**AMD's stop-gap solution** Edit: Plus a long-term solution

[lopri]

http://www.dailytech.com/article.aspx?newsid=2638

I predict we will not see anything like this in near future. I mean.. 2 sockets with 2 dual-core FX CPUs.. in 2006(!) that's $2,500 right there. I wonder if this presentation is even legit? And even in my imagination, I don't know how this will boost performance in current desktop applications.

Edit: AT posted AMD's loadmap analysis article now and there are a couple more interesting things.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2768

Edit 2: More about HTX.

Originally posted by: lopri
Regarding the new HTX, excuse me for quoting myself from article comments:

I totally agree that the "direct connect" is the most desirable way but I cannot help but think AMD is somewhat daydreaming. That is, what's showing in the slides seems way ahead of today's "practicality".
I mean, we've had this PCI Express which has been strongly pushed by core logic vendors, but so far all we practically have are video cards. I sometimes think all these mobo makers pay more attention to "asthetic" point when they design PCI-E slots so the boards look prettier. (lol)
If my understanding is correct, AMD will introduce a new type of slot, HTX, on motherboards. Will other technology/market follow? Or will it just give another chance to graphics card manufacturers to push us to buy new cards? On today's desktop boards, basically everything is "integrated", sans video. I know that a video card has its own core and frame buffer, and transfers data via PCI-Express, but if a physics card can utilize the HTX, what stops a video card from connecting directly to CPU, without passing the core logic or system memory?
I think this will also be closely related to the available bandwidth of HTX per CPU core (or cores), and I can't really think of any add-in board that'll prioritize the bandwidth other than video cards, (OK and the physics cards) even though the HTX will be an open standard. (look at the lazy/lame Creative)
A very desirable case would be where storage (hard disks) can take advantage of this "direct" connection but then again there is a such thing called "memory", so my imagination stops there. (maybe solid-state/I-Ram type of storage can make use of the HTX? Then what's the use of memory? Taking care of I/O?) Talking about I/O, I just thought it'd be interesting to see keyboards/mice connect to CPU via HTX. (Sorry I couldn't resist)
All in all, like the article says, this roadmap seems just too broad/ambiguous/futuristic. I'm not a CPU engineer so my thinking could be totaly off, though. If so, please enlighten.

And here is AT's impression on this new technology:

Stepping past the enthusiast platform, we have arguably the most exciting announcement of the day: Torrenza. Along with K8L, AMD plans on openly licensing it's (until now proprietary) coherent HyperTransport technology. At first glance, this may not seem exciting, but AMD is throwing in a little twist: HTX slots. These HTX slots will be standard interfaces connected directly to an AMD CPU's HyperTransport link. If both of these links are coherent, the device and the CPU will be able to communicate directly with each other with cache coherency. Because of this, latency can be reduced greatly over other buses as well, enabling hardware vendors to begin to create true coprocessor technology once again

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2768&p=2

Basically AMD wans to connect everything via HyperTransport, it seems.

 

[Viditor]

Originally posted by: lopri
http://www.dailytech.com/article.aspx?newsid=2638

I predict we will not see anything like this in near future. I mean.. 2 sockets with 2 dual-core FX CPUs.. in 2006(!) that's $2,500 right there. I wonder if this presentation is even legit? And even in my imagination, I don't know how this will boost performance in current desktop applications.

This is basically a faster quad core MCM solution for the desktop using unbuffered Ram.
While it's true that 99% of the people won't need it, it's also true that 99% won't need the Kentsfield either.

 

[Looney]

The enthusiasts market is what put AMD where it is today... it's what people talked about, what people bought, and what people told others. But i don't think AMD entirely understands who the enthusiasts are. They're not just the ones popping down $1000 for a CPU, they're the ones that buys $300 Athlon64 and sees how it kicks a P4s ass.

My prediction still stands... it looks like AMD is doing exactly what Intel did with the P4. They couldn't compete in the actual consumer market, so they're creating EE-type CPUs that only 0.1% of people are going to buy.

