Sign in with
Sign up | Sign in
Your question

Realistic AMD predictions for 2006-2010

Last response: in CPUs
Share
July 10, 2006 7:09:33 PM

have done alot of digging....

2006

and AMD has no answer for the CORE 2 as of now


4x4 is proof of this.... this is just a face saving manuever to get hardcore AMD fanbois with the cash to buy an AMD that will rule in some tests.

AMD is slow with .65 tech..

AMD will be able to battle INTEL on mid to high level on the server side ...but thats about it.

dekstop side is lost to the CORE 2 DUO hordes


2007

AMD will still be behind the eight ball on most fronts....AMD will slash prices deep... with no answer to CONROE other than a ridicoulously costly 4x4 setup.

there will be no K8L saviour in 07. not on the desktop side anyway

Intel should pretty much run unabaited through 2007.... AMD has no real answer.

2008

Enter AMD's answer... the long awaited K8L

will it be enough.... AMD will not be resting on its laurels waiting on the K8L to come knocking... CORE 2 DUO will be stronger

and @ 45um


For the forseeable future its an INTEL world

2010

K10 is born.... no further info
a c 446 à CPUs
a c 111 À AMD
a c 110 å Intel
July 10, 2006 7:36:47 PM

2007

45nm Conroe

2008

Nehalem - Conroe's successor

2009

32nm Nehalem

2010

Nehalem's successor will appear.

2011

The End Of The World

Quote:

Based on the predictions of Nostradamus as of now... WorldWar 3 is most likely to happen around 2011~2012. Until then both the communities... the Christians and the followers of the Islamic dharma are silently in preparation.



A rise in the level of waters around 10~12 feet in the seas and the oceans is predicted in World War 3. The radio active vapors would result in melting of the Arctic ice on an unprecedented scale. World War 3 would result in countries like Australia almost getting wiped out from the face of the Earth. World War 3 would result in waves raising to a height of 1000 feet... something unheard of!

July 10, 2006 7:38:45 PM

Realistic my ass! Who are you to judge how are thins going to happen based on some stupid outside news. Intel seems to have conquered the world already with a processor wich isn't even in stores. AMD shud just quit making procesors by now. They are plain fools not to lissen to geniuses like yourself, and find some other business like weaving.

Now cut on that crack you just smoked, wake up, and get a life, man!
Related resources
July 10, 2006 7:53:49 PM

Heh, where've U got the crystal ball man; got an exam this week end, just want to see how it ends. Feet on the ground!
July 10, 2006 8:15:42 PM

i have been accused of being the KING OF THE AMD fan bois

now when i make a realistic view of what AMD will have in the pipe

i am attacked


amazing
July 10, 2006 8:28:57 PM

Quote:
i have been accused of being the KING OF THE AMD fan bois

now when i make a realistic view of what AMD will have in the pipe

i am attacked


amazing


i think its your wording.

since everything about the future is speculative, you shouldn't say "the realistic view of AMD's future", because no one knows for certain. even though there are release date, but it doesn't mean AMD will definately release the product right on the date they promised.

however, you can say, "my opinionated view of AMD's future".
July 10, 2006 8:40:37 PM

Realistic AMD predictions for 2006-2010


the title of the thread said predictions
not fact or gospel truth
July 10, 2006 8:43:26 PM

Intel's roadmap is ambitious and aggressive, by their own admission.
The giant has woken from its slumber, it has reorganized priorities and is presumed to be investing heavily in research and development.
But that's no guarantee that Intel will be able to hit each and every one of its development targets on the bullseye.
Maybe they will, maybe they won't. But I think it's foolhardy to take all of Intel's target dates as "done deals."
Some may not even be achievable. But it looks good on paper. Helps settle down the investors, you know?
July 10, 2006 8:53:50 PM

No they won't hit every target projection, but at least they are projecting!
This does instill confidence from investors, and it bolsters that confidence when those targets are met...
AMD on the other hand seems to be producing BS projections and fluff like RHT, so when the time is right they will consume Conroe... :roll:
BS projections in the face of stiff competition with no working samples or even detailed schematics makes investors leary...
Combine that with reducing retail pricing and a drop in market share and the picture starts to get ugly, and quickly...
July 10, 2006 9:01:33 PM

Quote:
have done alot of digging....

2006

and AMD has no answer for the CORE 2 as of now


4x4 is proof of this.... this is just a face saving manuever to get hardcore AMD fanbois with the cash to buy an AMD that will rule in some tests.

AMD is slow with .65 tech..

AMD will be able to battle INTEL on mid to high level on the server side ...but thats about it.

dekstop side is lost to the CORE 2 DUO hordes


2007

AMD will still be behind the eight ball on most fronts....AMD will slash prices deep... with no answer to CONROE other than a ridicoulously costly 4x4 setup.

there will be no K8L saviour in 07. not on the desktop side anyway

Intel should pretty much run unabaited through 2007.... AMD has no real answer.

