Sign in with
Sign up | Sign in
Your question

Kentsfield or 4x4 or K8L ... Performance Predictions?

Last response: in CPUs
Share

Which Platform will give you the best Performance and or Price/Performance in 6 to 8 months or when launched? Kentsfield or AMD 4x4?

Total: 70 votes

  • Intel''s Kentsfield ...
  • 65 %
  • AMD''s 4x4 ...
  • 20 %
  • Either or ... Performance will be about the same ...
  • 16 %
July 19, 2006 10:37:32 PM

With Intel's quad core Kentsfield (2 dual core Conroes on a multi-chip module with 2x4MB cache on a single processor motherboard) due within 6 to 8 months, and AMD's 4x4 (2 dual core K8's on a dual processor motherboard) due around the same timeframe brings some questions to mind...

Which of these concepts do you believe will be the better performer?
What are your feelings as to the reason for your selected CPU out-performing the other?
What will it cost at launch approximately, and when do you expect its launch?

Initially, iirc 4x4 will not be based on K8L, but on the K8G. Is this correct?
Will they roll this 4x4 concept onto the K8L core shortly after release?

Supposedly AMD's K8L is due sometime in 2007 and in the 2nd half at that... Has this CPU been tapped out yet for engineering sampling? Are there any benchmarks illustrating what to expect peformance wise?
July 19, 2006 10:46:23 PM

You just started this to debate roadmaps again and again.... why? We know you like intel's so stop guessing at the future with no basis. I think AMD will perform better next year, and if both those systems involve 2 seperate cpu's i think AMD will do better.
July 19, 2006 11:12:49 PM

I think in single core based applications the performance of Kentsfield should be better, but in multithreaded they should be a little more even thanks to HTT. Either way I think Intel will have a serious advantage thanks to 65nm and the fact that it's going to be 4 cores on 1 chip vs. 2 chips with 2 cores each.
Related resources
July 19, 2006 11:17:23 PM

he just wanted people's opinion... its not so bad...

well if the performance of AMD is still the same (core) with 4x4, i think i would rather have 4 core similar to conroe
a c 99 à CPUs
July 19, 2006 11:20:40 PM

From what is being said around the Web, the Kentsfield will run on a shared 1333 MHz bus for all 4 chips. That equates to a 333 FSB for each chip and will likely result in a situation where if all four cores are working, there will be bandwidth starvation. I am guessing that the 4x4 will be very much in practice like the Opteron DP setups with inter-core communications over HTT and two RAM banks and NUMA. That will have much more memory bandwidth for the CPUs than the Kentsfield would have.

Now if the Kentsfield magically had a dual FSB to one chip setup (which it likely won't as it's going to be an LGA775 chip and 775 pins is not supposed to be enough for that) then I'd say Kentsfield. K8L will likely be a little more than 6-8 months away, more like a year from what's being said, so it is not in consideration.
July 19, 2006 11:21:02 PM

Too many unknowns to make a real sort of prediction at this point in time.
July 19, 2006 11:30:48 PM

First, how can you say you know when 4x4 is coming out? AMDs internal documents and public interview say sooner rather than later. I did find a site that had the tests used on the system.
CineBench, Maxon, WinSAT.

I'll look for the link.
July 19, 2006 11:32:20 PM

Quote:
I did find a site that had the tests used on the system.
CineBench, Maxon, WinSAT.


Hmm, I believe it was the sock puppet parrot that told you about the tests.
July 19, 2006 11:36:25 PM

Quote:
I did find a site that had the tests used on the system.
CineBench, Maxon, WinSAT.


Hmm, I believe it was the sock puppet parrot that told you about the tests.

I would have said that. I found the site with it this afternoon.
July 19, 2006 11:37:13 PM

Then provide the link.
July 19, 2006 11:39:53 PM

Here is the link. It's from what some say is an anti-AMD article.

Linkage!