 

[bunnyfubbles]

Originally posted by: Looney
The enthusiasts market is what put AMD where it is today... it's what people talked about, what people bought, and what people told others. But i don't think AMD entirely understands who the enthusiasts are. They're not just the ones popping down $1000 for a CPU, they're the ones that buys $300 Athlon64 and sees how it kicks a P4s ass.

My prediction still stands... it looks like AMD is doing exactly what Intel did with the P4. They couldn't compete in the actual consumer market, so they're creating EE-type CPUs that only 0.1% of people are going to buy.

Except $300 is more expensive than intel counterparts because the chip is actually better. It used to be the AMD chips were cheaper.

There was this chip I remember that AMD released, you might have heard of it and the line it started, it was called the FX51 - that one seemed to work...maybe you forgot about it, but the FX line is just as expensive and as absurd as Intel's EE.

You might think you know all the answers and that you know AMD is doing something wrong, but while you're playing armchair CEO, I'll play critic and question how well you understand marketing. Flagships aren't exactly released to be cash cows - the real money is made on the budget/mainstream side of things. HOWEVER, the flagships are your crown jewel, your bragging rights - its what gets people interested in you, if you can't afford the mighest of the flagships, at least you can be part of their fleet and claim, "oh yeah, I own AMD"

If AMD is in for 2nd place to Conroe like things seem most likely to be, are they to just give up? Are they to go lock themselves into a bathroom until someone magically craps out new CPU that is faster than Conroe clock for clock? The fact is, this actually seems like a plausible idea (mostly thanks to HTT). It would allow AMD to at least brag about "towing more payload" should they lose the race - and just like the FX51, it is possible the technology could trickle down closer to the mainstream.

 

[Looney]

Originally posted by: bunnyfubbles

Originally posted by: Looney
The enthusiasts market is what put AMD where it is today... it's what people talked about, what people bought, and what people told others. But i don't think AMD entirely understands who the enthusiasts are. They're not just the ones popping down $1000 for a CPU, they're the ones that buys $300 Athlon64 and sees how it kicks a P4s ass.

My prediction still stands... it looks like AMD is doing exactly what Intel did with the P4. They couldn't compete in the actual consumer market, so they're creating EE-type CPUs that only 0.1% of people are going to buy.

Except $300 is more expensive than intel counterparts because the chip is actually better. It used to be the AMD chips were cheaper.

I'm not actually comparing Conroe. I'm just saying what made AMD so successful. It wasn't just the price. Cyrus and the old AMD K6s back when the Celerons were around were cheaper, but they didn't do well because performance sucked. Even until recently when people found out how well Intel OCed, Intel is significantly cheaper than X2s and enthusiasts didn't flock to Intel.

There was this chip I remember that AMD released, you might have heard of it and the line it started, it was called the FX51 - that one seemed to work...maybe you forgot about it, but the FX line is just as expensive and as absurd as Intel's EE.

You might think you know all the answers and that you know AMD is doing something wrong, but while you're playing armchair CEO, I'll play critic and question how well you understand marketing. Flagships aren't exactly released to be cash cows - the real money is made on the budget/mainstream side of things. HOWEVER, the flagships are your crown jewel, your bragging rights - its what gets people interested in you, if you can't afford the mighest of the flagships, at least you can be part of their fleet and claim, "oh yeah, I own AMD"

I probably know more about marketing than you do. Flagships line does help, but not with the enthusiasts, who knows what the actual product can do. Flagships helps those who are ignorant, and read some review that was based simply on the flagship product. We're in the internet age. Go get an nvidia FX GPU and brag how awesome of a performance it is based on your 'mightest flagships' and you'll be quickly put in place. AND you'll realize how ripped off you really were. You're not going to spread the word of how great a buy you got, if anything you'll do the opposite.

If you doubt it's the enthusiasts which made AMD what they are today, then i suppose you can go with your flagship theory.

If AMD is in for 2nd place to Conroe like things seem most likely to be, are they to just give up? Are they to go lock themselves into a bathroom until someone magically craps out new CPU that is faster than Conroe clock for clock?

