Sign in with
Sign up | Sign in
Your question

Future predictions in CPU wars. Only 2007 predictions

Last response: in CPUs
Share
December 11, 2006 1:50:37 AM

Here is mine:

AMD is first to market with a 1 die solution for Quad Core, but it is a paper launch.

Intel launches its 45nm 1 die quad core within weeks of AMD's launch and it out performs AMD's processors. Intel shortly after releases a 2 die 8 core processor to combat the 4X4 and then just before the end of 2007 annouces it will soon release an 8 core 1 die solution that puts AMD so far behind that Intel stock reaches $40+/share and AMD stock slumps to $13/share range.

In late 2007 Sandman posts "AMD is now slightly behind Intel, but there processors outperform Intels in all MS Word benchmarks so I will still stick with AMD." The post is secretly typed on an Intel Penncrest system running word. :lol: 

The Moderators should sticky this Thread and lock it on NewYears so people can revisit peoples predictions throughout 2007 and can see who got what right. :wink:
December 11, 2006 7:31:51 AM

Quote:
Here is mine:

AMD is first to market with a 1 die solution for Quad Core, but it is a paper launch.

Intel launches its 45nm 1 die quad core within weeks of AMD's launch and it out performs AMD's processors. Intel shortly after releases a 2 die 8 core processor to combat the 4X4 and then just before the end of 2007 annouces it will soon release an 8 core 1 die solution that puts AMD so far behind that Intel stock reaches $40+/share and AMD stock slumps to $13/share range.

In late 2007 Sandman posts "AMD is now slightly behind Intel, but there processors outperform Intels in all MS Word benchmarks so I will still stick with AMD." The post is secretly typed on an Intel Penncrest system running word. :lol: 

The Moderators should sticky this Thread and lock it on NewYears so people can revisit peoples predictions throughout 2007 and can see who got what right. :wink:



well its rather simple, AMD has been fucking up alot lately. i think at best AMD's quad core will match the core 2 quadro, and be able to beat it at power efficiency (a natural result of the 4 cores being fused on 1 die)

torrenza while novel in its socket implantation, it has been done before in various ways (DVD accelerators for video cards, sounds cards that take over the brunt of audio decoding and processing tasks from the CPU, even the GPU is like this in that its an accelerator for graphics.... you know they once called them that too) :lol:  torrenza will mostly make it easy for workstations, and the average joe to use the cell processor.

AMD's Fusion will probably only lead to more power efficient laptops, and pave the way for more multi core dies with cores that are physically different.

AMD seems to be pushing for a less expensive upgrade route with DDR standards, one thing i agree with alot.

4x4 or quadFX is bust, it might (and i stress might) put it above intel when you can put 2 quad cores in it. but that will be something that few would indulge in.

what AMD should be doing, is redesigning their K8 core almost from the ground up, because short of making it scale higher in frequency (and all K8's basically hit the wall at 3.2ghz) it cannot compete with core 2 in any way but price performance ratio.

otherwise AMD is getting their ass handed to them by intel

intel if they are smart will press their advantage and make a native quad core as to not let AMD get the one up on them in any way. they will also have to match torrenza, simply because the cell is to good to just let AMD get all the money in it.

i think what will happen is AMD will return to what its good at (the enterprise sector), make a ton of money off of torrenza, and do a complete core redesign, and then let Intel really have it in a couple years, and then the consumer sector will reap the benefits of this (kinda like how intel spent all that time getting its butt kicked with netburst, then finally released something awesome). in the mean time, intel will make alot of money, like it always does, and AMD will be the underdog that peeps love to root for, whether or not it really is better then intel :wink: .
December 11, 2006 8:05:29 AM

December 2007.

After the fallout from the DOJ investigations and the precipitous drop in share price from the cancellation of the Dell supply contract, AMD was purchased at firesale price by Intel.

:twisted:
Related resources
December 11, 2006 8:11:52 AM

Quote:
December 2007.

After the fallout from the DOJ investigations and the precipitous drop in share price from the cancellation of the Dell supply contract, AMD was purchased at firesale price by Intel.

:twisted:


intel wouldn't dare do that, because the DOJ would be on their tales for anti trust violations in a heartbeat. losing the contract to dell wouldn't hurt AMD much at all, because most of their money comes from the enterprise sector, intel is the one that needs to be in the desktop market to survive, if all goes to hell then AMD will just go back to selling opterons and drop their whole desktop line. and we wont see them for a couple years.
December 11, 2006 8:29:38 AM

I don't think you'll see a multisocket desktop or more than 4-cores from Intel until 2008-09. Intel plans on CSI (their version of HyperTransport) and an IMC in 2008. I think going more than 4 cores will finally start stressing the fsb.

In 2007, I think Intel will push up the clockspeeds (they have plenty of room to push), and release their new "value" chips such as the E4300.

Intel will also be pushing out new mobo's such as the Bearlake - and moving to DDR3 and PCIe 2.0.

AMD will release Barcelona and socket AM2+. Barcelona will be a server (and 4x4) chip - I don't know if they will be able to bring the technology fully to the desktop by the end of 2007. I think they will increase clockspeed for their desktop chips by a little once they're fully at 65nm, but I'm not sure how much play their current µarch is going to give them.

In 2007, I think the big news for developments will be with Nvidia and ATI. They have already announced megapower GPU updates, and hopefully after Vista is released with DX10, they will focus on optimizing their CPUs with better thermal/power levels.
December 11, 2006 8:56:35 AM

Quote:


intel wouldn't dare do that, because the DOJ would be on their tales for anti trust violations in a heartbeat. losing the contract to dell wouldn't hurt AMD much at all, because most of their money comes from the enterprise sector, intel is the one that needs to be in the desktop market to survive, if all goes to hell then AMD will just go back to selling opterons and drop their whole desktop line. and we wont see them for a couple years.


Ok, I'll change my prediction.

December 2007.

After the fallout from the DOJ investigations and the precipitous drop in share price from the cancellation of the Dell supply contract, AMD was purchased at firesale price by Microsoft!