Excerpt:

AMD is developing a new platform for the PC enthusiast market. It’s a decidedly multi-core processor, the multi-threaded application future for the PC industry. AMD believes it will have the best solutions for the market in 2006. “In our technology development for the PC enthusiast market, AMD is seeing an 80 percent average performance uplift from four core over two core AMD64 processor-based systems on multi-threaded benchmarks including Futuremark 3DMark 06, Maxon Cinebench R9.5, and WinSAT running on the latest beta release of Windows Vista.”
July 19, 2006 11:43:13 PM

Why is it when someone posts what they like to call "proof" it is either the Inquirer or some obscure site?
July 19, 2006 11:49:20 PM

Wow, haven't been there since the R520 scandal.
July 19, 2006 11:53:16 PM

Quote:
Why is it when someone posts what they like to call "proof" it is either the Inquirer or some obscure site?



I didn't say it was proof I said it was a link to the apps used - supposedly.
July 20, 2006 12:06:34 AM

Quote:
Which of these concepts do you believe will be the better performer?

I would have to say Kentfield by a small margin 5%-7%, just because the K8 will essentially have an open pipe as it were to main memory. But if the processors all share the same cache similar to Core 2's configuration I don't see bandwidth being a real issue at this point.

Additionally they could theoretically allow two independent buses, for Kentfield with regards to the Xeon 5100's are 774 so pin compatibility wouldn't be a issue.

But quad core K8's aren't going to be something to easily beat, with regards that AMD will be on a 65nm node which will greatly help thermal issues and electrical draw but whether or not they will have good or half decent yields is also a question yet to be answered.

All in all I would have to say they will be close.

Quote:
What are your feelings as to the reason for your selected CPU out-performing the other?

Above that was stated.

Quote:
What will it cost at launch approximately, and when do you expect its launch?

I would hope they would launch both parts Q2 07, but that isn't up to me to decide.
July 20, 2006 12:15:28 AM

tying to predict the future eh.

it all depends on what processors are better with price really.

kentsfield goes on the 755 socket right?

k8l goes on the am2 socket right?
July 20, 2006 12:53:19 AM

Quote:
tying to predict the future eh.

it all depends on what processors are better with price really.

kentsfield goes on the 755 socket right?

k8l goes on the am2 socket right?


I don't predict the future, I predict trends.

Kentsfield is 775LGA. K8L will be AM3 but work in AM2. It won't be a desktop chip though. Indications are that FX will go quad core around the time that 65nm X2 comes out.
July 20, 2006 1:19:15 AM

I think 4x4 will be faster Just due to the limit on the fsb of the kentsfield will have to split the fsb.
July 20, 2006 1:33:25 AM

Will you have to buy a windows license for each socket? Or will Microsoft change their fees for desktop licensing?
July 20, 2006 2:05:51 AM

Quote:
From what is being said around the Web, the Kentsfield will run on a shared 1333 MHz bus for all 4 chips. That equates to a 333 FSB for each chip and will likely result in a situation where if all four cores are working, there will be bandwidth starvation. I am guessing that the 4x4 will be very much in practice like the Opteron DP setups with inter-core communications over HTT and two RAM banks and NUMA. That will have much more memory bandwidth for the CPUs than the Kentsfield would have.

Now if the Kentsfield magically had a dual FSB to one chip setup (which it likely won't as it's going to be an LGA775 chip and 775 pins is not supposed to be enough for that) then I'd say Kentsfield. K8L will likely be a little more than 6-8 months away, more like a year from what's being said, so it is not in consideration


Not good for 4x4 pricing mind you. Dual processor supporting chips, with two CPUs, two, two channel memory, meaning minimum would be 4 DIMMs, probably it will have 8 DIMM slots.
July 20, 2006 2:27:07 AM

Quote:
Will you have to buy a windows license for each socket? Or will Microsoft change their fees for desktop licensing?


