backtop


Print 62 comment(s) - last by latrosicarius.. on Dec 5 at 1:14 PM

AMD diverts resources from the Quad FX platform development the same year it promises Quad FX is the enthusiast future

AMD excited many technology enthusiasts last year when it introduced the Quad FX platform. AMD representatives touted this platform as the next big thing from AMD.

AMD apparently talked up its Quad FX enthusiast platform so well that Intel decided to roll out a competing product in its unreleased Skulltrail platform. When the Quad FX platform first hit market in January of 2007 it seemed doomed from the start to many with steep price premiums for the mainboards and the processors. These price premiums led to the lethargic adoption of the platform.

According to The Tech Report AMD representative Suzy Pruitt commented on the future of Quad FX. “The short answer is that while there are still engineering resources focused on future platform offerings that build off Quad FX, the current energy and effort has gone into programs and product initiatives like Spider and AMD has discontinued future planning and development of its eight-core enthusiast platform at this time.”

Pruitt continued, “We will continue to support customers that have an existing Quad FX with DSDC and are also working on an upgrade path for those customers. While AMD is not actively promoting AMD Opteron processor as a 2P enthusiast solution, we recognized that there are enthusiasts who are looking for two-socket solutions and think an Opteron platform is well-suited to meet that demand at this time.”

After all the promises ad statements by AMD that Quad FX was the companies enthusiast future, AMD has apparently decided to all but kill the platform off. The few enthusiasts who plunked down the big dollars required to adopt the platform should be feeling a bit uncomfortable right now.

AMD promises to continue support for the platform. However, AMD also promised the platform was the future and the company has all but killed it off the same year. The best Quad FX owners can look forward to is an upgrade to Opteron processors that work with the Quad FX mainboards.



Comments     Threshold


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

Bad news for the junkie
By TimberJon on 11/29/2007 4:44:30 PM , Rating: 3
I REALLY feel bad for the guy that bought all the top stuff just to brag about it. Like the article said, Quad-FX was hyped up to the max.

Get me interchangeable processors on a Graphics card and I will invest in that.




RE: Bad news for the junkie
By MPE on 11/29/2007 5:15:52 PM , Rating: 2
quote:
Get me interchangeable processors on a Graphics card and I will invest in that.


That sounds good but I am unsure how effective that would be when considering things like memory, controllers, video processor and etc would remain the same.

Not to mention, I doubt video card manufacturers would extend their products life cycle (aka less cards sold).


RE: Bad news for the junkie
By MonkeyPaw on 11/29/2007 5:51:27 PM , Rating: 3
Not to mention that the cost of an entire graphics card is less than that of the system it goes into. For example, an 8800GT is $220-250, and is considered upper-mid-range. Comparably, an upper-mid CPU+Mobo+RAM combo is going to easily cost $400+. Buying an 8800GTX for $500-799 is like buying the best motherboard and CPU you can buy, and we all know that costs way more. I know it's not the most elegant example, but the reality is that while interchangeable parts are great for customization, they don't make things cheaper than if the entire system is integrated. The moral of the story is that it's cheaper in the long run because the manufacturer mass-assembled the card for us. Incorporating slots and sockets on a daughterboard would increase prices across the board, not to mention that now cards would have to have BIOS's with more complexity to detect changes. No thank you.


RE: Bad news for the junkie
By 16nm on 11/29/2007 8:04:06 PM , Rating: 3
I want the entire system integrated onto the processor. I want to just plug my monitor, keyboard and mouse directly into the processor and go. I want it cheap and FAST! How long do I have to wait for this?


RE: Bad news for the junkie
By GeorgeOrwell on 11/29/2007 8:51:27 PM , Rating: 5
You can buy it today. It's called One Laptop Per Child.

It is so advanced, you do not even have to plug in your monitor, keyboard, or mouse.

You just turn it on and your future will be so fast, you'll have to stomp on your disc brakes.


RE: Bad news for the junkie
By timmiser on 11/29/2007 6:56:55 PM , Rating: 2
In the multicore cpu future, I think the era of the graphics card will come to an end. With more available cores on the cpus, I can see the graphic processing being done on the cpu and therefore no need for a graphics card.


RE: Bad news for the junkie
By afkrotch on 12/3/2007 12:36:14 PM , Rating: 2
I don't see a cpu core being used for graphics. General procs just weren't designed for such. What I can see is leaving out a core and replacing it with a gpu core. I see the standard home proc being more along the lines of a mainframe style proc.

Take for example this old mainframe proc.

http://www.computermuseum.li/Testpage/CPU-Chip-Mai...

Imagine each block being a seperate core and being capable of placing whatever style of processor you want. You can have 50 of them set to be a general cpu, 20 for gpu, 20 for physics, and 10 for system memory. Or you can simply customize it to whatever suits your needs.