Good job! Strawman's statements are a great way to argue! :thumbsup:

The fact is, this actually seems like a plausible idea (mostly thanks to HTT). It would allow AMD to at least brag about "towing more payload" should they lose the race - and just like the FX51, it is possible the technology could trickle down closer to the mainstream.

I suppose we'll see who's right in a few months. Just like when Intel released the EE because that's all they could do at the time, but it didn't help them. Yeah, reviewers liked to use the EEs, but if you were on the interweb at that time, you would know what the word really was in the forums. People weren't telling others who were asking for rig help 'go get a prescott, you should see how awesome the EE does in reviews'.

And I guess you know something that i don't... but what technology in the FX line trickled down to the 'mainstream'?

Anyways, stick around the forum, and when people start recommending A64s over Conroe because of these quad core CPUs, PM incase i miss it, ok?

 

[Viditor]

Originally posted by: Looney
The enthusiasts market is what put AMD where it is today... it's what people talked about, what people bought, and what people told others. But i don't think AMD entirely understands who the enthusiasts are. They're not just the ones popping down $1000 for a CPU, they're the ones that buys $300 Athlon64 and sees how it kicks a P4s ass.

My prediction still stands... it looks like AMD is doing exactly what Intel did with the P4. They couldn't compete in the actual consumer market, so they're creating EE-type CPUs that only 0.1% of people are going to buy.

It may have started it, but I disagree that it's the reason...
AMD's server line is what put it at where it is today (and is it's biggest revenue share).
In fact, while it may not have garnered much attention with enthusiasts (it is rather boring for us), AMD's new business platform has had major design wins with all but one of the top tier OEMs...that fact alone is worth many times more (in income) than losing a speed crown for half a year.
Then there's the vastly underrated Torrenza...IMHO, that will be as important a breakthrough as x86-64 ever was!

You should look very closely again at Anand's piece on the analyst's day...it's much bigger than many of us realize so far.
http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2768&p=1

 

[Looney]

AMD's server line really only took off with the Opteron, which is only 2 years old. And it was popular with enthusiasts... it was so popular in fact that AMD raised the prices on them so that enthusiasts would buy the consumers CPUs and not the Opterons.

If Torrenza really does as well as it claims, then that's significant news. I assume this is part of why there's rumors they're buying ATI. Anyways, this is definitely good news.

 

[n7]

Originally posted by: Looney
The enthusiasts market is what put AMD where it is today... it's what people talked about, what people bought, and what people told others.

I'd have to agree with this statement.

Take the retail market, for example.
I realize server market played a large part, but really, the average consumer doesn't even know what an Opteron is.
The server market had nothing to do with the retail market.

If you didn't have AMD fan sales guys at Best Buy etc., customers would still only be buying Intel, since it's only recently that people have finally accepted AMD as being an equal, & even today, the average user is nearly always more comfortable with buying the Intel name, cause they know it better.

AMD owes those who believed in them (namely that enthusiast sales guy who ran an AMD for his games) in huge way.

And btw, i werk in retail sales, so i'd venture to say that i have a pretty good perspective on the average consumer.

I have sold AMD to hundreds, maybe thousands, of disbelievers who w/o me would have gladly spent the extra couple hundred it used to be (& still somewhat is) for the Intel-based system which would not have performed any better.

So don't question the enthusiast's power in the market; it's been enormous IMO.

 

[lopri]

These big corporations are all aware of the influence of enthusiasts, and of course big websites like AT. Towards the end, that's why Intel showcased the Conroe like 6 months ahead of its launch, that's why NV is bringing out stuff like "linkboost", and this AMD's 4x4 thing is also on the same token. But this 4x4 is just too far away from reality that I've known of. Basically it's like a beefed-up/tweaked-up workstation and its suspicious introduction is just too obvious and purposeful. When Anand said "stop-gap", he's actually right. I'm guessing AMD management have been taking some marketing class from NV. Come on, AMD. Bring that Energy Efficient SFF X2s right now! On Socket 939!