2008

Enter AMD's answer... the long awaited K8L

will it be enough.... AMD will not be resting on its laurels waiting on the K8L to come knocking... CORE 2 DUO will be stronger

and @ 45um


For the forseeable future its an INTEL world

2010

K10 is born.... no further info



What you seem to forget is that AMD always concentrates on SERVER first. That's why Woodcrest won't even TRY 4Way. Also, why drop prices? Core 2 won't ame a splash until next year and if AMD continues their mad power cutting they will be at 30W for special case mid2007 chips and all by the end of next year.


power consideratiosn will help AMD through this. There are still millions of NetBust chips out there and now no one really wants them.

I won't speculate as to what chip they'll release when but I wll say that Intel better enjoy the lead while they can. I'm confident that AMD is not concerned about top performance right now except in high end servers.

Remember that K8 was Opteron first. that's why I had no troubl believing K8L would be next year. the smart move woul dbe to design thsi new arch around 45nm an intro the 65nm L next year. It's not like it won't be worth anything in 08.

09 will see 8 core Opteron based on DeerHound with HT4 at 10.4 Gt/s. we'll see though.

I don't think they should try to match Intel's 2year core timetable. Just continue to make good partnerships ( especially by not buying or merging with ATi) and execute in their roadmap.

I've said ti before and I'll say it again Intel had more than two years with CRAP, AMD gets two years, though K8 isn't crap.
July 10, 2006 9:07:26 PM

If AMD and ATI have merged as reports say

then that changes the whole game

it would even make sense that AMD is dragging its feet on new products due to the Ati tie up...



i will say this ...and i have said many times...IF AMD-ATi merge ..then it means they going for INTELS throat...

they can afford to loose ground now for that long term goal ....
July 10, 2006 9:14:11 PM

Quote:
Intel's roadmap is ambitious and aggressive, by their own admission.
The giant has woken from its slumber, it has reorganized priorities and is presumed to be investing heavily in research and development.
But that's no guarantee that Intel will be able to hit each and every one of its development targets on the bullseye.
Maybe they will, maybe they won't. But I think it's foolhardy to take all of Intel's target dates as "done deals."
Some may not even be achievable. But it looks good on paper. Helps settle down the investors, you know?


Guess you're right: getting to the 45 and 32 nm targets in not as easy as [year] - 45nm <ENTER> [year] - 32nm <ENTER> at these levels you can count interconnect atoms with your fingers.
At least after ages, Intel is going to sell something more than it's name and plain, stupid gigahertz ,but AMD is not longer the tiny dwarf and is holdong the pressure pretty well.
July 10, 2006 9:16:54 PM

i agree

and i stated ...AMD will be right there with INTEL on the server side

its the desktop side that will be in trouble for some time


and this is AMD predictions.... intel will f@()@ up... we all know that

but when?
July 10, 2006 9:18:48 PM

Quote:
No they won't hit every target projection, but at least they are projecting!
This does instill confidence from investors, and it bolsters that confidence when those targets are met...
AMD on the other hand seems to be producing BS projections and fluff like RHT, so when the time is right they will consume Conroe... :roll:
BS projections in the face of stiff competition with no working samples or even detailed schematics makes investors leary...
Combine that with reducing retail pricing and a drop in market share and the picture starts to get ugly, and quickly...




Ummm, AMD NEVER mentioned this RHT thing. NEVER. that was the Inq and some others. When InStat reports then we can say they lost share. I don't think they will becuase THERE WILL NOT BE ENOUGH CORE 2 this year to affect AMD. it will affect PD though and PD will affect it.

I noticed you mentioned nothing about Intel's insane price drops. They're selling 805 for less than $100. X2 3600+ will take the air out of that and I predict that a $500 5000+ will make some real waves. Especially for lowend wksta.
July 10, 2006 9:22:22 PM

Quote:
i agree

and i stated ...AMD will be right there with INTEL on the server side

its the desktop side that will be in trouble for some time


and this is AMD predictions.... intel will f@()@ up... we all know that

but when?


Desktop meant as high end but the middle (and value layer especially) will still be dominated by AMD, just because a conscious person / company williing to boy business/home ~$50 CPUs will opt for a sempron instead of a celeron and this is dead market for Intel because it has nothing at this price/performance ratio.
July 10, 2006 9:25:30 PM

Baron, Intel can afford to reduce prices and still make employees and stockholders happy...
The same can not be said for AMD...