RE: Bad news for the junkie
By latrosicarius on 12/5/2007 1:09:21 PM , Rating: 2
It matters what instruction set the processor is designed to execute.

When the x86 architecture first came out, there were very few "graphics-intensive" applications like games, so screen-rendering instructions were not a big-enough deal to build into the CPU's instruction set.

After games started requiring more and more speed, they had to "invent" (or at least improve) hardware acceleration in order to keep up--thus, the GPU was born.

If they can simply add GPU instructions to the CPU's instruction set, you will be able to have "true hardware" support for games without a graphics card. The CPU will be like a graphics card.

And with multiple cores, the OS could dynamically assign more cores to execute rendering threads if the game requires more resources, just as you can dynamically assign normal threads to available cores today.


RE: Bad news for the junkie
By mmntech on 11/29/07, Rating: 0
RE: Bad news for the junkie
By Gul Westfale on 11/29/2007 10:53:48 PM , Rating: 2
weren't the original quadFX chips little more than rebranded opterons? so what is stopping the owners of these boards from upgrading to faster opterons?


RE: Bad news for the junkie
By murphyslabrat on 12/3/2007 12:29:51 PM , Rating: 2
Actually, I have owned two ATI RAGE 128 PRO AGP's that had expandable memory via so-dimm socket. Also, the same GPU integrated into mobos often had the same feature. Also, some mobos with IGP's had expandability via headless AGP VRAM card. An example of this was with dell's optiplex series.

My point is that this has been done, but to buy a low-volume part to marginally increase performance (look at comparisons between 256MB and 512MB graphics card variants), while increasing cost to an, at best, relative ratio....

Not a good tradeoff


RE: Bad news for the junkie
By DallasTexas on 11/30/2007 8:17:29 AM , Rating: 4
Agree but it's fun to do a forum search for 'Quad FX' into Nov 2006 and see what posters commented on it. It is hysterical.

"...it will wipe Conroe ..."
"..Intel is dead..."
"..this is the system to get.."


RE: Bad news for the junkie
By Strunf on 12/1/2007 9:21:33 PM , Rating: 2
Funny if you do a forum search for "Phenom" you get just about the same comments.


RE: Bad news for the junkie
By MaK2000 on 12/1/2007 11:17:15 PM , Rating: 2
And Phenom is getting the same rave reviews. All hype and still slower than Conroe, muchless Penryn.


RE: Bad news for the junkie
By Denithor on 11/30/2007 8:49:12 AM , Rating: 1
quote:
Get me interchangeable processors on a Graphics card and I will invest in that.

This is actually similar to a thought I've had for a while now. How come they don't put a GPU socket directly on the motherboard instead of an entire card that connects by a relatively slow socket with limited bandwidth?

I mean come on, if we can install a CPU in a socket why not a GPU as well? Just chose a motherboard with the applicable sockets, buy the chips, and away you go. And a year down the road when the GPU is getting old replace just the chip itself at far lower cost than replacing an entire card.

This would also result in less e-waste going into the landfills (I know, I would never toss electronics but many people don't even know better or are too lazy to be bothered with proper disposal).


RE: Bad news for the junkie
By Gul Westfale on 11/30/2007 9:38:34 AM , Rating: 2
because a new chip like an 8800 would be handicapped by the small memory interface and slow memory chips of an older card.


RE: Bad news for the junkie
By ceefka on 11/30/2007 11:10:56 AM , Rating: 2
Then perhaps GPUs should also have HT(2/3) and On Board Memory Controller, maybe even their own memory slot. It makes sense, but I doubt if it is any cheaper in the long run.


RE: Bad news for the junkie
By latrosicarius on 12/5/2007 1:14:22 PM , Rating: 2
Video cards nowadays use GDDR3 RAM, anywhere from 256MB to 768MB. You would have to add more RAM to the system to make up the difference.


Better luck with NVIDIA
By Mitch101 on 11/29/07, Rating: 0
RE: Better luck with NVIDIA
By GeorgeOrwell on 11/29/2007 6:29:34 PM , Rating: 2
Nvidia is doing very well in the market. If anything, the AMD buyout has helped them in the short run. One might expect this as digesting a large company such as ATI is not an easy task.

Perhaps in the long run, AMD can offer extremely cheap combos of processor + chipset + graphics to OEMs. If executed well, it would put pressure on Nvidia. It would be no surprise if things went too well for AMD, Nvidia would file an anti-trust lawsuit against AMD. That would make Intel laugh I am sure.

However, AMD has basic execution problems -- like how to make a competitive processor -- that call into question the core viability of AMD as a long term technology player.