 

[TekDemon]

This is really a terrible solution if AMD wants to keep market share...

I mean it's nice that you can theoretically be running 4 cores and even 4 GPUs (apparently this is part of those crazy ATI acquisition rumors?), but uhh...seeing as how just plain 2 GPU SLI systems don't exactly make up some huge portion of the market, and seeing as how no games are really even stressing those systems...I don't know what AMD thinks is going to happen with this.

So yes, they'll keep like, 0.03% of people from buying Conroe but I don't see this as stopping them from losing a big portion of the desktop market.

Even worse they'll now have to fab 4 cores at their higher costs per core...I don't really get this, because if intel decided to pursue a similarly silly strategy they'd just have an even bigger cost advantage. Which is probably why this is being locked to FX cores only, which are high margin enough that it doesn't really matter.

Of course there's also the fact that this could be cutting into their own server market too...

Mostly this looks like a desperate attempt to keep competitive with technologies that they already had. Just take your server tech and put it into the desktop market and proclaim it a gaming solution?
Sad.

 

[lopri]

Regarding the new HTX, excuse me for quoting myself from article comments:

I totally agree that the "direct connect" is the most desirable way but I cannot help but think AMD is somewhat daydreaming. That is, what's showing in the slides seems way ahead of today's "practicality". I mean, we've had this PCI Express which has been strongly pushed by core logic vendors, but so far all we practically have are video cards. I sometimes think all these mobo makers pay more attention to "asthetic" point when they design PCI-E slots so the boards look prettier. (lol) If my understanding is correct, AMD will introduce a new type of slot, HTX, on motherboards. Will other technology/market follow? Or will it just give another chance to graphics card manufacturers to push us to buy new cards? On today's desktop boards, basically everything is "integrated", sans video. I know that a video card has its own core and frame buffer, and transfers data via Hyper Transport, but if a physics card can utilize the HTX, what stops a video card from connecting directly to CPU, without passing the core logic or system memory? I think this will also be closely related to the available bandwidth of HTX per CPU core (or cores), and I can't really think of any add-in board that'll prioritize the bandwidth other than video cards, (OK and the physics cards) even though the HTX will be an open standard. (look at the lazy/lame Creative) A very desirable case would be where storage (hard disks) can take advantage of this "direct" connection but then again there is a such thing called "memory", so my imagination stops there. (maybe solid-state/I-Ram type of storage can make use of the HTX? Then what's the use of memory? Taking care of I/O?) Talking about I/O, I just thought it'd be interesting to see keyboards/mice connect to CPU via HTX. (Sorry I couldn't resist) All in all, like the article says, this roadmap seems just too broad/ambiguous/futuristic. I'm not a CPU engineer so my thinking could be totaly off, though. If so, please enlighten.

And here is AT's impression on this new technology:

Stepping past the enthusiast platform, we have arguably the most exciting announcement of the day: Torrenza. Along with K8L, AMD plans on openly licensing it's (until now proprietary) coherent HyperTransport technology. At first glance, this may not seem exciting, but AMD is throwing in a little twist: HTX slots. These HTX slots will be standard interfaces connected directly to an AMD CPU's HyperTransport link. If both of these links are coherent, the device and the CPU will be able to communicate directly with each other with cache coherency. Because of this, latency can be reduced greatly over other buses as well, enabling hardware vendors to begin to create true coprocessor technology once again

http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=2768&p=2

Basically AMD wans to connect everything via HyperTransport, it seems.

 

[Viditor]

So yes, they'll keep like, 0.03% of people from buying Conroe but I don't see this as stopping them from losing a big portion of the desktop market

Frankly, I don't think they're that worried about losing much market share at all (and I don't blame them). 95% of the market has never heard of a Conroe, nor are they likely to (and frankly they wouldn't really care either).

A64 is now a well established and very successful platform, and it takes a lot of time to reduce the momentum of that...Intel has proven this by having maintained significant sales even after the Prescott disaster.

Conroe will need to prove itself over time (and by time I mean a year or more) in order to turn the tides on A64, and by then we will have K8L, etc...
What is far more damaging to both companies is the huge price drops on the 9xx chips...big price cuts always get EVERYONE's attention!

When I say it hurts both companies, I mean especially Intel...while it's great for consumers, Intel shareholders are going to have it rough this year.

Article

"SEMI FINANCIAL ANALYSTS at Handelsbanken Capital said it is expecting another profit warning from Intel because of poor chip sales.
Handelsbanken claimed Intel "stuffed the channels with chips in February and March, the floor fell out in April, and sales dropped 52% year on year".

The firm said in a note to its clients that Intel processor prices fell 40 per cent year on year in April, with volumes dropping by 21 per cent. That leads Handelsbanken to the view that Intel will deliver another profit warning, guiding Q2 sales to $7.9 billion, versus a consensus figure of $8.4 billion.

Average selling prices fell from $109 in March to $52 in April and were down 40 per cent year on year, said the firm. That had an effect on computer and peripheral logic sales.

The note continued: "Intel has obviously given up on making any money on its current generation of processors and has started a price war with AMD." The Intel (tick: INTC) share price closed at $18.04 on NASDAQ yesterday, a rise of two cents on the day"

 

[mamisano]

Well, if Intel had a robust, low-latency, high bandwidth interconnect like HT then I am sure they would try and implement similar technologies. What this means, if the manufacturers do adopt HTX like they did x86-64, is that it will FORCE Intel to follow. Problem with Intel is that they refuse to adopt an open standard HT interconnect in favor of their CSI which has been delayed and had features removed on several occassions. CSI is still no where to be seen, and by the time it is released this so-called "HT Killer" will be inferior to the future HT variants.

What the HTX connector will allow is a video card to talk directly to the processor and only be 1 hop away from main memory. This is a similar configuration to todays consoles like XBox, etc. Video card makers can produce both PCI-X and HTX based cards like they do today with AGP and PCI-X. The HTX will not be limited to video card, as it can be populated with specialty co-processors for applications like Physics processing, Rendering, etc depending upon the need of the system.

This might all seem like a stop-gap solution for Conroe, but I see it as the first step in a long-term plan from AMD to introduce new and exciting technologies.

By the way, what has Intel brought us lately to get excited about? I know the new Core architecture is going to be a great processor, but it seems to me like Intel has been concentrating all their resources on only one aspect of a system. AMD has been the one leading the way and pushing Intel to utilize newer technologies.

 

[trivik12]

I think there is a big difference between intel and amd. Intel can launch a marketing campaign saying that core 2 duo cpu's are best things since sliced bread. I am also sure Dell/Apple would launch major ad s saying they have the fastest desktop in the world. that would impact retail segment where AMD has been the strongest.

 

[TekDemon]

Originally posted by: Viditor
A64 is now a well established and very successful platform, and it takes a lot of time to reduce the momentum of that...Intel has proven this by having maintained significant sales even after the Prescott disaster.

Conroe will need to prove itself over time (and by time I mean a year or more) in order to turn the tides on A64, and by then we will have K8L, etc...
What is far more damaging to both companies is the huge price drops on the 9xx chips...big price cuts always get EVERYONE's attention!

When I say it hurts both companies, I mean especially Intel...while it's great for consumers, Intel shareholders are going to have it rough this year.

Well, the thing is that Intel still lost a lot of market share-especially the more profitable market shares because they ended up selling the lower performing chips with low margins. AMD might not lose the market massively in terms of percentage, but what's going to happen is that anybody considering anything medium to high end is going to pick up some PC magazine or go on the internet, and read that Core2 is the fastest out there for their budget (seeing as how most people will not be looking at a budget high enough for 4 cores). Then they'll go and buy a Core2 system because PC Magazine told them to. And this is going to be where the bulk of the profits are going to come from.

And while Intel shareholders are going to have it rough for a bit longer, if you look at AMD's stock price performance over the last 3 months it's not exactly like AMD shareholders aren't crying along with Intel shareholders lol.

One last thing too-Intel managed to keep market share decently in the low-medium end because Intel has a better cost structure than AMD-it costs them less money to make those chips, so Intel just went price war on AMD and managed to keep a healthy stake in the low-medium end.

The thing is though, that since Intel is now back on top in performance they're on the warpath-so they're launching Core super cheap-which is going to push their own P4's down the chain even cheaper. Intel can afford this since their 65nm fabs mean chips are cheaper to make. Plus they have tons of money in the bank for a rainy day. AMD can't use the same strategies that Intel used to hold onto market share-they can't just afford to go nuts with the price warring-they're not as big a company, and their chips cost more to make.

AMD's not intel, so you can't really assume that just because intel held on that AMD is equally resilient. Intel also hedged their bet on P4 by continuing to develop the p3 core on the sidelines-in case the P4 plan didn't go so well. And this in turn helped them grow the mobile market by leaps and bounds and provided a solution to strike back at AMD. But AMD doesn't really have the same kind of resources intel does, so whether or not they're going to be able to weather the storm is definitely unclear.

 

[F1shF4t]

Originally posted by: TekDemon

Originally posted by: Viditor
A64 is now a well established and very successful platform, and it takes a lot of time to reduce the momentum of that...Intel has proven this by having maintained significant sales even after the Prescott disaster.

Conroe will need to prove itself over time (and by time I mean a year or more) in order to turn the tides on A64, and by then we will have K8L, etc...
What is far more damaging to both companies is the huge price drops on the 9xx chips...big price cuts always get EVERYONE's attention!

When I say it hurts both companies, I mean especially Intel...while it's great for consumers, Intel shareholders are going to have it rough this year.

Well, the thing is that Intel still lost a lot of market share-especially the more profitable market shares because they ended up selling the lower performing chips with low margins. AMD might not lose the market massively in terms of percentage, but what's going to happen is that anybody considering anything medium to high end is going to pick up some PC magazine or go on the internet, and read that Core2 is the fastest out there for their budget (seeing as how most people will not be looking at a budget high enough for 4 cores). Then they'll go and buy a Core2 system because PC Magazine told them to. And this is going to be where the bulk of the profits are going to come from.

And while Intel shareholders are going to have it rough for a bit longer, if you look at AMD's stock price performance over the last 3 months it's not exactly like AMD shareholders aren't crying along with Intel shareholders lol.

One last thing too-Intel managed to keep market share decently in the low-medium end because Intel has a better cost structure than AMD-it costs them less money to make those chips, so Intel just went price war on AMD and managed to keep a healthy stake in the low-medium end.

The thing is though, that since Intel is now back on top in performance they're on the warpath-so they're launching Core super cheap-which is going to push their own P4's down the chain even cheaper. Intel can afford this since their 65nm fabs mean chips are cheaper to make. Plus they have tons of money in the bank for a rainy day. AMD can't use the same strategies that Intel used to hold onto market share-they can't just afford to go nuts with the price warring-they're not as big a company, and their chips cost more to make.

AMD's not intel, so you can't really assume that just because intel held on that AMD is equally resilient. Intel also hedged their bet on P4 by continuing to develop the p3 core on the sidelines-in case the P4 plan didn't go so well. And this in turn helped them grow the mobile market by leaps and bounds and provided a solution to strike back at AMD. But AMD doesn't really have the same kind of resources intel does, so whether or not they're going to be able to weather the storm is definitely unclear.

Thing is u assume that people will read about comps, a lot of people buy semprons and celerons, and if some people i know is an indication they dont care and dont know whats inside their comp as long as it does what they want. Also majority of sales are not in the mid to high end, its in the low end. Here in australia i see a lot more semprons being sold in shops than celerons. And even people buying mid to high end, some of them are clueless.

Sure they can advertise that conroe is the best thing to sliced bread, but isnt thats what they already were advertising with ALL their prescots, PD etc.

Having the fastest cpu is not everything, what amd is gonna do is release a new platform which will allow to do some nifty things in the future.