I do believe MS will charge additional fees if using more than 2 cores... I saw something about it on MS's website but can not recall details.
July 20, 2006 2:51:38 AM

Quote:
With Intel's quad core Kentsfield (2 dual core Conroes on a multi-chip module with 2x4MB cache on a single processor motherboard) due within 6 to 8 months, and AMD's 4x4 (2 dual core K8's on a dual processor motherboard) due around the same timeframe brings some questions to mind...

Which of these concepts do you believe will be the better performer?
What are your feelings as to the reason for your selected CPU out-performing the other?
What will it cost at launch approximately, and when do you expect its launch?

Initially, iirc 4x4 will not be based on K8L, but on the K8G. Is this correct?
Will they roll this 4x4 concept onto the K8L core shortly after release?

Supposedly AMD's K8L is due sometime in 2007 and in the 2nd half at that... Has this CPU been tapped out yet for engineering sampling? Are there any benchmarks illustrating what to expect peformance wise?


There are a ton of folks on this forum who know a lot more than I do, so I will ask the following about AMDs upcoming (?) 4x4 idea....

Why?

The cost difference between the 4x4 concepts and a legitimate dual socket motherboard seems, at least to me, to be pretty dam small.

2 cpus still need 2 sockets, still need a power supply and regulation to feed two cpus, still need bandwidth to feed two cores.

The PCI slots, PCI Express slots, memory slots, all cost the same on 2 sockets versus 4x4..

I just don't seee the cost advantage versus a legit dual socket board?

You know those kids drink commercials that boast "Contains 25% real fruit juice!"..? My reactiion has always been "Well if 25% real fruit juice is so great, why not just buy, well, real fruit juice...?"

If dual CPUs is so amazing... why not just go to a mainstream, lower cost consumer oriented Dual cpu board and save all the conversion and hassles needed to market, develop and tweek the 4x4 idea?

Maybe the high IQ AMD fanboys on this site can clue me in, but versus a legit dual socket motherboard, every time I hear 4x4 I just go "Why....?"
a c 99 à CPUs
July 20, 2006 2:57:41 AM

I bet that 4x4 will be treated like a regular dual-socket setup, i.e. you must use XP Professional to use the second socket. So you will pay more, at least for now. I do *not* see MSFT changing this as 4x4 will not be that common that it will dictate a rewrite of how per-processor fees are charged as what happened with dual-core processors.
a c 99 à CPUs
July 20, 2006 2:58:34 AM

No, but RAM does scale pretty linearly: 4 512MB sticks cost pretty much the same as 2 1GB sticks.
July 20, 2006 3:01:56 AM

Quote:
I bet that 4x4 will be treated like a regular dual-socket setup, i.e. you must use XP Professional to use the second socket. So you will pay more, at least for now. I do *not* see MSFT changing this as 4x4 will not be that common that it will dictate a rewrite of how per-processor fees are charged as what happened with dual-core processors.


Won't you have to pay for 2 copies of XP? As stated earlier, don't you have to have a license per socket?
a c 99 à CPUs
July 20, 2006 3:05:56 AM

This is a big "if:" if the 4x4 can use regular Athlon 64s and Athlon 64 X2s and not just FX-line chips, this allows one to build a dual-socket computer without paying the significant premium for DP-capable server chips. The Athlon X2 4400+ for Socket AM2 costs $460 while an Opteron of equivalent speed and cache- the Opteron 275- is $708. Also, the 4x4 will likely use cheaper unbuffered, unregistered RAM vs. registered ECC that the Opteron 200/800 series needs. If a dual socket 940 Opteron board costs the same as a 4x4 board (it likely won't as dual 940s are generally $300-400- twice that of a good AM2 board) you still save $500 on the CPUs. THAT's the appeal. And I bet that it won't be gamers but people who want a 4-core workstation on the cheap that buy this. I know I would if I were looking for a new unit.
a c 99 à CPUs
July 20, 2006 3:14:07 AM

No. For regular consumers, XP Home has a license for one socket and XP Pro can be installed on dual-socket computers. You do not have to pay for more copies of XP to run on more sockets- if there are more sockets than the OS supports, it will not recognize those additional sockets and your processors in them will not be used. You have to buy a version of Windows that supports as many sockets as your computer has to be able to use them all, so I guess the OS prevents you from being in violation of its EULA that way.