Use your analytical skills and look at what profits the two companies are reporting over the past several years, it will sink in soon enough...
a c 446 à CPUs
a c 111 À AMD
a c 110 å Intel
July 10, 2006 9:27:36 PM

Quote:
Intel's roadmap is ambitious and aggressive, by their own admission.
The giant has woken from its slumber, it has reorganized priorities and is presumed to be investing heavily in research and development.
But that's no guarantee that Intel will be able to hit each and every one of its development targets on the bullseye.
Maybe they will, maybe they won't. But I think it's foolhardy to take all of Intel's target dates as "done deals."
Some may not even be achievable. But it looks good on paper. Helps settle down the investors, you know?


Who cares?

According to this reliable source

Quote:

World War 3 shall be a fight between Dharma (moral values) and Adharma (wicked tendencies). None of all over the world shall be able to escape the effects of World War 3. The catastrophe shall be of such a magnitude that 1200 million people will vanish in World War 3.

After World War 3 the population of the world shall be limited to 4800 million. Primarily, India and China would not be aggressively involved in World War 3. With a combined population of about 2000 million... it shall mean that apart from India and China... the whole world would be badly affected by World War 3.


Most of us are going to be dead anyway.
July 10, 2006 9:28:18 PM

Quote:
If AMD and ATI have merged as reports say

then that changes the whole game

it would even make sense that AMD is dragging its feet on new products due to the Ati tie up...



i will say this ...and i have said many times...IF AMD-ATi merge ..then it means they going for INTELS throat...

they can afford to loose ground now for that long term goal ....



The ATi deal will AMD more because intel won't see it as them going for their throat BUT NVIDIA SURE WILL. nVidia made AMD with it's chipsets. That would make for alot of bad blood. BAD MOVE. they are in a great position now that Dell is selling Opteron. As people who wanted AMD AND Dell will demand desktops. That could mean 4% share INCREASE.

And HP is embracing the TurionX2 and it shoudl fit pretty well between Core and Core 2. AMDs long term goal is a FLEXIBLE platform based on the best ideas from the industry.


They dont' want to eclipse Intel, they just want to sell chips.
July 10, 2006 9:32:58 PM

Quote:
If AMD and ATI have merged as reports say

then that changes the whole game

it would even make sense that AMD is dragging its feet on new products due to the Ati tie up...



i will say this ...and i have said many times...IF AMD-ATi merge ..then it means they going for INTELS throat...

they can afford to loose ground now for that long term goal ....



The ATi deal will AMD more because intel won't see it as them going for their throat BUT NVIDIA SURE WILL. nVidia made AMD with it's chipsets. That would make for alot of bad blood. BAD MOVE. they are in a great position now that Dell is selling Opteron. As people who wanted AMD AND Dell will demand desktops. That could mean 4% share INCREASE.

And HP is embracing the TurionX2 and it shoudl fit pretty well between Core and Core 2. AMDs long term goal is a FLEXIBLE platform based on the best ideas from the industry.


They dont' want to eclipse Intel, they just want to sell chips.

AMD doesn't want to eclipse Intel, they just want to sell chips... :lol:  :lol: 
That is BS and you know it... What you really mean is they can not eclipse Intel, and so will settle for please buy our chips even tho they are not the fastest nor cheapest!
July 10, 2006 9:35:42 PM

When is AMD going to pull that rabbit out of their a$$e$?

That Reverse Hyper Threading should be getting the master switch flipped any day now, right? :lol:  :lol: 
July 10, 2006 9:37:15 PM

Quote:

Who cares?

According to this reliable source


World War 3 shall be a fight between Dharma (moral values) and Adharma (wicked tendencies). None of all over the world shall be able to escape the effects of World War 3. The catastrophe shall be of such a magnitude that 1200 million people will vanish in World War 3.

After World War 3 the population of the world shall be limited to 4800 million. Primarily, India and China would not be aggressively involved in World War 3. With a combined population of about 2000 million... it shall mean that apart from India and China... the whole world would be badly affected by World War 3.


Most of us are going to be dead anyway.

Time to start racking up charges on my credit cards.
July 10, 2006 9:40:56 PM

Quote:
Baron, Intel can afford to reduce prices and still make employees and stockholders happy...
The same can not be said for AMD...

Use your analytical skills and look at what profits the two companies are reporting over the past several years, it will sink in soon enough...



Losing money is losing money. Why throw it away? The plan for releasing this chip SUCKS. NetBust will burn a huge hole in their pocket as it has already. AMD isn't competing with Core 2 NetBust is.

AMD is sitting off to the side increasing their profits and OEM support. I'm sure they realize what will keep them in the black and won't lower prices more than that.


They dont' really have to lower prices. I'd try a month without it and see what happens. I'll stillpay the same price for AM2 cause it's worth it not because it's faster or slower than Intel. besides, they have the retail low end market wrapped up tight.

All Intel can do is lose money this year. AMD will not lose as much (ratio)because they dont' have warehouses full of inventory that NO ONE WANTS.

How would you sell PD with core 2 out at such a good price. Anything below $100 is going to hurt (BAD) when talking about 40 million chips (estimate of Netbust stock) How will that please investors?


I mean I have I only seen ONE DUAL CORE PC in best Buy, etc (a 3800+) so Sempron wil continue to rule. X2 3600+ is my new recommendation for the low end.
July 10, 2006 9:47:04 PM

I've said it time and time again, AMD is a server chip company first and foremost. That's where the high margins are. High margins make investors happy. They're making 4 core chips to better compete in the 1P and 2P space with intel's upcoming offerings. They've got the 4P space locked up tight.
July 10, 2006 9:48:46 PM

Quote:
If AMD and ATI have merged as reports say

then that changes the whole game

it would even make sense that AMD is dragging its feet on new products due to the Ati tie up...



i will say this ...and i have said many times...IF AMD-ATi merge ..then it means they going for INTELS throat...

they can afford to loose ground now for that long term goal ....



The ATi deal will AMD more because intel won't see it as them going for their throat BUT NVIDIA SURE WILL. nVidia made AMD with it's chipsets. That would make for alot of bad blood. BAD MOVE. they are in a great position now that Dell is selling Opteron. As people who wanted AMD AND Dell will demand desktops. That could mean 4% share INCREASE.

And HP is embracing the TurionX2 and it shoudl fit pretty well between Core and Core 2. AMDs long term goal is a FLEXIBLE platform based on the best ideas from the industry.


They dont' want to eclipse Intel, they just want to sell chips.

AMD doesn't want to eclipse Intel, they just want to sell chips... :lol:  :lol: 
That is BS and you know it... What you really mean is they can not eclipse Intel, and so will settle for please buy our chips even tho they are not the fastest nor cheapest!


if more comapnies were happy with their Lot in life there woul dbe a lot less a-holes workig for them. yo don't have to run people out of busines to be in business. There are probably close to 1 BILLION PCs/Servers/Laptops in the world. More than one company can be in the billions($) supplying them. i dont' KNOW what AMD is thinkign but from what they say, they re just happy to provide the service they do. Sure who wouldn't want Intel's share but they can feed their families on what they make so that should be enough.
July 10, 2006 9:51:02 PM

Quote:
When is AMD going to pull that rabbit out of their a$$e$?

That Reverse Hyper Threading should be getting the master switch flipped any day now, right? :lol:  :lol: 


Can you say Dell and HP and nVidia business platform?

How can Intel have been great in your eyes though they were getting their ass handed to them but now AMD is behind somewhat so they(AMD) shouldn't sell any?


You need help. Really.
July 10, 2006 9:52:09 PM

Are you implying that Intel is losing money?
Intel can manufacture chips cheaper than AMD and still make profits, period.
Are you implying AMD is sitting pretty in the face of this Conroe competition?
Is AMD in the Black? Or do they owe more than they are worth?
July 10, 2006 9:58:15 PM

Some how I think AMD has made some very long plans for it's processor roadmap. And I think right now it is sacrificing it's superiority as a chip producer to assure it's future. The New York fab should be enough prof of this, and also if the rumors about Ati are true, they should also confirm this.

Time will tell, but right now time is a real killer. I can really feel the pressure rising as Conroe launch day aproaches. Can't wait for some "genuine TomsHardware benchmarks" on this. :D 
July 10, 2006 10:14:44 PM

Quote:
Are you implying that Intel is losing money?
Intel can manufacture chips cheaper than AMD and still make profits, period.
Are you implying AMD is sitting pretty in the face of this Conroe competition?
Is AMD in the Black? Or do they owe more than they are worth?


AMD historycaly alway had better yield rates than Intel. But Intel had alot of fabs that's why they had lower costs for they processors, R&D costs spead over a lot more procs for Intel than for AMD.
BTW AMD's Fab 36 (opened in 2003) is the factory with the best yield rates in the world and also a record breaker using 300mm wafers.

Conroe is very expensve to produce due to it's very high transistor count and large cache and posibly not very good yield rates at the moment. AMD just halved the chache on it's processors wich means it can bloom production. Maybe one of the reasons behind the partenership with DELL.

Anyway they'r big companies and thy have equaly big secrets and plans for the future, and turnaaarounds have been before, this is why I wold't really bet my money on either company.
July 10, 2006 10:19:14 PM

Quote:
Are you implying that Intel is losing money?
Intel can manufacture chips cheaper than AMD and still make profits, period.
Are you implying AMD is sitting pretty in the face of this Conroe competition?
Is AMD in the Black? Or do they owe more than they are worth?


AMD historically alway had better yield rates than Intel. But Intel had a lot of fabs that's why they had lower costs for they processors, R&D costs spread over a lot more procs for Intel than for AMD.
BTW Amd Fab 36 (opened in 2003) is the factory with the best yield rates in the world and also a record breaker using 300mm wafers.

Conroe is very expensive to produce due to it's very high transistor count and large cache and possibly not very good yield rates at the moment. AMD just halved the cache on it's processors which means it can bloom production. Maybe one of the reasons behind the partnership with DELL.

Anyway their big companies and thy have equally big secrets and plans for the future, and turnarounds have been before, this is why I would really bet my money on either company.

Please link exactly where you found Intel's yield rates which you are mentioning... or AMD's yeild rates either...
I have yet to find any factual information alluding to that number... only amature speculation...
July 10, 2006 10:55:21 PM

Hi,

I don't see any intel CPU for 100 euros to beat AMD CPUs at the same price.
And I and many other people, offices, etc they do not buy the fastest processors, we/they buy the best for the less money possible.
So, i don't think thats really fair to discuss always and only the top of speed CPUs.
July 10, 2006 10:55:53 PM

Quote:
Please link exactly where you found Intel's yield rates which you are mentioning... or AMD's yeild rates either...
I have yet to find any factual information alluding to that number... only amature speculation...


This is a known fact for a while you can event read abou it in http://en.wikipedia.org/wiki/AMD (read under "Athlon / K7") AMD's history. I've also read about it in some magazines some years ago.

They have also declared it :
http://www.tgdaily.com/2005/04/19/behind_the_closed_doo...

"THG: Which brings us to the thorny subject of the yield rate. Could you comment on it? Our estimates assume greater than 50 percent?

Wallers: Our yield rate is very high compared with the competition. Right now we're in the same ballpark as the top chip manufacturers. We don't want to reveal details."
July 10, 2006 11:02:00 PM

All I know is speculation, and it appears Intel's yeild rate is significantly higher than AMD's is...
July 10, 2006 11:36:30 PM

Quote:
have done alot of digging....

2006

and AMD has no answer for the CORE 2 as of now


4x4 is proof of this.... this is just a face saving manuever to get hardcore AMD fanbois with the cash to buy an AMD that will rule in some tests.

AMD is slow with .65 tech..

AMD will be able to battle INTEL on mid to high level on the server side ...but thats about it.

dekstop side is lost to the CORE 2 DUO hordes


2007

AMD will still be behind the eight ball on most fronts....AMD will slash prices deep... with no answer to CONROE other than a ridicoulously costly 4x4 setup.

there will be no K8L saviour in 07. not on the desktop side anyway

Intel should pretty much run unabaited through 2007.... AMD has no real answer.

2008

Enter AMD's answer... the long awaited K8L

will it be enough.... AMD will not be resting on its laurels waiting on the K8L to come knocking... CORE 2 DUO will be stronger

and @ 45um


For the forseeable future its an INTEL world

2010

K10 is born.... no further info


So you are seeing the light. I have to say I am disappointed that AMD isn't going to be able to do anything to Intel for some time. But alas they pissed them off and made Intel look very bad for 3 years.

Then with Dell using AMD parts the monopoly appears to be nonexistent and the price wars that will remain will leave AMD where they were before 2 Fabs and a leased office building.

I think everyone can agree here, it will be a brutal few years for AMD as Intel goes full force.
July 11, 2006 12:03:23 AM

BS, quote some real numbers.
July 11, 2006 12:59:44 AM

Quote:


I think everyone can agree here, it will be a brutal few years for AMD as Intel goes full force.


Everyone? Hmm... I think the Baron would disagree.

Quote:
Most people in the industry agree that AMD will continue to increase share even in the face of Core 2 which won't be the majority of Intel's shipments until Q2 07.


AMD got Intel, just where they want them. LOL

Sorry, Baron - even theinquirer wouldn't print that.

I commend your loyality, and I hope AMD pulls it out.
I just see them becoming a leaner company in order to compete with
Intel. Thereby, maintaining a good reputation, while R&D comes up with
the answers.
July 11, 2006 2:23:22 AM

Quote:
have done alot of digging....

2006

and AMD has no answer for the CORE 2 as of now


4x4 is proof of this.... this is just a face saving manuever to get hardcore AMD fanbois with the cash to buy an AMD that will rule in some tests.

AMD is slow with .65 tech..

AMD will be able to battle INTEL on mid to high level on the server side ...but thats about it.

dekstop side is lost to the CORE 2 DUO hordes


2007

AMD will still be behind the eight ball on most fronts....AMD will slash prices deep... with no answer to CONROE other than a ridicoulously costly 4x4 setup.

there will be no K8L saviour in 07. not on the desktop side anyway

Intel should pretty much run unabaited through 2007.... AMD has no real answer.

2008

Enter AMD's answer... the long awaited K8L

will it be enough.... AMD will not be resting on its laurels waiting on the K8L to come knocking... CORE 2 DUO will be stronger

and @ 45um


For the forseeable future its an INTEL world

2010

K10 is born.... no further info



What you seem to forget is that AMD always concentrates on SERVER first. That's why Woodcrest won't even TRY 4Way. Also, why drop prices? Core 2 won't ame a splash until next year and if AMD continues their mad power cutting they will be at 30W for special case mid2007 chips and all by the end of next year.


power consideratiosn will help AMD through this. There are still millions of NetBust chips out there and now no one really wants them.

I won't speculate as to what chip they'll release when but I wll say that Intel better enjoy the lead while they can. I'm confident that AMD is not concerned about top performance right now except in high end servers.

Remember that K8 was Opteron first. that's why I had no troubl believing K8L would be next year. the smart move woul dbe to design thsi new arch around 45nm an intro the 65nm L next year. It's not like it won't be worth anything in 08.

09 will see 8 core Opteron based on DeerHound with HT4 at 10.4 Gt/s. we'll see though.

I don't think they should try to match Intel's 2year core timetable. Just continue to make good partnerships ( especially by not buying or merging with ATi) and execute in their roadmap.

I've said ti before and I'll say it again Intel had more than two years with CRAP, AMD gets two years, though K8 isn't crap.
...compare to Conroe it is...
July 11, 2006 2:39:28 AM

Quote:
Are you implying that Intel is losing money?
Intel can manufacture chips cheaper than AMD and still make profits, period.
Are you implying AMD is sitting pretty in the face of this Conroe competition?
Is AMD in the Black? Or do they owe more than they are worth?


AMD historycaly alway had better yield rates than Intel. But Intel had alot of fabs that's why they had lower costs for they processors, R&D costs spead over a lot more procs for Intel than for AMD.
BTW AMD's Fab 36 (opened in 2003) is the factory with the best yield rates in the world and also a record breaker using 300mm wafers.

Conroe is very expensve to produce due to it's very high transistor count and large cache and posibly not very good yield rates at the moment. AMD just halved the chache on it's processors wich means it can bloom production. Maybe one of the reasons behind the partenership with DELL.

Anyway they'r big companies and thy have equaly big secrets and plans for the future, and turnaaarounds have been before, this is why I wold't really bet my money on either company.

This one is just too juicy to pass up, I will return to thoroughly trash this post. Digging the links now.

I wish I had the perseverance (and time; I'm a crazed-college student, ahh!) that JumpingJack has to post a rebuttal to these baseless statements made by Cryogenic, so I will leave it to JumpingJack to find the numbers/figures/facts. But there are some things that are just ridiculous that Cryo's said:

1. "AMD historycaly alway had better yield rates than Intel"
WRONG!

2. "BTW AMD's Fab 36 (opened in 2003) is the factory with the best yield rates in the world and also a record breaker using 300mm wafers."
Okay, I will give you the best yield rates in the world statement, but how is it a record breaker using 300mm wafers? Intel did this YEARS ago and has 300mm wafers in MULTIPLE fabs across multiple CONTINENTS.

3. "Conroe is very expensve to produce due to it's very high transistor count and large cache and posibly not very good yield rates at the moment."
There is no indication of this ANYWHERE. I don't believe that Intel has published any numbers indicating transister count or size, but the most commonly believed figure is 150 mm^2 for the die size. This is SIGNIFICANTLY lower than AMD's X2 chips, which stands at 230mm^2 at a 90 nm process. There indeed is a reason that AMD has cut their cache on ALL X-2s from 2x1mb to 2x512KB...

4. "AMD just halved the chache on it's processors wich means it can bloom production."
Read the above. AMD has LOWERED predicted revenue forecasts, there is no reason to believe that they wish to increase production. Most analysts see this move as a reason to believe that yields are not what they had expected and AMD needs to halve the cache so that there is a smaller chance for errors to crop up; why can YOU not assume this as well??

Oh boy, there is just too much to say.
July 11, 2006 3:04:09 AM

For another's perspective; Motley Fool's analysis

That article was cited here, AMD Nomentum

We do not need to look to 2010 but we will know outcome of the war within 2 years. By the way Intel wants to be at 32 core CPU by 2010.Link to DailyTech article

It doesn't look good for AMD, financially or technologically.
July 11, 2006 3:10:00 AM

i dont know about you guys... but that end of the world crap prediction inserted at the last part was really funny :lol: 
July 11, 2006 3:16:57 AM

Here is my prediction:

AMD will sell cpus.
July 11, 2006 3:20:27 AM

Quote:
Are you implying that Intel is losing money?
Intel can manufacture chips cheaper than AMD and still make profits, period.
Are you implying AMD is sitting pretty in the face of this Conroe competition?
Is AMD in the Black? Or do they owe more than they are worth?


AMD historycaly alway had better yield rates than Intel. But Intel had alot of fabs that's why they had lower costs for they processors, R&D costs spead over a lot more procs for Intel than for AMD.
BTW AMD's Fab 36 (opened in 2003) is the factory with the best yield rates in the world and also a record breaker using 300mm wafers.

Conroe is very expensve to produce due to it's very high transistor count and large cache and posibly not very good yield rates at the moment. AMD just halved the chache on it's processors wich means it can bloom production. Maybe one of the reasons behind the partenership with DELL.

Anyway they'r big companies and thy have equaly big secrets and plans for the future, and turnaaarounds have been before, this is why I wold't really bet my money on either company.

This one is just too juicy to pass up, I will return to thoroughly trash this post. Digging the links now.

i agree, this post is very amusing. No where does he give proof about either companies yields.

1. while transistor count may go up on conroe, the overall die size is very small which allows higher yields. (a single defect affects a smaller population, pretty easy to understand, 1 failure/200 parts is greater than 1 failure/400 parts).

2. While large cache size can increase CPU size, cache is a very simple and therefor redundant design. This means that unlike normal logic, any failures can be accounted for in design. This means a cache actually has LESS failures than normal logic.

3. Does AMD test their units to the same extent that Intel does and the other way around? Does one company test their CPUs more extensively than the other?
July 11, 2006 3:27:21 AM

Baron corrected me on this... 8O AMD will sell chips and chipsets, heck, even the whole silicon wafer, (currently working on multi-colored ones, oooh! :p  ) sort of diversefy if you will, and this has been known from way back into this quarter, while Intel is cornering themselves as only a chip company :? really sad... Think of how much money they are losing due to the P4 pricing and low initial Conroe pricing... humiliating too... :oops:  [/sniffs] Would be nice if they would make em fast, I am really disappointed to see Conroe's running slower than AMD chips... 1.66GHz to 2.67GHz is the best you can do Intel? AMD makes em up to 3GHz... but I guess they have to start small... :roll:
July 11, 2006 3:44:39 AM

but if I had to guess the answer, Intel would be the one getting higher yeilds due partly to thier extensive experience and design techniques... not even to mention profits...
July 11, 2006 4:04:34 AM

Quote:
but if I had to guess the answer, Intel would be the one getting higher yeilds due partly to thier extensive experience and design techniques... not even to mention profits...


I would tend to agree -- however, as I am working to discredit his statements, I cannot in fairness and honest postively conclude Intel's are highest without the actual data, thus I will take the position that it remains unknown :) 

While i am not sure of the actual yield numbers, both companies would be wise to invest as much as possible into yields (both design and testing).
July 11, 2006 4:06:14 AM

Whom do you think is spending more in the areas of design and testing?
July 11, 2006 4:19:20 AM

Quote:
Whom do you think is spending more in the areas of design and testing?


Unfortunately that information is kept very close to the chest by both companies, so that comparison is not possible. There are so many factors that play into this. Design for test and redundancy in design, testing at various conditions, optimizing test for yield/demand, and overtest/undertest. Factors such as design, test socket types, and test time, and test coverage all play a huge roll in this.
July 11, 2006 4:39:39 AM

Quote:
but if I had to guess the answer, Intel would be the one getting higher yeilds due partly to thier extensive experience and design techniques... not even to mention profits...


I would tend to agree -- however, as I am working to discredit his statements, I cannot in fairness and honest postively conclude Intel's are highest without the actual data, thus I will take the position that it remains unknown :) 

While i am not sure of the actual yield numbers, both companies would be wise to invest as much as possible into yields (both design and testing).

Yield, above all, is the difference between profit and loss. I completely agree -- both companies seek new and every ingenious ways to increase yield.

Yield is also the key to planning and scheduling, so that the companies knows that amount of WSPW to meet orders and correctly load the fabs. This is extremely difficult as there are so many factors that affect the yield and the yield is always changing. With the length of time it takes from wafer start to having a product, these predictions need to be fairly accurate.
July 11, 2006 5:01:27 AM

Quote:
but if I had to guess the answer, Intel would be the one getting higher yeilds due partly to thier extensive experience and design techniques... not even to mention profits...


I would tend to agree -- however, as I am working to discredit his statements, I cannot in fairness and honest postively conclude Intel's are highest without the actual data, thus I will take the position that it remains unknown :) 

While i am not sure of the actual yield numbers, both companies would be wise to invest as much as possible into yields (both design and testing).

Yield, above all, is the difference between profit and loss. I completely agree -- both companies seek new and every ingenious ways to increase yield.

Yield is also the key to planning and scheduling, so that the companies knows that amount of WSPW to meet orders and correctly load the fabs. This is extremely difficult as there are so many factors that affect the yield and the yield is always changing. With the length of time it takes from wafer start to having a product, these predictions need to be fairly accurate.

See -- you are thinking along the correct lines. In fact, it is a plight both AMD and Intel suffer from because they need to anticipate market demains weeks or even months ahead of time in order to load the factories correctly, with the right product mix.

Actually, with a unified architecture over the last 3 years, AMD has had an avantage in this area since, with a unified arch, a company can dynamically apportion wafers in works to the appropriate product mix as market demans fluxuate and change, Intel will now be able to do this too.

You brought up a good point.

Jack

While i do agree somewhat with the unified architecture, i do not think that had as much of an impact on Intel. Unless you are doing core disabling/cache disabling, then once you have the masks, it is just as hard/easy to change wether its a unified architecture or not. But it will make the masks easier to make for the different versions since no architecture changes are being made.
July 11, 2006 7:06:49 AM

WOW!!! That was the most thorough job i have seen very informative and a great read. I registered to this forum just so i can post that jumping jack you are the man real impressive stuff!
July 11, 2006 7:09:58 AM

Quote:
Third and final installment ---

Core 2 large cache, by your argument, will yield low. This is rubbish for two reasons. First is a small concept called 2 meg shared cache which is nothing more than a 4 meg cache with large defects in the SRAM with 2 megs deactivated for the defective blocks. The second is a small but amazing term called redundancy.

Intel or AMD for that matter create their processors with more than the specificed cache -- why? Because there are sitatuons where it would ba shame for one small particle screwing up on small bit in the SRAM to render the CPU useless.

This master thesis has a nice blurb (with real reference, not some interview in THG with some AMD dude with AMD signing the paycheck). Read through the introduction for an overview, he also does his reasearch and analysis on defect and yields for SRAM:

http://etd.lib.ttu.edu/theses/available/etd-11212005-13...

Sram is constructed in very regular arrays, called columns and rows, in fact, SRAM is indexed for tracelines based on the word line and bit line when it is actually read or written to. If a block or section (row(s) or column(s)) are defective (be it systematic or random particles), then the block is simply deactivated and the cache table re-indexed. There are several papers available that you can read on yield analysis, maximizing yield, and the concept of redundancy to allow an otherwise non-functional section to be "hidden" and save a die rather than throwing it away. This paper discusses some of the yield implications of redundant cache:
http://www.iccd-conference.org/proceedings/1999/0406029...

Here are a few more links to understand and comprehend yield:
http://www.gigascale.org/pubs/873/TVLSI_bestPaper_Jan05...
http://www.micromagazine.com/archive/00/01/segal.html (AMD authored discussion of reducing random defects).
http://research.cs.tamu.edu/eda/papers/sbala97.pdf (here is one dealing with packaging, remember Intel has the most advanced packaging technology in the world, AMD is just now catching up).


So overall, even with a large cache, the die size of Core 2 Duo is nominally small, not much larger than a Presler die, certainly smaller than a Athlon die at 90 nm. 65 nm for AMD will drive die size smaller and why it is critical to AMD to get to 65 nm. It is so critical, they will even be willing to release/produce on 65 nm even when the bin splits give such poor speeds (see recently published roadmap). Die size is also the reason AMD discontinued the 1x2 meg processors, relying on the IMC to keep performance in line and giving up on the larger dice altogether. AMD's behavior is one of a company looking for ways to cut costs no matter what, makes sense as they now have no other method for competing other than price.

They have lost performance, they have lost performance/watt majorly, and even with the price cuts they will lose price/performace and combined price/performance/watt -- well, let's say they really need K8L before 2008.

The level of detail I provided here is actually quite brief, the subject of yield is so important in the industry that volumes have been written on it. In fact, a small company essentially grew into a major empire in the equipment supplier area focusing only on defect detection for yield, that company was called Tencor (there is an interesting story on the Tencor name). Tencor later merged with KLA and is what we know today as KLA-Tencor.

However, the next time you are requested to support your statement a link to wikipedia and an interview of an official of the company touting the company prowess will not cut the mustard.


Jack


Cache redundancy is a great thing. Takes up a little extra die size and greatly improves yield. It really makes cache disabling almost unnecessary. You would need to have many defects in the cache in order to create enough failures that it can't be fixed using redundancy. Smaller cache versions of a CPU are usually done to reduce die size as long as there is a market segment that will purchase the cheaper version.
!