorrect...
While the profit numbers of this last quarter were indeed abysmal, AMD did gain marketshare. In fact, market penetration may be the one and only upside from Q4.


By kilkennycat on 1/27/2007 2:38:32 PM , Rating: 2
Quote:-

If you sold AMD short when the buyout happened, then you must have been sweating it out for some time now! The stock went up from a mid-day low of ~$16.90 to a high of ~$27.90 in September.
-------------------
Actually I sold my ATi stock and AMD short just before the the time the ATI-AMD deal was finally consummated -- when Core2 was quantity shipping and it was glaringly obvious that a huge mistake was made. November time-frame, iirc...
--------


Quote:-

AMD learned the hard way that in order to secure a launch of a CPU, they MUST develop their own chipset (remember when Via burned them at the original Athlon launch?). It costs AMD more to develop a chipset than it does to develop a CPU because they have to hire a contract team to do it...with ATI, they no longer need to do this and will keep the long-term costs down.
------------------

I didn't know that nVidia and (the original) ATi were "contract teams" for AMD... that AMD actually paid them money to build their AMD-compatible chip-sets. Doesn't nVidia and ATi make lots of money on their AMD-compatible chipsets... at least nVidia do! Maybe AMD still pay a pittance to scrabbling third-tier suppliers like Via and SiS ? Please refer me to the ORIGINAL sources of all of this information with regard to contract design of chipsets for the current AMD64 family of processors.

And as for a chipset costing AMD more than a CPU to develop (even if they did it themselves) .. which brand of tobacco are you smoking ??


By Viditor on 1/28/2007 11:09:41 PM , Rating: 2
quote:
Actually I sold my ATi stock and AMD short just before the the time the ATI-AMD deal was finally consummated

So you watched your short position heading south by more than 10%...I am sure you were relieved when the earnings warning came! Congrats on a successful gamble.

quote:
I didn't know that nVidia and (the original) ATi were "contract teams" for AMD

I guess you also didn't know that AMD has developed 9 of their OWN chipsets, and that the majority of Opterons are still sold with AMD chipsets...
AMD has had a self-developed chipset for every model of release...

quote:
And as for a chipset costing AMD more than a CPU to develop (even if they did it themselves) .. which brand of tobacco are you smoking ??

That was actually directly from JS3 the year he retired...
He was talking about how big a strangle hold Intel really had, and said that it cost AMD much more to develop the chipsets than the processer itself.
Try doing some Googling on it...


Wow
By ajfink on 1/26/2007 11:35:31 AM , Rating: 2
It looks to me that the new Semprons will completely destroy the current lineup of even high-end X2s in performance (per clock, and they will probably overclock to a solid 2.8-2.9Ghz, further burying the X2s in the ground), as long as L3 isn't an extremely integral part of the new architecture's performance. AMD will have to steeply discount its old processors to make room for the Rana-based ones.

The Agena FX's better be fast as all Hell to still warrant a 125 watt TDP (but I guess the QX6700 isn't really that cool running either).




RE: Wow
By kenji4life on 1/26/2007 12:02:38 PM , Rating: 2
If that's true it looks like A64 all over again!! woo!


RE: Wow
By coldpower27 on 1/31/2007 10:26:04 PM , Rating: 2
The new Semprons are still going to be Single Core processors so they won't be destroy the higher end Dual Core very much.

The Rana Core is actually the replacement for the Athlon 64, just like how the Pentium E2xxx line is replacing the price points of the Pentium 4 HT's.

Yes both companies are replacing their mainstream Single Core with Dual Core processors.

It's Quad Core on the 65nm process it cannot be helped, even Intel's Kentsfield QX6700 has a TDP of 130W, not to mention the Presler XE 965 had the same TDP too, and it's only a Dual Core.

Going forward for both companies the only Single Core lines left will be the budget Sempron Spica core for AMD, and the Celeron 4xx Conroe-L for Intel. There will not be a Core 2 Solo, as Core 2 is Intel's flagship brand.