8)
December 11, 2006 9:48:06 AM

Well, I wouldn't count on K8L not scaling. It might not be as much of a rework as Core is of the P3. But, once, a while ago, the P3 wouldn't scale, so then came the P4. So, K8L may, or may not scale, can't base the judgement on current scaling of K8. But, you guys just keep on trying to stir crap up, because anyone with common sense, knows that AMD is about to release a redisgn of K8, and is not going to just get bought out after years of barley hanging on.

wes
December 11, 2006 9:52:19 AM

Quote:
But, you guys just keep on trying to stir crap up, because anyone with common sense, knows that AMD is about to release a redisgn of K8, and is not going to just get bought out after years of barley hanging on.


Why do we stir crap up? Well, mostly 'cuz it's fun! What's the point of coming onto a forum like this if you can't throw a Moulinex into the manure pile every once in a while? :D 

I think you have identified it. That is the problem right there. If AMD had been hanging onto CPU development and superb marketing instead of barley, they'd be in much better shape.

JUST KIDDIN' WES! DON'T RIP MY HEAD OFF!!! :lol: 
December 11, 2006 10:02:38 AM

I could tell what you guys were doing. It was obvious, just don't want the idiots coming in here and ruining what could be some interesting predictions.
Not going to rip your head off, normally don't react to people like that in here.
I am just interested in some legit ideas, and legitamate rumors of what is to come. Thats all. Don't worry, I try not to be a fanatic, and normally try not to provoke them.

wes

Edit: we are the only two people posting, toms needs a chat room I guess.
December 11, 2006 11:09:23 AM

Quote:
I could tell what you guys were doing. It was obvious, just don't want the idiots coming in here and ruining what could be some interesting predictions.
Not going to rip your head off, normally don't react to people like that in here.
I am just interested in some legit ideas, and legitamate rumors of what is to come. Thats all. Don't worry, I try not to be a fanatic, and normally try not to provoke them.

wes

Edit: we are the only two people posting, toms needs a chat room I guess.


I guess it's still pretty early in North America... they'll all rise and shine soon, I'm sure! As for you ripping my head off it was about your "barley" typo, I was just having a bit of fun at your expense. :D 

In all honesty, I see Dec. '07 not being that different from Dec. '06. I can see the middle part of the market being migrated to dual cores to an even greater degree but I would be highly surprised if there is any platform that can significantly exceed QX6700 performance. That may likely come in '08. Just my 2 cents!
December 11, 2006 11:18:42 AM

personally I don't do the I see in my crystal ball thing so being the patient dude that I am, I'll wait and see.
December 11, 2006 11:22:49 AM

Quote:
personally I don't do the I see in my crystal ball thing so being the patient dude that I am, I'll wait and see.


Hell, I'd be damn happy just to be sure that I would be alive in Dec. 07! :lol: 
December 11, 2006 11:37:43 AM

Haha,

Didn't even catch my typo, good one though.

As far as QX6700, if K8L can compete with Core 2, then AMD quad will compete with the Intel Quads. At least it seems logical that they would. But who knows, K8L might experience major issues, and it might not perform like we hope.

wes
a b à CPUs
December 11, 2006 11:42:33 AM

Quote:
I could tell what you guys were doing. It was obvious, just don't want the idiots coming in here and ruining what could be some interesting predictions...I am just interested in some legit ideas, and legitamate rumors of what is to come. Thats all.


I'm with you in that I would be interested in some informed speculations as well as some intelligent discussion about processor developments in 2007. But unfortunately, given the bias that permeates these forums, with the OP's message being a perfect example, I do not have any expectations of anything productive coming out of this thread.

+1 post for me
December 11, 2006 12:25:54 PM

Yeah, wes, the typo was a good one! :lol: 

K8L seems to be held as the Holy Grail by "some" and I'm really wondering whether the whole thing is just gonna be lunchbag letdown when it finally arrives. I remember being really excited about Windsor 5000+ and by the time it finally shipped it was as thrilling as a peanut butter sandwich. Then everything got shifted onto 4x4. "Wait until you see this..." and it was somewhat of a damp squib. QuadFX is a disappointment pretty well across the board. A multifarous kludge of brobdinagian complexity designed for a severely circumscribed market.

Chunky, I may not be able to provide intelligent discussion but who else do you know that can put multifarous, brobdinagian, and circumscribed into the same sentence? :twisted:
December 11, 2006 12:37:23 PM

AMD will be down to 5% again. Unless Intels HQ explodes, theres no hope for AMD cause it looks like Intel has taken the story very seriously this time.
December 11, 2006 12:41:46 PM

I understand what you are saying, but, I really wasn't to excited by 4x4, and didn't really expect the performance difference to be any better than X2 to Core 2 Duo. I mean, you are putting two X2's against 2 Core 2's. How would it be better all of the sudden? I don't know, maybe I missed something.

Anyway, just from looking at the arch. changes, the common theme seems to be that K8L should either match, or beat core 2. That is, if all is at it seems, and if nothing get's botched.

wes
December 11, 2006 12:53:56 PM

My prediction, based on a bit of truth, a bit of history, and a shiat-ton of speculation is that AMD will start focusing more and more on the enterprise/server market and maybe low-end consumer. Their fusion might dominate mobile technologies for a while, but Intel will most likely strike back.

Either way, AMD has the upper-hand in servers right now, so if they keep up the work there, then that will last them a long time. Their consumer market will start to see serious problems becase of the C2D, but if they downplay their consumer market in general, or focus on the low-end where performance doesn't matter as much as cost, they might squeak by. Mobile wise, Turion still can't compare to Intel's Mobile assets, although merging with ATI will help them a lot in this category. Fusion might turn the tables around depending on when it comes out and how well it does, but that remains to be seen.
December 11, 2006 1:02:55 PM

Quote:
AMD will be down to 5% again. Unless Intels HQ explodes, theres no hope for AMD cause it looks like Intel has taken the story very seriously this time.


Yeah, it's kinda like when everybody and his brother were writing off Apple as a lunatic fringe computer company and then Jobs came back and kicked some serious bollocks. Intel has taken it on the chin for a very long time in the enthusiast market, but if we back off and get some perspective, C2D/C2Q is some incredibly impressive technology that makes its impact where it counts: In sheer performance. And I repeat, I've been an AMD guy for many years and over a dozen systems! But there is nothing coming out of AMD, even the paper 65nm, that is going to knock Intel off the throne right now anyway.

Quote:
I understand what you are saying, but, I really wasn't to excited by 4x4, and didn't really expect the performance difference to be any better than X2 to Core 2 Duo. I mean, you are putting two X2's against 2 Core 2's. How would it be better all of the sudden? I don't know, maybe I missed something.

Anyway, just from looking at the arch. changes, the common theme seems to be that K8L should either match, or beat core 2. That is, if all is at it seems, and if nothing get's botched.

wes


K8L matching C2D is going to be IMHO a day late and a dollar short. It will be just as hohum if they come in at 5% or 10% over in the price/performance. Let's face it. C2D trounces AMD topend right now. AMD needs a dealt royal flush to kick Intel's a$$ and I just don't see any on the horizon. I may be wrong and in a way I hope I am so that all the years I spent loyal to AMD would be worth something in the future.

Again IMHO, Quad FX is only going to be considered king of the heap once I can plug in 2 x quadcore FXs in the slots. Now we're cooking with gas. But if you try to put C2Q up against Quad FX dualcore in the vast majority of "enthusiast" utilizations, it's like propping up Joan Rivers in a bikini against Pam Anderson. Which one you gonna go for? :D 
December 11, 2006 1:07:46 PM

Quote:


...

Chunky, I may not be able to provide intelligent discussion but who else do you know that can put multifarous, brobdinagian, and circumscribed into the same sentence? :twisted:


He has a good point, I would have never thought of doing that...:) 
December 11, 2006 1:14:33 PM

Quote:


...

Chunky, I may not be able to provide intelligent discussion but who else do you know that can put multifarous, brobdinagian, and circumscribed into the same sentence? :twisted:


He has a good point, I would have never thought of doing that...:) 

:trophy: :lol: 
a c 99 à CPUs
December 11, 2006 1:20:47 PM

Here is mine:

1. AMD 65nm chips become widespread just after New Year's. They overclock decently so that the low-end Athlon 64 X2s become roughly as popular as the Core 2 Duo E6300 as it can OC to rival the E6300's OCing performance but is less expensive.

2. AMD also releases 65nm Opterons. These will have a 65W TDP for the normal units (up to 3.0 GHz,) a 35-45W TDP for the HE units, and there will be a 3.2 GHz SE unit with about an 80W TDP. AMD will sell quite a few of these as they will run cooler than a Woodcrest/Tulsa platform and not take hot, expensive, slow FB-DIMMs.

3. RAM prices creep slowly back down to roughly $85-90/GB for midrange RAM by the middle of the year. Not quite down to the early 2006 levels of $70-75/GB, but more reasonable than $110-120+/GB.

4. NVIDIA will finally release their 650i chipset and sells a bundle of them. Intel is forced to release a new performance chipset to replace 975X. It will support a 1333 MHz FSB.

5. Intel hurries up the switch of the socket 771 Xeons from FB-DIMMs to buffered DDR3 to battle the Opteron DP and MPs.

6. Intel boosts the Core 2 Duo Extreme's clock speed to 3.2 GHz (X6900) and relabels the X6800 as the E6800. Drops prices slightly on the Core 2 Duo line.

7. Intel also releases the Core 2 Quadro Q6600/1066 FSB and downlabels the QX6700 as the Q6700, drops the price a little. They replace the QX6700 with the QX6800, a 3.00 GHz on a 1333 FSB and it runs on the new Intel 1333 FSB chipset. The chip performs well but has a 160W TDP and runs very hot.

8. AMD releases 65nm QuadFX chips, the FX-80 (2.8 GHz) FX-82 (3.0 GHz) and FX-84 (3.2 GHz.) These run a lot cooler than the FX-70s and with Vista supporting NUMA, the 3.2 GHz FX-84s slightly beat the QX6800 overall and consume only a little more power than the QX6800 unit. The QuadFX also runs on an AMD/ATi chipset now.

9. Intel ships some Xeon Clovertown 5300 series CPUs but most people go with the Woodcrests or Opterons because FSB choking is limiting scaling on some server-type apps. Intel responds by releasing the DDR3 Xeon chipset with dual 1600 MHz busses and the monolithic-die Xeon 5400s with 8MB L2 + 8MB L3 in one fell swoop. This improves performance significantly and Intel sells quite a few.

10. AMD releases the K8L arch. The first chips released are the Barcelona and Shanghai Opterons. These have HT 3.0 and support registered DDR2-1066 or DDR3-1333. They outperform the Xeon 5400s by a significant margin. The Agena desktop chips come next and outperform the Kentsfields by 10-20% clock for clock in most cases and run much cooler.

11. Intel starts to make 45nm quad-core chips on 1600 MHz FSBs. There is a new desktop chipset that has a 1600 MHz FSB to support them as well as DDR3-1333. These chips are released at up to 3.6 GHz on the desktop and use their clock speed advantage over the ~3.0 GHz top Agena to perform roughly similar on non-memory-hogging applications. They lose to the Agena on memory-bound applications. The 45nm Intel server chips run cooler than the 65nm QCs but still suffer from FSB choking at higher speeds and heavier workloads.

12. Intel states plans to abandon the FSB and introduce a chip with an IMC and in the case of the Xeons, they adopt CSI but it suspiciously looks a lot more like HT 3.0 than the original CSI. They make plans for the new chips but they won't ship until midway through 2008.
December 11, 2006 1:38:04 PM

Quote:
AMD will be down to 5% again. Unless Intels HQ explodes, theres no hope for AMD cause it looks like Intel has taken the story very seriously this time.


There is little chance for AMD to get 5% market share only unless it executed really badly.
December 11, 2006 2:20:00 PM

My prediction...

Well, look at the last 20 years of development...

For 2007 I see more of same. There will be successes and failures from both camps. Research will march on, lessons will be learned and, ultimately, we the computer tinkerers will benefit in a big way.

Ya gotta love progress even when there are potholes in the road! 8)
December 11, 2006 2:27:16 PM

Quote:
Here is mine:

2. AMD also releases 65nm Opterons. These will have a 65W TDP for the normal units (up to 3.0 GHz,) a 35-45W TDP for the HE units, and there will be a 3.2 GHz SE unit with about an 80W TDP. AMD will sell quite a few of these as they will run cooler than a Woodcrest/Tulsa platform and not take hot, expensive, slow FB-DIMMs.

6. Intel boosts the Core 2 Duo Extreme's clock speed to 3.2 GHz (X6900) and relabels the X6800 as the E6800. Drops prices slightly on the Core 2 Duo line.

8. AMD releases 65nm QuadFX chips, the FX-80 (2.8 GHz) FX-82 (3.0 GHz) and FX-84 (3.2 GHz.) These run a lot cooler than the FX-70s and with Vista supporting NUMA, the 3.2 GHz FX-84s slightly beat the QX6800 overall and consume only a little more power than the QX6800 unit. The QuadFX also runs on an AMD/ATi chipset now.

9. Intel ships some Xeon Clovertown 5300 series CPUs but most people go with the Woodcrests or Opterons because FSB choking is limiting scaling on some server-type apps. Intel responds by releasing the DDR3 Xeon chipset with dual 1600 MHz busses and the monolithic-die Xeon 5400s with 8MB L2 + 8MB L3 in one fell swoop. This improves performance significantly and Intel sells quite a few.

10. AMD releases the K8L arch. The first chips released are the Barcelona and Shanghai Opterons. These have HT 3.0 and support registered DDR2-1066 or DDR3-1333. They outperform the Xeon 5400s by a significant margin. The Agena desktop chips come next and outperform the Kentsfields by 10-20% clock for clock in most cases and run much cooler.


2. Opteron lines are the last to be converted to 65nm

6. E6800 may be cancelled, according to the author of HKEPC

8. Since A64 FX 4x4 are renamed Opterons, 65nm versions will be announced late in 2007.

9. It seems that the FSB is not the limiting factor for Core Architecture NOW.

10. Agree with you :wink:
December 11, 2006 5:16:39 PM

Quote:


intel wouldn't dare do that, because the DOJ would be on their tales for anti trust violations in a heartbeat. losing the contract to dell wouldn't hurt AMD much at all, because most of their money comes from the enterprise sector, intel is the one that needs to be in the desktop market to survive, if all goes to hell then AMD will just go back to selling opterons and drop their whole desktop line. and we wont see them for a couple years.


Ok, I'll change my prediction.

December 2007.

After the fallout from the DOJ investigations and the precipitous drop in share price from the cancellation of the Dell supply contract, AMD was purchased at firesale price by Microsoft!

8)

:lol:  haha

do you suppose that AMD will sell ATI to Nvidia to cut their losses?
December 11, 2006 5:32:43 PM

Quote:
Well, I wouldn't count on K8L not scaling. It might not be as much of a rework as Core is of the P3. But, once, a while ago, the P3 wouldn't scale, so then came the P4. So, K8L may, or may not scale, can't base the judgement on current scaling of K8. But, you guys just keep on trying to stir crap up, because anyone with common sense, knows that AMD is about to release a redisgn of K8, and is not going to just get bought out after years of barley hanging on.

wes


im sry i have to correct you, the Pentium 4 is far far far far from anything like the P3. the Pentium 4 is completely different, actually if you want to trace the P3 architecture it it something like this

1 Pentium pro (mostly original design, not related to the original Pentium in many ways at all)

2 becomes the Pentium 2 (core shrink, some redesigns and instruction sets)

3 pentium 3 (the PII redesigned a bit more, but mostly just an added instruction set)

4 the Pentium M ( basically a higher quality, die shrunk added instruction set, slightly improved architecture PIII)

5 core architecture (is an improved pentium M, again added instruction sets, die shrinks, and such)

5 core 2 duo (takes some of the features of the PIII (almost directly from that line actually) completely redone prefetching system, lots of mobile technology (from the Pentium M) used to reduce power consumption and heat, designed to produce high overall IPC's and the easy frequency scalling is actually more to do with the fact that its a mobile processor pumped up for desktop use)

now the pentium 4 was completely original, was based around netburst and long (eventually 31 stages) pipelines. as a very low overall IPC average, and produces alot of heat.
a c 99 à CPUs
December 11, 2006 5:34:24 PM

I'd more predict for MS to go under than AMD to. MS is a house of cards, only propped up by vendor and file format lock-in. We're already starting to see it falter with the big Vista/Office 2007 launch making a deafening silence in the corporate world, which is not only MS's biggest market but its key to survival. Throw in a lot of high-profile companies and groups switching to other office suites and operating systems and you'll see the house of cards come a-crashing down. They really can't do much else to save themselves as the Zune is a joke and so are all things MSN. They have the Xbox...woo. MS is also so big that to be able to change what they're doing would take an enormous amount of time and be more or less impossible to do. AMD on the other hand does not depend any sort of strangle-hold on the market to keep afloat. They are small and could very easily switch to doing other things in the semiconductor market if they got pushed out of the desktop CPU market. They could only make server chips or embedded CPUs or whatnot. They'd do very well at that.
December 11, 2006 5:44:18 PM

Quote:
I'd more predict for MS to go under than AMD to. MS is a house of cards, only propped up by vendor and file format lock-in. We're already starting to see it falter with the big Vista/Office 2007 launch making a deafening silence in the corporate world, which is not only MS's biggest market but its key to survival. Throw in a lot of high-profile companies and groups switching to other office suites and operating systems and you'll see the house of cards come a-crashing down. They really can't do much else to save themselves as the Zune is a joke and so are all things MSN. They have the Xbox...woo. MS is also so big that to be able to change what they're doing would take an enormous amount of time and be more or less impossible to do. AMD on the other hand does not depend any sort of strangle-hold on the market to keep afloat. They are small and could very easily switch to doing other things in the semiconductor market if they got pushed out of the desktop CPU market. They could only make server chips or embedded CPUs or whatnot. They'd do very well at that.


i agree with this, Microsoft is again just barely keeping up with the innovation of others. IE 7 is a joke compared to Firefox 2, the Ipod and Zune ect. anyone that perpetuates that windows is a better platform (for any other reason other then its large install base, which in itself is a poor reason, not to mention the whole homogenius effect) for anything over Unix or Linux is either ignorant or stupid, and you know what they say about that :wink:. i dont think they will fall in 2007, but the insane prices for vista, and the growing market of even average joe building his own computer, coupled with what will probably be another lackluster OS (im willing to bet there will be a security update the day it is officially released), is not helping them with their monopoly.
December 11, 2006 6:34:21 PM

What most in this discussion fail to realize so far is that AMD may just continue to Gain market share in the consumer space (and overall) because they now have product in every channel. AMD has never had this before. Even though they do seem to have what I think is temporary supply constraints because of the switch to 65nm.

Also, unless consumer behavior changes all of a sudden, the price first mentality will still win most of the time in the lower to mid-range consumer space. And AMD systems that perform great for most people are available with all the brand names for slightly less than it's Intel counterpart - when all you can spend is $---.00 + tax - that's all one can spend.

And on the enterprise side AMD has garnered much mindshare over the last 3+ years and now that they have market share to back up the mindshare in the server space - so they are in the door, and remember, AMD chips already had very acceptable power envelopes so even though C2D has better, the Corporate shops that have started using AMD are not pressed on that side either. The server market is more platform specific, so performance is high on the list but is not the final decision maker. If it were AMD would have captured as much as 50% of the server market since Opteron was intro'd.

What I am hearing are predictions from an enthusiasts perspective. So those are some of my predictions above and here are mine as they relate to AMD Intel;

AMD will make a good comeback with K8L
AMD's 65nm process will allow some additional OC headroom
Intel will maintain the overall desktop performance crown despite 65nm
The K8L will win back overall server performance and will still be competitive even after Intel moves to 45nm tech
The Fusion project will yield announcements only in 2007, but the tech will be formidable
Torenza product announcements will start
Stream processing on ATI video cards will take off
AMD stock will rise by 30% in 2007
Intel won't lose much of its overall market share but AMD will make inroads in laptops and further inroads in the corporate desktop market

These are just predictions for this little corner of the IT world.

Well we will see - I've never edited a post but I'll add more if I think of any.
December 12, 2006 12:53:19 AM

Well,

I never said it was like the P3. All I said was the P3 did not scale very high, so then they released the P4. The P3 scaled less than a ghz, from 450 to 1.4, the P4, due to it's different architecture, scaled from 1.3 to 3.8. I can understand how it was confusing, I should have left the P4 out, I was backing up the P3 not scaling, by Intel releasing the P4. All I meant, is they couldn't get the P3 to scale, now, with Core, they basically have goten it to scale. My own post confused me a bit, so sorry for the confusing post.

wes
December 12, 2006 1:58:44 AM

Quote:
Well,

I never said it was like the P3. All I said was the P3 did not scale very high, so then they released the P4. The P3 scaled less than a ghz, from 450 to 1.4, the P4, due to it's different architecture, scaled from 1.3 to 3.8. I can understand how it was confusing, I should have left the P4 out, I was backing up the P3 not scaling, by Intel releasing the P4. All I meant, is they couldn't get the P3 to scale, now, with Core, they basically have goten it to scale. My own post confused me a bit, so sorry for the confusing post.

wes



its ok, it just wasn't that good of an example, the k8L is no a major redesign, and will probably be like one of those stages from the Pentium pro to the core 2. the athlon has been that way since it came out, and hasn't had a total core redesign in quiet awhile, but has had alot of little improvements over time.
December 12, 2006 11:24:53 AM

Well, I would not call this a simple revision. This is much more substantial than what they have done going from say Sledgehammer all the way to Windsor.

Yeah, Core microarch. is a ground up design, but that design is based around the P3.

AMD might have called this K9, but for obvious reasons they did not. Anyway, this is a redesign of the core, that is much more substantial of than anything they have done with K8 so far. It isn't ground up, but it might not need to be. K8 is obviously good, and they might be able to get even more performance out of it as well as make it scale substantially higher. Who knows?

wes
December 12, 2006 11:58:44 AM

Quote:
But, you guys just keep on trying to stir crap up, because anyone with common sense, knows that AMD is about to release a redisgn of K8, and is not going to just get bought out after years of barley hanging on.


Why do we stir crap up? Well, mostly 'cuz it's fun! What's the point of coming onto a forum like this if you can't throw a Moulinex into the manure pile every once in a while? :D 

I think you have identified it. That is the problem right there. If AMD had been hanging onto CPU development and superb marketing instead of barley, they'd be in much better shape.

JUST KIDDIN' WES! DON'T RIP MY HEAD OFF!!! :lol: 


SO why when AMD buyers do this do you get all righteously indignant?
December 12, 2006 12:50:08 PM

Here's my prediction:

Chuck Norris sells off bits of his muscle to AMD and Intel, and both chip companies make a ton of superpowered processors from the muscle provided, creating a sort of "Golden Age" for computing, where everything is clocked at 300Ghz yet only consumes 10 watts of power. SuperPi calculations have officially reached 0 seconds by now [/joke]

Well, enough with the messing around... here's a more serious prediction:

1. AMD releases K8L with performance about on par with C2Q, though will run much cooler. Enterprise market suck up initial supply, as the K8L provides the necessary scaling and power efficiency which servers require for optimum performance.
2. Intel releases 45nm early in order to compete with K8L, providing a small performance boost to give it a small edge over K8L, but because of early release, supply is extremely limited and prices much higher than K8L's. The 45nm chips run as cool as the K8L chips, though they still don't scale as well.
3. ATI releases R600 just on time as the retail version of Vista is released into the desktop market, and proves to be just as powerful as Nvidia's G80 cards, however, because of a smaller die manufacturing process, R600 is quite a bit less power hungry and runs significantly cooler. Soon after the first high-end models are released, ATI follows through with mainstream versions of the R600, and ATI profits from the quick uptake on the graphics cards.
4. Intel, despite a solid comeback from AMD, will still hold the majority of the market share, but because of increasing pressure from AMD, will suffer more hiccups similar to their Employee Lay-offs.
5. AMD will struggle to meet demand with the supply of K8L and Agena chips, and will only really be able to finally keep up until maybe Xmas 2007.
6.The Enterprise market welcomes the Stream processing solution from AMD/ATI, providing a more competitive market for processing accelerators. Even though demand is small to start off with, AMD/ATI will profit from this greatly.

Can't think of anything else I wish to add... so I'll leave this as my two cents.
December 12, 2006 1:19:18 PM

My prediction.

There is nothing really interesting from AMD until K8L is released in the middle of the year. It catches up with Intel in the quad-core performance vs. Hz, sporting slightly lower energy usage overall. The K8L is an improvement but not enough to reclaim the crown on the high-end desktop.

Out of respect people stop making stupid infantile AMD jokes, this emboldens the AMD fanboys to new highs who are then promptly crushed(hopefully).


AMD finally gets the 65-nm SOI just right in the last quarter of the year just breaking 3.2 Ghz.

Quad cores fail to take off by the end of the year. Leaving AMD's modest gains a hollow victory.

Meanwhile, Intel transitions to 45-nm production and keeps tremendous pricing pressure on all AMD's offerings. Intel also pushes their high-end Dual cores to 4.0 Ghz. Nothing else really interesting comes from Intel except talk of future technologies.

DDR3 begins to see early adoption in the last quarter.
a c 99 à CPUs
December 12, 2006 1:53:15 PM

You may be right with quad cores not taking off this year- dual cores took a year or so to come down in price and up in performance enough for most people to want to bite. So maybe 2008 is the year of the quad core.

When you say that AMD likely won't break 3.2 GHz on 65 nm, I assume that you're talking about the quad-core CPUs. Those are supposed to go up to about 2.9 GHz at launch, so maybe AMD will offer an Agena FX at 3.2 or so. It might run a touch warm at speed, but much cooler than the Kentsfield, so they very well may give it a go. The process should scale up much farther than 3.2 GHz as AMD has had a little while to work on it by then. They did get 3.0 GHz out of 90nm- more than anybody predicted. If AMD decides to make a run of fast dual-cores and not just midrange dual-cores, there is no doubt that you'll see units going above 3.2 GHz as the process will support it and so will the thermal characteristics. However I doubt that AMD will try to release performance dual-core parts once the quads start shipping in some volume. Once the Athlon X2s shipped in volume, development ceased on fast single-core CPUs. That's why you can't buy a single-core CPU faster than 2.4 GHz on the AM2 socket.

Intel very well might start to sample some 45 nm chips by the end of the year but they're not supposed to ship until 2008. This would keep pricing pressure on AMD if they could accelerate their switch to 45 nm, but Intel would have to sell for less than AMD as they are not due for a core refresh until 2008-2009. The Core 2 arch will have to battle with K8L, and I don't think that Intel will have the upper hand in that battle like it does with the K8. Intel also seems to be jumping on the quad-core bandwagon (heck, they *started* the quad-core bandwagon!) and you'll probably not see significant dual-core development from Intel once the Kentsfields become widely shipping. Just look at the Prescott -> Cedar Mill P4s once the Pentium Ds shipped. There was no increase in clock or anything and you know that they could have made a Cedar Mill 4.0 GHz or faster if they wanted to. The Core 2 Duos tend to be able to reach 3.6 or so GHz overclocked, so there is some headroom left. Not 4.0 GHz worth, but I can easily see Intel making a 3.20 or 3.33 GHz Core 2 Duo X6900. It will run a bit warmer than the other Core 2 Duos, but it should be less than 85-90W, which is not unreasonable at all for a performance part.

I'd predicted that the Kentsfield will not get much above 3.0 GHz on 65 nm as it's already at a TDP of 130W (173W max. dissipation) at 2.67 GHz. That's as hot as any chip they've ever released- 90nm Pentium Ds included. Pushing it to 2.93 or 3.00 GHz would mean its TDP would get to 150-160W (200-213W max. dissipation) and that would probably even be too much for the progenitor of NetBurst to handle.

So the Kentsfield is stuck at 2.67-3.00 GHz and the Agena is at roughly the same clock speeds but will have better IPC and thermal characteristics. So yeah, Intel does not have much interesting for 2007. 2008 ought to be a better year for them with the 45 nm switch and heavy development on the Nehalem micro-architecture and DDR3 adoption.
December 12, 2006 1:59:36 PM

Quote:
We're already starting to see it falter with the big Vista/Office 2007 launch making a deafening silence in the corporate world, which is not only MS's biggest market but its key to survival.


That's way overly dramatic. MS will make a bunch of Vista money from new PC sales at Best Buy, Dell, etc. Big biz is always slow to jump on the upgrade wagon. I know IT admins that are still using Windows 2000 and are in no rush to go to XP, let alone Vista. Once a network is tweaked, it's a good idea to stick with it because lost productivity can be so crippling. What about Vista is tantalizing enough to push a large corp to upgrade, say from XP? Don't say security because corp IT is not likely to use only MS security apps. You think a large corp is going to upgrade just for the bling graphics? Many corp workstations would need significant hardware upgrades just to minimally run Vista.
a c 99 à CPUs
December 12, 2006 2:24:24 PM

That proves my point. MS makes a ton of money from yearly corporate software license subscriptions. Since you gave yet another example of how businesses see nothing in it for themselves to jump on the upgrade bandwagon, quite a few are pondering whether or not to keep paying for annual MS software licenses when W2K becomes unsupported by MS.
December 12, 2006 2:36:48 PM

MU-Eng,

Seems like a pretty logical prediction. It all relies on how well K8L performs.
If it performs well in dual, then obviously single die 4core solution should do the same unless something gets botched.

Once K8L(K9) in my opinion, comes out, it should be a pretty good battle, and it will probably go back and forth with what App which cpu performs better than the competitions at.

Interesting times ahead, just like every year. Last year, we were awaiting Core, now, K8L.....

The interesting thing is, now the Intel fanboy's are trying to say K8L is going to suck, not any better than K8, just like the AMD fanboys were saying Core would suck. Comes with the territory I guess.

I will only speculate K8L, until I see some benchmarks, and one actually running, I won't say for sure that it will be any better, I can only speculate.

wes
December 12, 2006 3:28:27 PM

Quote:
That proves my point. MS makes a ton of money from yearly corporate software license subscriptions. Since you gave yet another example of how businesses see nothing in it for themselves to jump on the upgrade bandwagon, quite a few are pondering whether or not to keep paying for annual MS software licenses when W2K becomes unsupported by MS.


Who knows? Maybe Google will be selling some of those licenses? I'd love to see some real competition in that marketplace.
December 12, 2006 4:00:32 PM

Quote:


intel wouldn't dare do that, because the DOJ would be on their tales for anti trust violations in a heartbeat. losing the contract to dell wouldn't hurt AMD much at all, because most of their money comes from the enterprise sector, intel is the one that needs to be in the desktop market to survive, if all goes to hell then AMD will just go back to selling opterons and drop their whole desktop line. and we wont see them for a couple years.


Ok, I'll change my prediction.

December 2007.

After the fallout from the DOJ investigations and the precipitous drop in share price from the cancellation of the Dell supply contract, AMD was purchased at firesale price by Microsoft!

8)

That sounds very possible.......
December 12, 2006 4:33:56 PM

Before i give my 2 cents i would like to note I prefer AMD over Intel, but i am no stretch a fanboy and conceed AMDs defeat to C2D.

my 2cents:

AM2+ with revisions for HT3.0 and DDR3 support will come out on 65nm and provide some cost/performance ratios that rival most C2Ds, while consuming less power and allowing a nice and smooth transition for AM2 owners.

Roughly around the same time K8L or AM3 or whatever its gonna be called will equal or exceed C2Q peformance but by just enough to brag, Intel shrinks to 45nm and the architectures revive the classic PD Hyperthreading -vs- K8 battle, where in some applications AMD wins and others Intel does.

The new Opterons come out and do what they do best own the server markets, with uHT HT3.0, DDR3 and AMDs drive for open platforms will only make them the preferred chip for servers around the world. Intel brings many new features into there chips to regain lost market, but AMD became the popular girl in high school and made it with the popular server building companies by letting them insert there products in there uHT open platform, and are loved for it.

QuadFX doesn't sell that great, but sells enough not fail, to exspensive till K8L SocketF(1207) desktop chips come out and you can drop in any AMD SocketF(1207) chip in there, and then it becomes a bargain. With 2 quad-cores people struggle to overtask there desktops and find themselves, decoding DVDs, ripping music, rendering 3D graphics in Lightwave 9, while editing movies in Premiere, and loading that 2gb photoshop file, and at the same time flying there Boeing 787 in Flight Simulator X on two 30" LCD screens with Vista Ultimate. And you get annoyed by the neighbors complaing that your computer is causing the neighborhood to black out.

Fusion dominates mobile space, late in 2007 maybe 2008 Intel develops there own Fusion and the mobile space becomes a fierce battle and we reap the benefits of cheaper laptops. But when Intel releases theres AMD announces a desktop version of Fusion in dual core form for the low end and business markets, and announces a successor in quad core form that offers video, audio, on two cores, leaving you with a dual core power. Many other variants of Fusion come out sporting physics, audio, video, and other tyes of processing in different variants, uses dependent.

Meanwhile Intel does what Intel does, waits for AMDs answer to Core. AMD delivers with something that equals or betters Intel, Intel answers back and runs pace for pace with AMD for about a year then drops a whole new architecture and awaits AMDs response to start the war all over again.


speculating is fun

edited for spelling errors
December 12, 2006 5:30:13 PM

My only prediction for 2007 is that AMD will continue to gain market share everywhere except the high-end/enthusiast market. At the low-end, the Semptron is by far the best value, and may even compete well against the upcoming Intel value CPU products. At the server side, it's not the CPU performance alone that makes or breaks the deal. HT alone is reason enough to go with AMD, and they will stay strong in the server segment.

Having the best performing desktop CPUs doesn't make the sale with most people. For example, A little over a year ago I was building my current rig, and I didn't want to spend the CAN$400+ on an X2. A simple Athlon would have been a good choice, but I wanted to have dual-core, so I went for the Intel "patch-job" PD820 which cost me CAN$300. AMD didn't have anything to compete at that price-point. Now, I'm so glad I didn't get the AMD CPU, because it would have been socket 939, and the hope to upgrade a couple years down the road would be shot down. Now, I can drop in a nice C2D into the same socket with a simple BIOS update.

But I digress... my main point is that the vast majority of the consumer segment is very value-oriented. Looking at total system cost, AMD looks very good because the motherboards tend to be cheaper. This must be where the CPU manufacturers make their real money. AMD's problem is keeping up with demand. If they get more fabs online, they are going to grow nicely.

Regarding everything else, I think that the second-half of 2007 will be a great time for you enthusiasts: both K8L and Intel 45 nm. For me, I'm waiting for 2008 before I consider my next big upgrade.
December 12, 2006 5:37:42 PM

Quote:
You may be right with quad cores not taking off this year- dual cores took a year or so to come down in price and up in performance enough for most people to want to bite. So maybe 2008 is the year of the quad core.

I think we need to watch the software.

Dual-cores are barely being utilized right now.

I don't think Dual-core saturation has reached a high enough density to push the software toward multi-core technology.

I do, however think Dual-core saturation will be sufficient in 2007.

Quote:

When you say that AMD likely won't break 3.2 GHz on 65 nm, I assume that you're talking about the quad-core CPUs. Those are supposed to go up to about 2.9 GHz at launch, so maybe AMD will offer an Agena FX at 3.2 or so.

It might run a touch warm at speed, but much cooler than the Kentsfield, so they very well may give it a go.


I believe I said, "AMD finally gets the 65-nm SOI just right in the last quarter of the year just breaking 3.2 Ghz."

I intended it to be with 65nm and either dual or quad core.

If they really wanted to they could probably reach 3.2 Ghz now with 90nm but no one would want them.

All indications say 2.9 core clock will be the best they can offer in the third quarter so I think predicting 3.2 Ghz for any AMD multi core regardless of heat dump might be a little optimistic of me, but AMD is due for a break.


Quote:

The process should scale up much farther than 3.2 GHz as AMD has had a little while to work on it by then.


Eventually, yes, I see the 65 nm process scaling well but I don't think they will get much past 3.2 by the end of 2007.


Quote:

If AMD decides to make a run of fast dual-cores and not just midrange dual-cores, there is no doubt that you'll see units going above 3.2 GHz as the process will support it and so will the thermal characteristics.

Maybe we will see one or two high-end dual-cores(3.2 and 3.4?) by the end of 07' but they will be scorchers, it really depends on how well K8L performers vs. the completion at the time of it's release.


Quote:

However I doubt that AMD will try to release performance dual-core parts once the quads start shipping in some volume.

As I said, quad cores are doubtful to start shipping in volume for 2007. AMD would probably be in trouble if quad-cores really caught on because their supply problems would multiply even with their shift to 65 nm.


Quote:

Intel very well might start to sample some 45 nm chips by the end of the year but they're not supposed to ship until 2008. This would keep pricing pressure on AMD if they could accelerate their switch to 45 nm, but Intel would have to sell for less than AMD as they are not due for a core refresh until 2008-2009.


Most roadmaps I have seen say we will start seeing 45 nm refreshes in the second half of 2007. Although, mass conversion to 45 nm is projected to happen in early 2008.


Quote:

The Core 2 arch will have to battle with K8L, and I don't think that Intel will have the upper hand in that battle like it does with the K8.

I agreed that K8L should combat most of Core 2's clock for clock advantage, but I expect Intel to keep pumping the Ghz to maintain an overall dual core dominance.


Quote:

Intel also seems to be jumping on the quad-core bandwagon (heck, they *started* the quad-core bandwagon!) and you'll probably not see significant dual-core development from Intel once the Kentsfields become widely shipping. Just look at the Prescott -> Cedar Mill P4s once the Pentium Ds shipped. There was no increase in clock or anything and you know that they could have made a Cedar Mill 4.0 GHz or faster if they wanted to. The Core 2 Duos tend to be able to reach 3.6 or so GHz overclocked, so there is some headroom left. Not 4.0 GHz worth, but I can easily see Intel making a 3.20 or 3.33 GHz Core 2 Duo X6900. It will run a bit warmer than the other Core 2 Duos, but it should be less than 85-90W, which is not unreasonable at all for a performance part.


Wouldn't Intel if it feels pressed by K8L push for(or past?) 4.0 Ghz on 65 nm dual cores.

If quad-cores don't take off then the dual cores will keep upping the high-end clocks.

As it stands now it is fairly common for low-end Core 2's generally make 3.4 overclocked and the high-end Core 2's often reach 4.0 overclocked.



Quote:

I'd predicted that the Kentsfield will not get much above 3.0 GHz on 65 nm as it's already at a TDP of 130W (173W max. dissipation) at 2.67 GHz. That's as hot as any chip they've ever released- 90nm Pentium Ds included. Pushing it to 2.93 or 3.00 GHz would mean its TDP would get to 150-160W (200-213W max. dissipation) and that would probably even be too much for the progenitor of NetBurst to handle.


I think that is fairly reasonable.

Quote:

So the Kentsfield is stuck at 2.67-3.00 GHz and the Agena is at roughly the same clock speeds but will have better IPC and thermal characteristics.


As I said, "It [K8L] catches up with Intel in the quad-core performance vs. Hz, sporting slightly lower energy usage overall. The K8L is an improvement but not enough to reclaim the crown on the high-end desktop."

Assuming that K8L can dispute the quad-core performance crown but, "Quad cores fail to take off by the end of the year. Leaving AMD's modest gains a hollow victory."

Quote:

So yeah, Intel does not have much interesting for 2007. 2008 ought to be a better year for them with the 45 nm switch and heavy development on the Nehalem micro-architecture and DDR3 adoption.

Agreed.

----------
Sorry for the really long post.

EDIT:
AMD might be able to answer a few questions on Thursday.
December 12, 2006 5:39:31 PM

Quote:

In late 2007 Sandman posts "AMD is now slightly behind Intel, but there processors outperform Intels in all MS Word benchmarks so I will still stick with AMD." The post is secretly typed on an Intel Penncrest system running word. :lol: 


LOL

I guess if you cant beat'em, flame'em :p 

I think we will see a roll reversal as K8L is released in qty around the same time frame as C2D was lauched this year. Intel will be responding with overclocked C2D and C2Q to keep par with AMD.

Or not, what the h*ll do I care, Im not BaronBS here. :lol: 
December 13, 2006 10:49:31 AM

@ almerac & jamesgoddard

AMD ending up in Bill's claws would be a very interesting situation. I wonder what the EU & DOJ would have to say if Windows Vista 2008 would run only on AMD CPUs. Now that is a helluva way to dismantle Intel and give Bill "666" Gates his long-desired control of the planet! :D 

@ BaronMatrix

Why do I get all righteously indignant? Because I have had this solemn, sacred power imbued and vested in me by the highest spiritual and temporal authority in the cosmos: Me. :p 
December 13, 2006 10:56:31 AM

Now that would be something...
December 13, 2006 11:05:56 AM

Quote:
Now that would be something...


Dade dude, that would be a balls-out move that would freeze the industry in its tracks. M$ uses the few quadtrillion dollars that it has left over from last year's coffee budget to buy out AMD, puts some proprietary widgitywidget in each CPU, and Vista 2008 won't boot on anything else. I wonder if the Feds could smack M$ for monopolistic practice because this would be an "exclusionary" and not monopolistic practice. M$ could reply that nobody is forcing people to buy AMD CPUs. They can run Linux, etc. on Intel all they want. That might fly!

Can you imagine what kind of flame (and maybe even real) wars that would start?

OUCH! :twisted:
December 13, 2006 11:16:40 AM

I never said good or bad I just said something. Couldn't find the right word.
!