Now businesses are a wholly different story, especially with server OSes that support lots of CPUs and clients. I am not that knowledgeable about proprietary OSes and their licenses in that arena- I know Linux. Linux can support 256 CPU cores, be they in 256 sockets or one with an unmodified kernel. There is no restriction built into the OS itself, however if you buy a supported version of a server Linux from Red Hat or Novell, there might be and you'd have to call them. However, from what I have seen, I highly doubt that there are per-processor fees. There are NO per-processor fees or restrictions in any consumer desktop Linux that I have ever seen or used and I've put it on multi-socket boxes.
July 20, 2006 3:32:15 AM

Thanks for the clarification.
July 20, 2006 3:34:06 AM

Quote:
otherwise it appears like they come from the BaronCaboose.
Say what now?
July 20, 2006 3:48:24 AM

Quote:
otherwise it appears like they come from the BaronCaboose.
Say what now?

Sorry, no offense --- maybe I should have written BaronBehind. :lol:  The comedy on these boards is sometime greater than those Blue Collar Comedy guys. Honestly, who could EVER compete with Shikarou and Mike?
July 20, 2006 4:14:42 AM

Get Mike and Shikarou talking about things everyone can relate with and you'd have a real hit. :lol: 
July 20, 2006 5:50:28 AM

So Kentsfield and 4x4 are coming out this year? I just saw a link and Intel stated they will release it by end of this year.
July 20, 2006 6:27:31 AM

Until K8L it's impossible for AMD to outperform Intel. Remember, dual core is better than dual CPU. That means Intel has a better performing CPU . . . except four of them on one CPU, whereas AMD has four pretty good CPUs on two dies separated over HTT. while HTT is fast, it's not on-die fast.
July 20, 2006 8:11:21 AM

sure about that? Lower fsb per cpu. let see 1 cpu with 4 cores. shares fsb. To me that would be slower. Vers two chips with there own memory control and two cores each.

I think the best idea is wait and see. Then we see what happens.
July 20, 2006 9:07:03 AM

Kentsfield i think would be the winner. If not overall at least in the price/performace steaks considering how expensive it would be to by 2 fx processors and a dual socket mobo (providing the "4x4" concept is limited to fx series cpu's)
July 22, 2006 4:47:45 AM

Quote:
This is a big "if:" if the 4x4 can use regular Athlon 64s and Athlon 64 X2s and not just FX-line chips, this allows one to build a dual-socket computer without paying the significant premium for DP-capable server chips. The Athlon X2 4400+ for Socket AM2 costs $460 while an Opteron of equivalent speed and cache- the Opteron 275- is $708. Also, the 4x4 will likely use cheaper unbuffered, unregistered RAM vs. registered ECC that the Opteron 200/800 series needs. If a dual socket 940 Opteron board costs the same as a 4x4 board (it likely won't as dual 940s are generally $300-400- twice that of a good AM2 board) you still save $500 on the CPUs. THAT's the appeal. And I bet that it won't be gamers but people who want a 4-core workstation on the cheap that buy this. I know I would if I were looking for a new unit.


Which again brings me to "why"...

Let's assume you are 100% correct and 4x4 takes off in the 4 core workstation market (Your reasoning by the way is totally sound I agree with it) - with 4x4 AMD will have managed to transform $1500 in Opteron sales into $500 of AM2 sales... (remember AM2s are now hugely cheap - AM2-5000+ are reported to be just $282 each...)