What new details?
By BasementWayne on 1/26/2007 10:06:26 AM , Rating: 2
I'm comparing this to what was said last November. I really don't see anything new. The biggest difference is the Agena (quad core) launch speed. It has dropped from "Clock frequencies of 2.7 GHz to 2.9 GHz" to "2.4 - 2.6GHz.".





RE: What new details?
By Regs on 1/26/2007 10:14:31 AM , Rating: 2
Pretty much. The only time we'll know anything new is upon the retail release.


so far away!
By lazyinjin on 1/26/2007 10:26:51 AM , Rating: 2
quote:
As the flagship, Agena will be the first next-generation desktop launch and is scheduled for Q3'07.


that sounds like forever to me(I really want to build a new PC soon...but I'm also waiting to look at R600). And I'm also really surprised at the 125W TDP, hopefully thats just overly conservative.




RE: so far away!
By Griswold on 1/26/2007 12:08:03 PM , Rating: 2
AMDs TDP figures have always been "conservative" as in, they represent the absolute maximum, which is somewhat difficult to reach for a consumer. As such, 125W for the first stepping of a premium(?) quadcore doesnt seem to be out of line.


I own an am2
By verndewd on 1/28/2007 6:05:10 PM , Rating: 2
It looks as if Am2+ could narrow the gap,recent news states that AMD is looking for 40% perf dominance over clovertown on barcelona.

This is great news ,but having purchased Am2 in july on the promise of am3 compatibility,I am feeling a bit used as a consumer.I can throw an am2+ in and sacrifice the ht3,But what will that mean in performance?I am concerned and upset at the purchase i made.

Having said that,i feel That barcelona may indeed be a great chip,and am2+ a decent platform for the agena counterpart,but it doesnt say much about am2's verility(if you will) as a product at all.As a matter of fact AM2's performance is not even bieng mentioned in the news releases.




RE: I own an am2
By Super XP on 2/4/2007 6:45:07 AM , Rating: 2
Don't feel bad, AM3 from what I've been reading is looking for a late 2008 release. I don't think we are ready for DDR3 just yet.

Also, you can always e-bay your AM2 motherboard. Though I personally think the AM2+ will be the one to get because we would be seeing DDR2 ram go down in price. They are also getting faster & faster.

Several websites say that Barcelona is 40% faster in a wide array of benchmarks. This would be awesome seeing that AMD completely molested there K8 with heavy modifications & additions. Good for them. But I do think they need something completely new in the near future.


PCI-E 2.0
By kenji4life on 1/26/2007 11:12:35 AM , Rating: 2
I'm guessing PCI-E 2.0 may show up on some AM2+ boards..

This whole thing reminds me of Socket7 and SuperSocket7...

Of course that was more AMD revamping an Intel platform, wasn't it?




By Griswold on 1/26/2007 12:09:20 PM , Rating: 2
quote:
Give it up and admit defeat.


Hehe, and here we have the fanboy from the other side of the fence. :P


By FITCamaro on 1/26/2007 12:25:10 PM , Rating: 2
While his last couple statements were fanboyish, he is largely right. AMD did overcharge at first for the X2 processors since Intel wasn't any real competition(granted it was still more than enough for the average joe). If AMD had kept their prices lower they would have sold a lot more in the beginning.

And yes C2D does vastly outperform AMD right now. Granted the prices have come down to where AMD is still a good buy. And though while AMD isn't the top performer anymore, they're still shipping more than ever before, probably mainly because they're now the cheapest in town when looking at both companies current gen products. You can get a 3800+ for about $50 cheaper than a 6300. The 6300 will outperform it, but idiots who buy Dells, Compaqs, Gateways, etc don't look at performance charts, they look at price.