It is not out of the question today to envision AMD merely as a reshaped ATI, offering chipsets + graphics. AMD might even have to sell off their processor division to IBM (which already makes POWER processors), Sun (which needs something better than SPARC), Samsung, etc.

In many ways, if AMD sold their processor division to a stronger technology company it would be better for the market. AMD apparently does not have the brains/resources/focus required for the timely development and delivery of modern processors. A stronger owner of AMD's x86 technology could offer better solutions at a better price.


RE: Better luck with NVIDIA
By Adonlude on 11/29/2007 6:48:19 PM , Rating: 1
quote:
Nvidia would file an anti-trust lawsuit against AMD. That would make Intel laugh I am sure.

Intel is already lauging at AMD. Intel has much better products and probably will for a long time. Intel's stock is up. AMD's stock is at a 52wk low. In fact, AMD's stock hasn't been this low since 2002.


RE: Better luck with NVIDIA
By timmiser on 11/29/2007 7:01:04 PM , Rating: 1
AMD was doing pretty well competing with Intel a couple years ago when only they had affordable 64bit processors and only they had real dual core cpus. It won't be long until the pendulum swings back the other way and AMD is on the rise again.


RE: Better luck with NVIDIA
By darkpaw on 11/29/2007 9:42:02 PM , Rating: 2
I like to see AMD remain competitive, but this is wishful thinking. The only reason they every got such an advantage was because Intel got lazy and sloppy with the Netburst and had thought wrongly that Itanium was the future.

I really don't see that happening again. Unless Intel just rolls over and decides not to release a new architecture or process for 2 years, AMD will never catch them. It's really a matter of resources, and in that area Intel blows AMD away.


RE: Better luck with NVIDIA
By MandrakeQ on 11/29/2007 10:23:58 PM , Rating: 2
I'm not sure Intel got lazy and sloppy with P4.

There were several simulation studies done about 7+ years ago that came to the conclusion that 30 pipe stages was the optimal logic depth for a processor. However, those same studies didn't account for power and heat, the main killers of P4.

Intel could make a mistake like that again and allow AMD a chance to compete, but it looks like they will be on top for the near future. Especially now that they don't completely ignore AMD, and they make sure to cover their bases by guaranteeing their road map has something to match everything on AMD's.


RE: Better luck with NVIDIA
By S3anister on 12/2/2007 4:30:44 AM , Rating: 2
Well i guess we'll see when they try to shrink processor die down past 32nm, it'll take more and more resources, not to mention R&D... let's just hope AMD can keep competetive.


RE: Better luck with NVIDIA
By 16nm on 11/29/2007 8:16:37 PM , Rating: 2
quote:
AMD has better luck competing with NVIDIA than with Intel. If AMD can wipe out NVIDIA then AMD stands a chance to be profitable.


It should be pointed out that ATI are executing quite well, thank you. I won't say that they can "wipe Nvidia out" but they are making all the right decisions right now. AMD is not doing so well. They should have been growing their fabs instead of buying ATI, frankly. They must dramatically cut CPU prices to compete with Intel so they must sell a huge number of chips to make a profit, but dang it, they don't have the capacity to. I bet when the Intel board heard that AMD was buying ATI, they were cracking open the champagne and celebrating.


RE: Better luck with NVIDIA
By GeorgeOrwell on 11/29/2007 8:43:32 PM , Rating: 2
ATI is doing well in the OEM market due to low prices.

These low prices would not exist if AMD were not running at a loss to maintain its market share, prop up existing contracts, not lose key OEM partners, etc.

The biggest problem with ATI is that it continues to ship products that have low quality drivers. This is before and after the AMD acquisition. I have used ATI products a long time, since ATI's 8514/A compatible boards. The hardware has been great, but the drivers have always been junky.

To their credit, AMD/ATI has begun to help the open source community build open source ATI drivers. This is the best move I have seen from AMD/ATI in a long time. It might be the move that changes the game for ATI.

Frankly, AMD should have been designing/building a great follow-on to the Opteron, not opening more fabs, not buying ATI, not fielding a massive lawsuit against Intel, etc.

It doesn't matter how many fabs you have if you have nothing worth building. FAB capacity is a far easier problem to solve vs. "design and debug a world class x86 processor".


RE: Better luck with NVIDIA
By 16nm on 11/30/2007 12:44:08 AM , Rating: 2
quote:
Frankly, AMD should have been designing/building a great follow-on to the Opteron, not opening more fabs, not buying ATI, not fielding a massive lawsuit against Intel, etc.


They did design the best follow up they possibly could have. The problem is that they can never match Intel's process. Intel's chips/processes are just better manufactured and can perform better when the designs are identical, which they pretty much are now. AMD must compete on price. You get less performance, that's OK, but you must pay less for it. This is how AMD succeeded for several decades and that is where they need to be again, waiting for Intel to stumble, again. Intel may never f-up the way they did with P4 but if they do, AMD need to be healthy and ready to jump. AMD must produce more and lower production costs through volume to lower selling prices.

AMD have a good design. They need to work on their process to get more speed out of it. The design is good. The 65nm process is poo-poo.

ATI are being smart. They are working hard at improving their products and driving down manufacturing costs. Kudos to ATI. The competition between ATI and Nvidia is strong. It's a good sign for the consumer.


RE: Better luck with NVIDIA
By GeorgeOrwell on 11/30/2007 5:03:11 AM , Rating: 1
I think we're on the same page, just it's 11x17 and there's a bit of room to wander around.

AMD doesn't have the best process -- which means it will be difficult to compete on price when your costs are higher. And it also means, generally, that power usage will be higher. This is a double whammy.

AMD doesn't have the best design -- B2 stepping had so many bugs that there will be an immediate B3 and then a quick follow with B4 (keep in mind the entire chip is being redesigned for 2H08).

Maybe the current design is good for some server apps, I think that is being shown. But to compete in other markets, the design need some work beyond just bug fixes. Maybe this will be a double pumped ALU, bigger L2/L3, support for unbuffered ECC, etc.

Overall, AMD's design is now comparable to Intel's current chips, but is so buggy that shipments are still being curtailed to the minimum. And even with comparable, you then need to compete on price, availability, support, etc. Areas that AMD cannot do for long (price), cannot do because of lack of competence (availability), and cannot do because of lack of resources (support).

Ask someone at Supermicro/Tyan who is willing to share the dirt and you will hear some scathing opinions of AMD as a company of complacent idiots.

When it comes to customer purchases, we need to look at the the key metrics, such as performance/watt. Here AMD is unable to compete except in a few niche markets.

We also need to look at price. And here AMD is unable to compete without running massive losses. Maybe the moves to maintain market share will ensure survival. But at what cost? There will be less stock for employees, more dilution. Morale will suffer.

ATI is being smarter than AMD (who isn't?), but still has mostly bad drivers, poor quality control, a completely confused product naming system, etc. It is still a company that makes a good graphics board, hardware wise, but does not do well anywhere else. It is not to say Nvidia is perfect. But ATI needs to improve tremendously if it is going to have a sunny future.

At the end of the day we still have a company unable to field a quality next generation processor in any quantity other than for "early adopters". And that is a very bad problem for a processor company.


RE: Better luck with NVIDIA
By Vanman345 on 11/30/2007 12:31:50 AM , Rating: 2
While I really want a healthy viable competitor to Intel, AMD is going to have very serious problems for many years to come I think. The largest obstacle immediately in front of them is the crushing debt load from the ATI purchase. Compound that with Intel's execution with the Core 2 duo CPU family and AMD faces a herculean task ahead of them. The only way they are able to move product against a clearly superior competitor is to cut prices. Those razor thin margins on the K8 processors don't make it real easy to pay off debt and invest much in R & D.

While Nvidia faces some challenges, I believe Intel is a far bigger threat than AMD/ATI. The more feasible scenario is that Nvidia as a possible buyer of AMD. Currently Nvidia has an 18 billion market cap, zero debt and a ton of free cash, while AMD has a market cap of 5.6 billion, just slightly higher than their debt of 5.35 billion dollars. But Jen-Hsun will never even consider that, aside from the debt load, aging product line and the anti-trust considerations of Nvidia buying their only real competition, it would really piss off Intel and cut off a large chunk of business.

I hope that AMD is able to pull something out of the hat that has a chance of competing with Intel, but the odds on that happening in the next 3-5 yrs are pretty slim IMO.


RE: Better luck with NVIDIA
By hardwareking on 11/30/2007 5:14:38 AM , Rating: 2
quote:
Maybe they will be back with fusion


As Goku from Dragonball Z said "Fusion is the only way you can beat Maajin Buu"

AMD really need somethin special now


RE: Better luck with NVIDIA
By Hawkido on 12/3/2007 4:57:21 PM , Rating: 2
quote:
Maybe they will be back with fusion but it doesn't look like the Phenom is a good Intel competitor even if it scales to 3ghz some 45nm Intel chips are scaling to 4ghz already.


While I agree that the Phenom needed to be more competitive at the get go, I don't see that your statment holds as much water as you think...

#1. The Scalability of AMD's Procs is better than Intel's procs Mhz for Mhz.

#2. The Desktop market is only one of 4 realms in competition between Intel and AMD (Server, Desktop, portable, integrated). Intel has the best Desktop part hands down, They both have good Integrated parts, Intel seems to have a good edge on the mobile performance but AMD has a better power profile with it's integrated memory controller and seperated powerplane, AMD has the best Server Part, and as speeds scale up (even if Intel scales speeds at a 33% faster rate) Intel will not be able to compete with the increadibly scalable AMD Architechiture. Intel is bottlenecked at one socket and 4 cpus on a socket. Also intel cannot put all 4 CPUs on a single package, due to the same FSB bottleneck. You cannot make blanket statments about all 4 based on observations of the others.

Of the 4 sections the Server sector is the most profitable (Largest Margin), however the Desktop sector generates the most actual profit (smaller margins, much higher volume)

However with the trend going to Virtual Servers and thin clients at work places you may very well see the server sector sit on a larger percentage of the market.

If AMD somehow manages to take over a larger portion of the market with Fusion they will be protected from most forms of anti-trust litigation because they offer a different type of product. Videocard integrated into CPU or CPU integrated into Videocard, depending on the viewpoint. This should stop nVidia or Intel from being able to say AMD is acting unfairly towards them as they could license the technology from AMD or develop it themselves.

(That last part... not likely to happen, but it is the dream of AMD execs and Fanbois)


FX is dead, quad or not
By Pirks on 11/29/2007 4:26:48 PM , Rating: 2
Who in their sane mind would buy FX-74 when E6850 is $20 cheaper?

R.I.P. FX, you were so cool and trendy way back in Netburst era.




RE: FX is dead, quad or not
By timmiser on 11/29/2007 7:03:43 PM , Rating: 2
You underestimate the power of a fanboy.


RE: FX is dead, quad or not
By Pirks on 11/29/2007 7:11:35 PM , Rating: 4
Wrong. I explicitly excluded fanboys from the picture by saying "who _IN THEIR SANE MIND_". Read my post again. Thank you.


RE: FX is dead, quad or not
By Strunf on 12/1/2007 9:24:32 PM , Rating: 2
sanity is in the eye of the beholder :P


The dead end..............
By crystal clear on 11/30/2007 6:20:29 AM , Rating: 2
quote:
the current energy and effort has gone into programs and product initiatives like Spider and AMD has discontinued future planning and development of its eight-core enthusiast platform at this time.”


Now link this statement to what their CEO says-

Ruiz said AMD aimed to return to profitability soon.

"That is our No. 1 goal right now," he said.

http://www.reuters.com/article/rbssTechMediaTeleco...

Dont expect AMD to return to profitability anytime before 2Q 2009.(that is if they dont messup their R&D/product launches/marketing/manufacturing/transition to 45nm etc.)

To achieve this major objective it has to drastically increase its revenues & more drastically cut its cost.

In cost cutting measures the future of Quad FX. is dead & will announce/or not announce the death of many more other projects in hand.

They have no choice but to scrap all those projects as funding them is scarce like water in the desert.

In addition to lack of funds, technically they AMD have reached a dead end with reports of serius problems in its transition to 45nm.
The not yet launched higher end Phenoms also are slow in coming.
AMD does not deliver in time-neither their R&D nor its manufacturing.

AMD Roadmaps in 2007 have been abruptly changed frequently with no explainations & neither its benchmarks can be trusted.
They have turned the serious business of Roadmaps to a sales gimmick.& benchmarking an advertising ploy or tool.

Their launches of Barcelona & Phenoms are more like paper launches rather than a hardcore product launches like Intel.
We have yet see an OEM of some repute selling Barcelona based systems.(All of them are complaining.)

Hector Ruiz, AMD's chairman and CEO, was in India Thursday to announce the opening of a new, 52,000 square-foot AMD design facility in Bangalore that will house up to 350 engineers.

AMD Expands R&D Presence in India

http://www.amd.com/us-en/Corporate/VirtualPressRoo...

They make it appear as an expansion when in fact it is drastic slimming down.
This is a cost cutting measures as its cheaper to operate in India rather than the USA.
AMD R&D in the USA will be scaled down drastically with a lot of staff being FIRED.

Their Asset lite stratergy is destined for a failure in its desperate bid to cut cost.

AMD share prices have fallen to their lowest levels & will need further funding mid 2008.

AMD is reaching a DEAD END.




RE: The dead end..............
By crystal clear on 11/30/2007 7:19:11 AM , Rating: 2
On Tuesday, Dell introduced Precision workstations using Intel's newest Quad-Core and Dual-Core Xeon 5400 and 5200 series processors. Both series are based on Intel's 45-nanometer manufacturing process. The workstations are available "today."

Dell has yet to announce computers based on the quad-core Opteron "Barcelona" processor despite being "launched" more than two months ago by AMD.


By crystal clear on 12/1/2007 3:24:36 AM , Rating: 2
Advanced Micro Devices, Inc. (AMD)
9.76 -.3800 ( -3.75%) Nov 30 2007 16:03 EST


RE: The dead end..............
By crystal clear on 12/1/2007 3:51:49 AM , Rating: 2
CEOs Who Need To Return To Business School

Looking around the wreckage of some of the big cap companies it is not hard to find a few CEOs who probably need to go back to business school. Wall St. would like to see them at Harvard so they would get the best education possible, but the school in Cambridge might not let them in.

First on the list is Hector Ruiz of AMD (AMD). He already has a PhD, but it may be the money order kind that you can get through the mail. Ruiz has effectively taken AMD from a high-margin chip company which had the technology to compete with larger rival Intel (INTC) to a sad shadow with big debt and small margins. He engineered the buy-out of graphics chip company ATI, which has been a disaster of their first order. Less than two years ago AMD traded at above $42. It now sits at under $10.



http://www.247wallst.com/2007/11/ceos-who-need-t.h...


RE: The dead end..............
By MandrakeQ on 12/1/2007 2:12:38 PM , Rating: 2
Hector Ruiz got his Ph.D. from Rice. That isn't mail order by anyone's standards. Also, your posts are really hard to read with the irregular use of bold and italics.


Opterons?
By ViRGE on 11/29/2007 4:49:23 PM , Rating: 3
While I had been expecting AMD to kill the Quadfather, their choice of an upgrade path is odd. The differentiating feature about the FX chips compared to the Opterons was that they used faster, cheaper unregistered memory that's commonly used in desktop computers. Meanwhile the Opterons require said buffered memory, meaning anyone that wants to drop in an Opteron to replace an FX chip would also have to replace their memory.

At that point, why not ditch the platform entirely? It's really not a viable upgrade path.




RE: Opterons?
By GeorgeOrwell on 11/29/2007 6:05:45 PM , Rating: 3
The Opteron 100/1000 series of processors (which support only one socket) have been capable of using plain unbuffered ECC (vs. registered/buffered ECC) for a long time.

Likely AMD would only have to tweak a bit on the Opteron memory controller to enable the use of unbuffered ECC RAM on the Opteron 2000 series processors. Something like this might be even doable via the BIOS.

If AMD had been intelligent, they would have done this long ago, further capitalizing on the Opteron's built-in memory controller.

This change have enabled dual socket (2S) systems to be built with cheaper/faster RAM with the limitation of fewer banks being supported. As an Opteron 1000 processor supports 4 DIMMs (at least), the limitation would be small, presuming that each Opteron 2000 could also support 4 DIMMs of unbuffered ECC RAM, giving the user 8GB to 16GB max (1GB/2GB DIMM size).

Of course, Opteron servers could keep using ECC REG to pick up the support for more banks.

AMD would likely have profited from being more aggressive in pushing support into their memory controller for newer/better RAM vs. playing it safe. Intel is usually trying the new memory years before AMD supports it (ala DDR3).

If AMD is willing to go to the lengths of making a three-legged CPU, I don't see why they would not spend the minimal effort to better segment the Opteron platform into workstation/high-end gaming and server platforms. Without the high cost of ECC REG DIMMs, two socket Opteron boards would be very appealing to many.

QuadFX could have done all this, offered well thought out improvements to the existing Opteron platform, but instead delivered nothing more than an ugly baby with an extreme thirst for the milk, ultimately a baby that AMD decided to throw in a dumpster.


amd is riding their pony over the hills
By GeorgeOrwell on 11/29/2007 5:46:57 PM , Rating: 5
If AMD had put 10% of the effort they put into Quad-FX into working with their partners to make dual Opteron motherboards better, that would have had a good payoff in making the Opteron platform stronger for content development, game development, and high-end gamers.

The erstwhile invention of a new platform only for gamers (and not content creators), Quad-FX, was pure stupidity.

Meanwhile AMD has sat on HTX for years and developed essentially nothing that would make HyperTransport a market force. If AMD had developed a game physics processor for HTX, that might have been interesting. And there were already a number of Opteron motherboards with HTX.

In a couple years Intel is hitting the beach with its Nehalem stormtroopers. You can be sure Intel will not be stupid and will have all sorts of cool stuff you can plug into a QuickPath socket.

Although I have a number of Opteron systems, they are all being retired and replaced with Intel. Today's Intel systems offer greater stability, greater compatibility, and better performance/watt.

Unfortunately AMD has done nothing of late other than show that they are riding a one trick pony -- the integrated memory controller -- into a dusty oblivion over the hills of Brokeback Mountain.




By mendocinosummit on 11/29/2007 7:11:22 PM , Rating: 1
Brokeback AMD.


Seems clever.
By Clauzii on 11/29/2007 5:21:46 PM , Rating: 2
Focusing on the spider platform might be a good move from AMD.

Already having GPU tech in-house, and maybe Physix too, with the right 'donation' of money put into refining the Phenoms quick, will, maybe, make it possible for AMD to actually put out some interesting SpiderFusion solutions.

Being able to deliver Chipset, CPU, GPU and Physix is probably needed to compete with intel after their Havoc purchase. And if someone other than AGEIA can do better at "Fusioning" the technoligy into some REAL (read 'better performing'!) hardware, it seems like a clever move.

There should still be 500+ billions left from the current almost 10% sale of stocks left, to cover some of the current wounds.

For AGEIA it is a good win situation, since the dies themselves would be fabbed in-house instead of at some remote facility, probably making changes to design cost less and take more time. So if there would still be PCIe Physix cards on the market (for those without SpiderFusion) but priced at~$60, I think that Physix as hardware would become more attractive to the mainstream user.

I COULD(!) imagine a future lineup of Fusion CPUs where the PHYSIX chip was bad. Then theese CPUs could be for the low end of the market but still having the ability to buy a cheap Physix card to make a-kind-of "SpiderFusion Light".

And last, I don't think AMD needs to compete with intel directly, but keeping their focus on what the K8 was so full of at the time: Innovation!




RE: Seems clever.
By Clauzii on 11/29/2007 5:23:54 PM , Rating: 2
Let me rephrase: "...cost less and take LESS time..."


Slight of hand
By BruceLeet on 11/29/2007 9:08:51 PM , Rating: 2
"Hey Mr.Ruiz, lets pretend to start on something "big" to make Intel look into it and start researching their own competition for it, and if they fall for it we know they've spent money and once they launch it as they always do before us, lets pull out and work on something that will work half-ass against Nvidia, then when the shit hits the fan you resign as CEO and I get your job" - Dirk Meyer Chief Operating Officer




RE: Slight of hand
By wordsworm on 11/29/07, Rating: 0
RE: Slight of hand
By BruceLeet on 11/30/2007 1:49:16 AM , Rating: 2
Take it easy Mr.Intel fanboy, I was simply poking fun at AMD no need to get all butthurt from someone else's mistake, on that topic I just got home from College so I got blazed up took the shortcut while typing doesn't everybody? I too noticed the mistake after I read others comments and saw "slight of hand" I was thinking lol wtf, opened it up sure enough there's my post, things are just funny when your blazed on stres, you need to loosen up :) (I didn't spell stres wrong, its slang you prude)

Btw I could care less about *8GB Graphics Chips and *5.4Ghz Processors, as long it can Compile my code then Im happy ;)*sarcasm

*sigh* gaming extremists *rolls eyes*


hmm...
By Etern205 on 11/29/2007 10:28:57 PM , Rating: 2
Since AMD ditched their Quad FX platform.
I wonder if Intel is going to drop their skulltrail or if not, then a lower in price as well as using DDR2s or 3s would be nice instead of FB-Dimms.




RE: hmm...
By GeorgeOrwell on 11/29/2007 11:03:21 PM , Rating: 2
This is a good question to ask, especially as FB-DIMM goes the way of the dodo when Nehalem hits the beach in 2009.

Skulltrail seems like a good way to waste a lot of money buying cores and threads that most apps and games will never use. The world is at least 5+ years out from the everyday utilization of four cores, not to mention EIGHT.

Not to mention all the extra electricity that a second processor and FB-DIMMs will use.

For most games, one/two great graphics cards + overclocking one quad-core (4Ghz?) will give you better performance than 8 cores and high-latency FB-DIMMs.


Octo FX
By Sureshot324 on 11/29/2007 11:16:59 PM , Rating: 2
Instead of killing this platform, AMD should have made it work with 2 quad-core Phenoms for 8 total processors! Unlike quad FX, they could truly claim to have the fastest platform and a lot of rich people would pay for for it.




RE: Octo FX
By chromoplastic on 11/30/2007 12:10:23 AM , Rating: 2
Quad, Octo, whatever they come up was just a despaired marketing trick. AMD has no follow up to the Athlon 64, it's just simple as that. They just marvelled at themselves with the x64 "innovation", but forgot that Intel by the time had that too... and it brought no money, just technical and fans recognition. Intel responded and AMD was caught with their pants down. My experience as a web/mail admin is that after almost 3 years we are replacing every AMD based server with an Intel one. It's just simple as that. And i'm very sorry because we need competition, and Intel has none for the next quarters... maybe years to come...


Who's Surprised?
By Vanilla Thunder on 11/29/2007 4:26:39 PM , Rating: 3
This was nothing more than a stop gap solution that NO ONE should have bought into. That is unless they had some cash burning a hole in their pockets, and had a nuclear power source in their home. As much as it sucks for owners of this platform, I'm glad to see that AMD is not going to sink anymore money into this pointless venture. Bring on the Fusion!!!

Vanilla




Seriously
By BruceLeet on 11/30/2007 2:14:35 AM , Rating: 1
Why would they do this if they promised Quad FX was "enthusiast future", A company just doesn't drop development/research on products that they "promised" was...promising. Haha but you know AMD! They do crazy things.

Now is it possible they have found/done/developed something more versatile/innovative with "Spider" that just blows Quad FX off the to-do list? Quad FX and Spider are both platforms, but one got sacked, Hmmmm. I can understand how internet people (mostly naive gaming enthusiast-younglings) who do not understand the Coporate world; somehow came up with the idea that AMD/ATI merger was going to blossom overnight, it takes years for something to come to fruition after a merger of such size, some fail to understand this.

Anywaywhohow, I think AMD wants to put more time and money into Spider and future projects




RE: Seriously
By GeorgeOrwell on 11/30/2007 5:33:42 AM , Rating: 2
For those of us who have come from the corporate world, we also know that many mergers fail or simply limp long, dying a long lingering death, as they were never meant to be.

As for "Spider", I'd put money down that it won't last a year. Even more money on "dead in two".

There is no "platform" here, just a random assemblage of parts that don't work together in any way except PCIe 2.0 video card + PCIe 2.0 chipset. That's not a whole lot to base an entire "platform" on.

There is not even an HTX slot or external connector for "Spider". Now that would have been interesting as you could have linked two "Spider" machines together as a transparent NUMA architecture. Each Spider's memory, video, peripherals, etc. would be part of one larger machine. This feature alone would have enabled innovative applications that could not be easily built otherwise.

I guess after buying so many Opterons and seeing the newest Opteron to be nothing more than a buggy dud that does well on a few isolated benchmarks, there is not a lot of confidence AMD will be shipping anything good until Barcelona+, the redesign, coming 2H08.

And there is still that inkling that if AMD doesn't pull off the redesign, we will be seeing some executives who will be sent to Allah. Maybe that will finally motivate the company out of their complacency... :-)


-_-
By Regs on 11/29/2007 5:04:03 PM , Rating: 2
Who the hell were they kidding?




Bring on Fusion
By MustangMike on 11/29/2007 6:46:34 PM , Rating: 2
I'm still waiting for day when they unveil a platform that's a quadcore cpu/ Graphics / Physx all in one package. If not use the Dual Socket Quad Core setup. Run the Quad Core CPU in socket 1 and the Dual Core Graphics and Single core Physx on Socket 2. Even IF it wasn't better then Intel I still would like to see that come out.




Another Stupid Move
By shabodah on 11/30/2007 1:40:53 PM , Rating: 2
I'm a big supporter of the underdog, but, AMD is really screwing the pooch lately. In any business, you have to use your advantages, not try and compete against your rival's advantages. So:

Dear AMD,

Please do the following you can get back to profitability and gain some of us enthusiasts back:

1)Get HT3 on the Opteron Platform IMEDIATELY. It does not matter that people in Taiwan don't like the idea, how well your platform scales should be your MOST important priority in the server market.

2)Forget Desktop Quad-Core Chips. Every good quad-core you make should be going into the server market. Period.

3)Expand Tri-Core on the Desktop. The ability to individually clock each core for max performance is an enthusiast's dream. Fewer cores per di means less total heat, which will allow better single-threaded performance, which is where you need to focus RIGHT NOW.

4)Release a Dual-Core processor based on K10 AS SOON AS POSSIBLE. Most of the SAME REASONS AS ABOVE.

5)Single-Core processors should end up as Semprons on the desktop for only ONE REASON; They did not make the power requirements cut for the lower power devices they were intended for.

6)4x4 is Actually a Good idea, but it will NEVER be successful without ATX support. VERY few people are going to buy an E-ATX case just to put one motherboard in, and the few that do are already alright with their Opteron oriented workstations. Furthermore, the processors promoted for these systems should also be lower-wattage than their normal desktop equivilents. I said at the original release, I have no interest in two dual-core 3.0ghz 120 watt processors, but I'll buy two dual-core 2.0ghz 35 watt processor RIGHT NOW.

7)Spider needs it's IGP version ASAP, preferably a hybrid solution that can work with a cross-fire setup. This would be especially fun on a mATX board, catering to Lanparty type folks. Furthermore, expand everything you can with this platform to high-performance notebooks.




just in case anyone cares....
By maverick85wd on 12/2/2007 8:51:35 AM , Rating: 2
this is a paper I wrote for economics has to do a little bit with microprocessors... hopefully I'll get an A!

http://blog.myspace.com/index.cfm?fuseaction=blog&...




QuadFX failed
By exdeath on 11/30/07, Rating: -1
"The Space Elevator will be built about 50 years after everyone stops laughing" -- Sir Arthur C. Clarke

Related Articles













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