Doesn't seem like a winning plan to me....
July 22, 2006 6:14:47 PM

Quote:
This is a big "if:" if the 4x4 can use regular Athlon 64s and Athlon 64 X2s and not just FX-line chips, this allows one to build a dual-socket computer without paying the significant premium for DP-capable server chips. The Athlon X2 4400+ for Socket AM2 costs $460 while an Opteron of equivalent speed and cache- the Opteron 275- is $708. Also, the 4x4 will likely use cheaper unbuffered, unregistered RAM vs. registered ECC that the Opteron 200/800 series needs. If a dual socket 940 Opteron board costs the same as a 4x4 board (it likely won't as dual 940s are generally $300-400- twice that of a good AM2 board) you still save $500 on the CPUs. THAT's the appeal. And I bet that it won't be gamers but people who want a 4-core workstation on the cheap that buy this. I know I would if I were looking for a new unit.


Which again brings me to "why"...

Let's assume you are 100% correct and 4x4 takes off in the 4 core workstation market (Your reasoning by the way is totally sound I agree with it) - with 4x4 AMD will have managed to transform $1500 in Opteron sales into $500 of AM2 sales... (remember AM2s are now hugely cheap - AM2-5000+ are reported to be just $282 each...)

Doesn't seem like a winning plan to me....

Those are certainly good questions, mostly, since there's such scarce info from AMD, mainly. However, the 4x4 concept is very interesting in itself: Apparently, it's not confined to CPUs and allows other processing/networking/storage IP companies to contribute to such a versatile approach.
FPGA (Field Programmable Gate Arrays), are starting to include DSPs & co-processors are also a serious option, within the 4x4 concept set up.
Most probably, Intel will remain with the performance per uArch crown and, whenever needed, some sort of IMC-like solution will be introduced, be it CSI or something else. For now, AMD has its IMC running, HT & ccHT working, the K8 & a hard time 300mm/65nm transition to make (and, they're late!); but, costs aside, Intel might follow a 4x4-like implementation in the future, even before the 3D stacking processing innitiative, since it can hardly compete with [cheap] multi-socket (or slot) solutions. Issues like availability, upgradeability & costs will erode, as soon as demand grows.
Whatever the outcome, either from AMD or Intel, I find the 4x4 solution a very interesting concept. My opinion, of course.


Cheers!
July 22, 2006 6:34:09 PM

Quote:
Those are certainly good questions, mostly, since there's such scarce info from AMD, mainly. However, the 4x4 concept is very interesting in itself: Apparently, it's not confined to CPUs and allows other processing/networking/storage IP companies to contribute to such a versatile approach.
FPGA (Field Programmable Gate Arrays), are starting to include DSPs & co-processors are also a serious option, within the 4x4 concept set up.
Most probably, Intel will remain with the performance per uArch crown and, whenever needed, some sort of IMC-like solution will be introduced, be it CSI or something else. For now, AMD has its IMC running, HT & ccHT working, the K8 & a hard time 300mm/65nm transition to make (and, they're late!); but, costs aside, Intel might follow a 4x4-like implementation in the future, even before the 3D stacking processing innitiative, since it can hardly compete with [cheap] multi-socket (or slot) solutions. Issues like availability, upgradeability & costs will erode, as soon as demand grows.
Whatever the outcome, either from AMD or Intel, I find the 4x4 solution a very interesting concept. My opinion, of course.


Cheers!


Won't it be possible for Intel to release a cheaper 2P compatible chipset for the Woodcrest? Why can't they release cheap Woodcrest motherboards with Crossfire support, for example, and other high-end desktop goodies? With, or without dual FSBs, depending on the costs involved.

Surely, Intel would not mind Woodcrests getting into desktop machines, would they?
July 22, 2006 7:27:33 PM

Quote:
Which again brings me to "why"...

Let's assume you are 100% correct and 4x4 takes off in the 4 core workstation market (Your reasoning by the way is totally sound I agree with it) - with 4x4 AMD will have managed to transform $1500 in Opteron sales into $500 of AM2 sales... (remember AM2s are now hugely cheap - AM2-5000+ are reported to be just $282 each...)


The Pro wksta market uses ECC and Quadro. If you're working on million dollar SW or graphics you won't buy something just cause it's fast. You want solid and dependable which is what Opteron is.

This will only appeal to th eenthusiast AMD buyer, though it may win some Intel fans since 8x8 is due early next year. If they all fit in AM2 that is a seamless upgrade to X4L late next year.

I keep saying that this will help unlock the additional bandwidth of DDR2. quad will do even more as there will probably be DDR2 1600 by then. GDDR4 is running at 3.2Ghz (Samsung).

I think it wll be close between 4x4 and Kentsfield, but X4L has the same specs as announced, Clovertown has no chance. the K8L arch is a little more powerful than Core 2.

And keep in mind that I knew the Core 2 arch would defeat X2 according to the specs.
July 22, 2006 7:31:04 PM

Quote:
Won't it be possible for Intel to release a cheaper 2P compatible chipset for the Woodcrest? Why can't they release cheap Woodcrest motherboards with Crossfire support, for example, and other high-end desktop goodies? With, or without dual FSBs, depending on the costs involved.

Surely, Intel would not mind Woodcrests getting into desktop machines, would they?


Well if the talk is true that AMD and ATI will merge next week...there will be no crossfire support for INTEL
July 22, 2006 7:35:08 PM

Quote:
Whatever the outcome, either from AMD or Intel, I find the 4x4 solution a very interesting concept. My opinion, of course.


You seem to be one of the most non-biased posters here.

AMDroids say 4x4 is straight from God.
Intelliots say 4x4 is crap.

Nice to see non-bias intrest for the 4x4 platform.
July 22, 2006 7:46:13 PM

Quote:
Whatever the outcome, either from AMD or Intel, I find the 4x4 solution a very interesting concept. My opinion, of course.


You seem to be one of the most non-biased posters here.

AMDroids say 4x4 is straight from God.
Intelliots say 4x4 is crap.

Nice to see non-bias intrest for the 4x4 platform. I am another non-biased poster. 4x4 isn't really anything. It may be an innovation yes, but who is going to buy it? I will not.
July 22, 2006 7:50:28 PM

me going to 4x4...hinges on a few things

1. can I use x2's on it?
2. Can i put K8L's on it later?
3. What big time games or Apps coming out will take advantage of it?


It has been said that Crysis will be optimized for the 4x4.

Crysis will be THEEEEE game of 07 and it would be a hell of a launch title for the 4x4.
July 22, 2006 8:08:07 PM

Quote:
me going to 4x4...hinges on a few things

1. can I use x2's on it?
2. Can i put K8L's on it later?
3. What big time games or Apps coming out will take advantage of it?


It has been said that Crysis will be optimized for the 4x4.

Crysis will be THEEEEE game of 07 and it would be a hell of a launch title for the 4x4.


1.) I believe that AMD said that you could only use FX processors on it, I don't remember if it is AM2 only.

2.) Yet to be seen, I would hope and think so.

3.) The majority do not, but 4x4 would sure as hell future proof you.

Jack and I took a look at Crysis, it seems very interesting. I can't wait for it.
July 22, 2006 8:08:41 PM

Quote:
Won't it be possible for Intel to release a cheaper 2P compatible chipset for the Woodcrest? Why can't they release cheap Woodcrest motherboards with Crossfire support, for example, and other high-end desktop goodies? With, or without dual FSBs, depending on the costs involved.

Surely, Intel would not mind Woodcrests getting into desktop machines, would they?