By Pirks on 1/26/2007 5:55:54 PM , Rating: 1
quote:
And yes C2D does vastly outperform AMD right now. Granted the prices have come down to where AMD is still a good buy. And though while AMD isn't the top performer anymore, they're still shipping more than ever before, probably mainly because they're now the cheapest in town when looking at both companies current gen products. You can get a 3800+ for about $50 cheaper than a 6300. The 6300 will outperform it, but idiots who buy Dells, Compaqs, Gateways, etc don't look at performance charts, they look at price.
Camaro, you got it nailed, I just wanted to add that a LOT of people still run all sorts of single threaded apps and games, I think multithreaded/heavily multitasked users are still MINORITY in absolute numbers, and since 1MB cache single core A64 San Diegos are dirt cheap now ($75 on newegg, or $60 on ebay) - one can overclock them easily up to 2.8GHz and get blazing fast single threaded machine for HUNDREDS less than Intel C2D machines, considering mobo upgrade and the move from DDR to DDR2 which is a NECESSITY for Intels but optional for AMD since you can go for DDR S939 or DDR2 AM2, both have vast selection of fast and inexpensive single and dual cores, especially if you shop at ebay. so, for those looking for best bang for a buck and not running any dual core friendly apps or games AMD is still number one choice. and guess what - in absolute numbers this is MAJORITY of the whole PC market. Intel fanboys can suck it up, but AMD becames even more and more popular now as its prices go down under Intel's pressure. funny part is this: if Intel ever loses performance battle to AMD again - they switch places just like two years ago - AMD will be leet CPU again with OC fanboys flocking to it in droves and Intels become popular choice again because they're gonna be CHEAP just like AMD now


By ElJefe69 on 2/15/2007 11:24:55 PM , Rating: 2
there exist hardly any benchmarks where a quad core intel chip beats the same clocked core 2 duo conroe chip.

it's sad as hell, but true. Only way K10 is going to win is by beating intel across the board in gaming. Their top end performing motherboards are normall 100 dollars or less compared to intels top performers. on die memory controllers allow for any manufacturer to be withing 3% of the best board speeds with minimal investment. IF amd could match pricing, it would crush...

then we could trust that it would obviously crush in a multicore environment with its hypertransport sexiness.



By just4U on 1/27/2007 1:14:54 AM , Rating: 3
As a Enthusiast I look at the price alot considering I also build good performing budgit machines for friends and family.

I wouldn't call those who buy from the big companies Idiot's either. NOR do they always look at the price sometimes it's just name recognition is all. <shrug>



By hubajube on 1/26/2007 1:51:35 PM , Rating: 2
quote:
Me a fanboy? Please, I'm not 12. My father helped build Intel and I am a stockholder.
Yet you make fanboy statements?


By cochy on 1/26/2007 2:05:25 PM , Rating: 3
We are consumers. Intel is to show us appreciation, not the other way around.


By PAPutzback on 1/26/2007 2:35:24 PM , Rating: 3
You got that right. All my current systems are AMD and I enjoyed the good years of the Socket A. But I think I know when a streak has coem and goen and most likely Intel is in it to win. They have the money to stay ahead now. Let's jsut hope AMD can stay in to keep the competition there so Intel doesn't go lazy again and prices stay down.

My next system will be an Intel Quad core for sure unless AMD really comes out with something special. But I don't see that happening.


VrUm vRuM VRum vrUM
By koglo on 1/26/07, Rating: -1
RE: VrUm vRuM VRum vrUM
By jondevelops on 1/26/2007 1:35:44 PM , Rating: 2
People need to realize that none of these companies will reamin on top. Are you forgetting that IBM acquired the technology to slow down light? They already have developed concrete plan for photonic processors. You may not have notice how the Intel-familiar Microsoft switched ti IBM for its Xbox and the IBM-familiar Apple witched to Intel. More like a "i got what yo u want, and you got what i want" swap. Not to mention Microsoft is in the process of making its OWN processors!! Competition is still growing.


"If you look at the last five years, if you look at what major innovations have occurred in computing technology, every single one of them came from AMD. Not a single innovation came from Intel." -- AMD CEO Hector Ruiz in 2007

Related Articles













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki