backtop


Print 67 comment(s) - last by Lerianis.. on Feb 14 at 8:42 AM

AMD benefitting from Intel's $1B mistake

Intel must be the gift that keeps on giving to AMD. In late 2009, Intel was forced to pay AMD $1.25B for its years of monopolistic behavior in the processor market. Now, AMD is benefiting directly from Intel's Series 6 (Cougar Point) chipset woes

In late January, Intel explained that it had uncovered a SATA problem with Cougar Point that could cause performance degradation over time. As a result, Intel stopped shipment of the affected chipsets, will take a charge of roughly $1B, and won't start shipping updated chipsets to its partners until later this month (volume shipments to customers will come later).

Not surprisingly, AMD is profiting greatly from Intel's gaffe according to Fox Business News. "We have some customers and retailers who have come to us specifically as a result of Intel's chip problem," stated AMD exec Leslie Sobon. "Some retailers have had to take things off their shelves, so they call us to ask what they could get from our OEMs that's similar. And OEMs are asking us for product, as well." 

Fox Business News goes on to report that computer manufacturers that sell both Intel and AMD systems are coming to the latter to supply more processors/chipsets to make up for the shortfall due to the Cougar Point woes. 

AMD earlier this year unveiled its Fusion processors that feature onboard DirectX 11 graphics. AMD says that netbooks using the chipset will have over 10 hours of battery life.

AMD has also been in the news recently thanks to the ouster of Dirk Meyer as CEO. Meyer was removed from his position due to a lack of vision when it comes to mobile computing devices like smartphones and tablets.



Comments     Threshold


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

RE: Serves them right...
By seamonkey79 on 2/9/2011 9:33:44 AM , Rating: -1
Socket A, every time a new chip was released, lost the ability to use the new chips... Thunderbird era boards didn't all run Palomino, Palomino era didn't all run Thoroughbred, Thoroughbred didn't all run Barton. Socket 754 was dead before it came out, 939 was replaced with AM2, AM2 don't run all AM2+

While I'll grant you we've been in an AM2+/AM3 world for awhile, it doesn't change the fact that same socket or not, we still, as a general rule, needed to replace the motherboard with our CPUs for many years, primarily because the new CPUs with the same socket layout didn't work with the board's from the previous CPU iteration.


RE: Serves them right...
By Da W on 2/9/11, Rating: 0
RE: Serves them right...
By Denigrate on 2/9/2011 10:13:28 AM , Rating: 2
939 did NOT last all that long. Not sure what you are smoking.


RE: Serves them right...
By Shadowmaster625 on 2/9/2011 11:25:51 AM , Rating: 2
Socket 939's life was shortened by forces outside its control. It was Intel that forced the move to DDR2, even though there was no measurable performance increase vs DDR400. The same happened with DDR3. It is amazing that intel has not forced a move to DDR4, to suck some extra money out of everyone after their latest billion dollar debacle.


RE: Serves them right...
By MonkeyPaw on 2/9/2011 1:03:58 PM , Rating: 2
I think Intels worst concept recently was FB-DIMMs. Hot, inefficient, expensive, and only useful until Intel could get QPI and an IMC in place to catch up to K8 Opteron. Can you even buy FB-DIMMs anymore? So much for top-dollar getting you something better.


RE: Serves them right...
By Cheesew1z69 on 2/9/2011 1:11:13 PM , Rating: 2
2 years in the computer world is actually quite a long time....


RE: Serves them right...
By YashBudini on 2/9/2011 4:52:17 PM , Rating: 2
Really, it's 14 dog years and mine isn't running anywhere near 3GHz.


RE: Serves them right...
By RjBass on 2/9/2011 10:17:08 AM , Rating: 5
I have an AM2/AM2+ board that I just upgraded from and Athlon 64 X2 to a Phenom II X3 and the upgrade was flawless. My buddy just finally switched from a socket 939 board to a AM2/AM2+/AM3 board. He can still use his DDR2 RAM, he got my Athlon 64 X2 and he can upgrade to a Phenom II X6 if he wants using a socket AM3 CPU. AMD has made it so easy to upgrade with their recent boards that switching to Intel for me or most of my friends would cost hundreds if not a thousand more dollars when we can just spend 1 or 2 hundred with a smaller simple upgrade and still achieve awesome performance. The only time in the last ten years that I have complained about AMD's sockets was when i purchased a socket 754 board that they quickly abandoned. Since then it has been all good.


RE: Serves them right...
By silverblue on 2/9/2011 11:02:22 AM , Rating: 3
754 was single-channel. I think AMD quickly realised that the Athlon64 could really make use of the extra channel.