Actually, it would - most probably - be overkill, bringing Woodcrest to the DT; and, Intel already declared that it won't introduce newer chipsets for the DT soon, which might mean that they'll focus more on the server/workstation & mobile, leaving the DT chipset market open to ATi & nVidia (and others...); hence, it's expected that both CrossFire & SLI implementations will begin to appear in volume, within the next few months.
This attitude makes sense, for several reasons (I'm not downgrading the YxY concept; just trying to stick with what's going on, right now):

a. Costwise yes, it would be [substantialy] expensive to bring a 2P chipset solution into the DT space - now - since price & performance wise, Conroe seems more than fit to the task;

b. It couldt happen in the near-to-mid-term; however, it would be up to third parties (ATi; nVidia) willing to cooperate with Intel, on that issue; I'm not sure if Intel would be interested, at the moment, to bring its high-end SRV/WKS high profit platforms to the DT space (and, they still have to prove they can be successful at the SRV/WKS);

c. Talking high-end features, every 1st-tier motherboard & chipset manufacturer (among other IP contributors) are actually bringing the best they've got to the DT space; letting the mobile space aside, for a while, not even the SRV/WKS has the versatility & # of features an off-the-shelf DT platform has except, obviously, the scalability & CPU performance...


Cheers!
July 22, 2006 8:15:40 PM

Quote:
honestly ,rich, its hard to say;in amd's own words 60% jump in 07,and 150% jump in 08 ,via the link baron provided to the V.P's video release.
i am unaware of Intels performance predictions,and AMD's remain to be seen as of this time.perplexing question given the lack of substance in the market.


What do you mean by "substance in the market?" Intel is abou 90% non-Core 2, so they have until next year before Core 2 is going out in every Dell.

IF AMD can strip down some nForce wksta boards pretty quick, they will deflate soem of Core 2 especially if they drop the prices of 5000+ the way it's been hinted. That would be a pretty good $1400 system.


$282 x 2 = $564
$150 mobo
$150 2 GB RAM


$864 for the bare bones.

$100 for the HD
$100 for the case
$200 for the video

That is a helluva deal. Anyone want to buy a decked out 4400+ system?
July 22, 2006 8:22:52 PM

Quote:
i may be mistaking you ,but 4x4 will be an am3 feature as well .and in order to retaliate against intels incredible performance/price manuever youd think that 4x4 couldnt be justifiable on an exclusively high end scenario.
that in reality ,to fully counter intels brilliant success,theyd have to open the whole line up for 4x4 to reply to conroe and take it up to a new level altogether,renerding intels success a pittence of its former self.
AM3 as well? I still will not buy it, but that is good to know for future reference. I believe K8L will do well against Conroe without the frighteningly priced 4x4, but that also remains to be seen. We will see.
July 22, 2006 8:30:23 PM

With you there. But I somewhat hate AMD now; if they buy my beloved ATi I am afraid I will go apesh!t. But even if I hate them my posts will not be tainted, and if they have the superior product I will not think twice about buying one.
July 22, 2006 8:34:19 PM

I pride myself in showing the non-biased side of myself, it can only help people. But for what it is worth, I'm with you on this particular subject regardless of what our peers believe.
July 22, 2006 11:37:47 PM

Quote:
Most probably, Intel will remain with the performance per uArch crown


thats pretty bold,and time will tell if that can be quantified,i wouldnt take that position ,but i wont try to argue against iy anymore than to say,its too early to make such a bold proposition.

Perhaps I should have been more emphatic on some time-frame; the prob is that, when AMD completely transitions its manufacturing process to 300mm/65nm mature yields (with - most probably - the same uArch), Intel will be upgrading its current one (Core) at a 45nm node process. If this truly happens (mind that I'm referring to uArch alone!), I don't see how AMD can regain a comparable (much less, superior) technological equity over its most direct competitor.
Things aren't, usually, that linear; uArch aside, that's why I find the 4x4 concept so interesting... and, AMD can still pull up some new interesting & feasible ideas, in order to remain attractive. Realistically though, so can Intel.
So, with as much as one's allowed to know outside those info-bunkers, I'd say that this time, AMD is in real trouble for the foreseeable future. In such circumstances, even simple, brilliant & effective ideas might be prohibitively costly & too late to implement. I guess this is a consensual view.
That, was my point.


Cheers!
!