RE: Serves them right...
By DanNeely on 2/9/2011 11:22:41 AM , Rating: 1
939 launched when AMD made the first x2 processors. A single DDR1 channel couldn't feed both cores. IIRC there were minimal gains for 939 single core CPUs.


RE: Serves them right...
By Cheesew1z69 on 2/9/2011 1:14:27 PM , Rating: 2
So you are saying it came out when X2 was released? I am pretty sure it came out before the X2 as they released single core when 939 was out.


RE: Serves them right...
By Indigo64 on 2/9/2011 4:33:28 PM , Rating: 3
Almost accurate.

Socket 939 was the answer to AMD's socket 754 "negative" of being able to only use one channel of memory at the time. Since Intel was using dual channels on the P4 platform, AMD needed an answer to fulfill the enthusiast market (more or less) and thus came Socket 939.

Since AMD had just perfected the Opteron at the time (Socket 940) and wanted to give folks on the non-server side an option for dual channel loving without needing to use Registered RAM meant for servers, Socket 939 was introduced.

The first Athlon 64's that inhabited 939 were the slightly faster (both in terms of clock and cache) single core Athlon 64's. The dual cores then followed, such as the Athlon 64 X2 3800+, 4200, 4400, 4600, etc.

On a personal note, It was a great platform at the time. I owned several 939 chips, and my current server 2008 R2 box uses a 4600+ to this day.


RE: Serves them right...
By Taft12 on 2/9/11, Rating: -1
RE: Serves them right...
By Tegrat on 2/9/2011 11:44:30 AM , Rating: 2
Asrock 939DualSata2 had an AM2 Riser card that supported DDR2 and AM2.....


RE: Serves them right...
By RjBass on 2/9/2011 11:48:27 AM , Rating: 2
Not true at all. Later 939 boards including boards manufactured in 2008 and 2009 supported both DDR and DDR2 as the 939 chips had the memory controllers for both, or it was written into the chipset (not sure which). Regardless my buddies old board had two slots for DDR RAM and two slots for DDR2.


RE: Serves them right...
By Alexvrb on 2/9/2011 4:33:08 PM , Rating: 3
No, he's right. Google is your friend. Socket 939 CPUs have an IMC that only supports DDR1. This has NOTHING to do with the motherboard. The IMC (integrated memory controller) is built into the socket 939 CPU itself and WILL NOT support DDR2. I should know, I still have a box with an FX-60, the fastest Socket 939 processor AMD released.

Your friend either had an early AM2 board, or a riser card, or you're mistaken altogether.

By the way, riser cards for socket 939 boards don't really count as "DDR2 support for socket 939", since they DO NOT add DDR2 support to your existing Socket 939 processor. Instead, you have to add a completely new AM2 CPU (which has a new, DDR2 IMC built in) to the riser card alongside the new memory. Basically the riser card was just a secondary (AM2) motherboard that shares some components with the primary board. You can't run both processors at the same time, either.


RE: Serves them right...
By BSMonitor on 2/9/11, Rating: -1
RE: Serves them right...
By RjBass on 2/9/2011 2:26:48 PM , Rating: 2
I can understand your argument, but seeing as how AMD is doing just that, and people seem to be loving it, I don't see how your argument is valid.


RE: Serves them right...
By Belard on 2/9/2011 10:24:41 AM , Rating: 5
UH.... your AMD history is screwed up.

Your talking about CPU types, If a new CPU comes out in which its FSB exceeded a chipsets ability, then its a problem that requires a new board. But overall, for Socket-A - the last Socket A type boards would run every Socket A cpu ever made.

754 was a stupid thing, not as bad as 940. This was time to market issues. 754 wasn't crap and actually lasted about 3 years, far longer than intel's 1st gen i3-5-7. Socket 939 lasted a very very long time and the reason it retired was that it didn't support DDR2, how could it?

AMD's AM2~AM2+~AM3 have worked great. If you bought an AM2 CPU, it would work on AM2~AM2+ boards. AM2 CPUs don't work with DDR3. See?
AM3 CPUs work on AM2+ and AM3 boards. Thats pretty cool.

So, unlike intel - AMD keeps their sockets far more compatible and longer than intel.

How about the first Pentium 4 CPUs? It was the dumbest thing people bought. Really, it was marketed by intel fans and corp "Buy the P4 for the future". uh, okay... not.
A) P4s were slower than AMD and P3 CPUs, until it hit at least 2Ghz.
B) You paid out the nose for RD-RAM
C) First Socket 423 (sp?) had about a 9month life-span... no CPU or mobo support after socket 478.

So how exactly was buying the first P4s, an investment into the future? It wasn't. Intel made a boat load of money off junk.

The i3-5-7 sockets are a nice mess. I5 or i7 series CPUs can fit into two different types of sockets, depending on which one you got. For many, the 6core option wasn't going to happen. Meanwhile, AMD AM3 sockets works with 6core CPUs.

Now a 2nd gen of i3-5-7 CPUs which require totally incompatible socket & motherboards are out. Making that $300 ASUS mobo totally USELESS for an upgrade. A typical upper end AMD mobo, with similar feature sets go for $100.

When AMD's Bulldozer comes out, it will require a new socket... AM3+
Its a vastly different chip than the current AMD Line, but still uses DDR3. Bulldozer won't work on AM3 boards, but AM3 CPUs will work in AM3+ boards. So getting SATA 6GB, USB3, modern mobo to work with a complete range of CPUs... is good for the industry and the consumer. Why? As a PC manufacture, a single board/socket model is needed to sell AMD products. Low-end to high end.

For the intel line, there are currently 4 different Socket types for gen1 and gen2 i3-5-7 CPUs that have to be managed. Run out of one type of board, theres a problem.

Needless to say, Core2, Corei5(gen1) won't fit or work on any of the latest Intel boards.


RE: Serves them right...
By BSMonitor on 2/9/11, Rating: -1
RE: Serves them right...
By omnicronx on 2/9/2011 4:14:07 PM , Rating: 4
quote:
LOL, Core i3, Core i5, and some Core i7's have gone through 3 such major north bridge changes caused by integration on the CPU die. F
And your point is? Does that change anything he said? Is the the problem of the consumer that Intel waited so long to do things such as put the memory controller on die? And then soon after quickly shift to GPU on die?

The fact remains, AMD's products over the last many years have had a great track record of backwards compatibility, and regardless Intel's reasonings, they have not.


RE: Serves them right...
By BSMonitor on 2/10/11, Rating: -1
RE: Serves them right...
By omnicronx on 2/10/2011 12:02:50 PM , Rating: 3
Yes bud, I take the time and log onto a bunch of accounts just to rate you down.(in spite of course)

As though I care about the DT rating systems or what people think of my posts.

The argument that AMD has not changed anything just makes me laugh, AMD has had their memory controller on die since 2005. As a result they were ahead of Intel so they did not have too, are you seriously trying to imply that is a bad thing?

Your CPU architecture comment is also incorrect. K7,K8,K10's all had significant changes. There were more if you consider non significant changes. (and thats all within the last 6 years)


RE: Serves them right...
By BSMonitor on 2/11/11, Rating: -1
RE: Serves them right...
By Hieyeck on 2/9/2011 10:36:55 AM , Rating: 3
Sockets gotta change SOMETIME.

I have one of the last AM2+ chips running on an AM2 board. All I had to do was spend 5 minutes using the Asus util to update the BIOS (straight from Windows to boot!). Best motherboard I've owned considerig I'm still rocking load times with a $250 mobo from 6 years ago.

AMD can only do so much, they provided the tools and the support, but it's ultimately up to the manufacturer to decide whether to update their BIOS or not.


RE: Serves them right...
By Drag0nFire on 2/9/2011 2:43:21 PM , Rating: 1
Equally noteworthy, the relative stability in AMD's motherboards in recent years may be tied with lack of innovation in their processors. The Phenom II line was released in 2008(!), and I don't believe their architecture has changed significantly since then (we'll see about Bobcat/Bulldozer).


RE: Serves them right...
By kc77 on 2/9/2011 5:10:29 PM , Rating: 2
That is NOT correct. New chips required BIOS updates nothing more. You did not have to throw the motherboard away for socket A. Socket A in particular lasted for an extremely long time. They supported processors from Athlon 600 to the Athlon XP's. Now you might not experience the fastest bus speeds by hanging on to your original Socket A motherboard but the top tier motherboard manufacturers supported that socket from the afore mentioned models.Mine in particular was a KG7 and it when from Athlon 800 (was the first chip I bought) and I stopped at Athlon XP 1800 and then went to the 754 socket.

Socket 754 was short lived. However, socket 939, lasted about as long as anything Intel had at the time (average for Intel seems to be 2-3 years).

AM2 -> AM2+ compatibility is determined by the amount of wattage required. Again with a cheap motherboard you might run into problems. Good quality boards supported everything from Athlon 64 to Phenom Quad Cores. I know they do ... well cause I have one.

For Intel your premises might hold true, but by and large AMD has been sticking to it's sockets for a much longer time.


"Folks that want porn can buy an Android phone." -- Steve Jobs














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