Dissappointing Year So Far for AMD

Status
Not open for further replies.
duby229 said:
Not so, becouse all you have to do is look at die size. 143^2mm*2 compared with 284^2mm Provided that both companies have the same yeild, they will both be able to produce the same number of quad cores from a wafer or pretty close to it, plus AMD has the added benefit of binning some defects as X3''s.

In the situation where yields are the same AMD's products do in fact cost less.
Click to expand...

Yes, but reality is that MCM gives better yields, so AMDs don't cost less, they cost more.
That's the part you don't want to get.
The same number of defects per wafer gives AMD less working quadcore than Intel, as has already been established. Hence, the yields will NOT be the same, they'll be in Intels favour.
So you are hanging up your story of "AMD costs less" on the assumption that they have equal yields, which they don't.
Which is not saying that Intel has better production technology... Not at all, even with the same number of defects per wafer, this would be true. Intel just has smarter production technology. With the same number of working transistors per surface area, Intel gets more working quadcore CPUs. As simple as that.

What you're saying is "Intel must have more defects per wafer, else it's not fair to AMD", which is ofcourse a load of BS.
 
Scali2 said:
Yes, but reality is that MCM gives better yields, so AMDs don't cost less, they cost more.
That's the part you don't want to get.
The same number of defects per wafer gives AMD less working quadcore than Intel, as has already been established. Hence, the yields will NOT be the same, they'll be in Intels favour.
So you are hanging up your story of "AMD costs less" on the assumption that they have equal yields, which they don't.
Which is not saying that Intel has better production technology... Not at all, even with the same number of defects per wafer, this would be true. Intel just has smarter production technology. With the same number of working transistors per surface area, Intel gets more working quadcore CPUs. As simple as that.

What you're saying is "Intel must have more defects per wafer, else it's not fair to AMD", which is ofcourse a load of BS.
Click to expand...

intel wafer = 100 cpus
amd wafer = 50 cpus
so 25 defects per = 25 working quad cores vs 37 working quad cores, but, intels dead cores are a complete write off, AMDs say 1 core is pooched, will at least be able to be resold as Tri-Core, I Don't get why you say this is bad? sure it's not IDEAL, but its not bad and might end up being better then having a dead single core chip, unless intel can make a single core chip out of their dead dual cores(what is that a celeron I guess?)

so it turns in to
intel 100cpus - 25 defects = 37 quads and 1 dual
amd 50cpus - 25 defects = 25 quads with 15 tri cores and 5 dual cores

Intel 37 quads @ $300 = $11,100
Intel 1 Dual Core @200 = $200
Intel 20 celerons @ 75 = $ 1,500
= $12,800 with 5 non-usable cpus

AMD 25 quads @ $300 $7,500
AMD 15 tri @ $225 $3,375
AMD 5 Duals @ $150 $0,750
= $ 11,625 with 5 non-usable cpus

so in that given scnario, AMD really isn't losing so much in this case, and you have no idea on how good INTEL or AMD has the "yields"

Also AMD chips have far less cache, which is pretty expensive right?

so you're lookin @ 4m per quad vs 8m per quad :|
 
Digital Viper-X- said:
intel wafer = 100 cpus
amd wafer = 50 cpus
so 25 defects per = 25 working quad cores vs 37 working quad cores, but, intels dead cores are a complete write off, AMDs say 1 core is pooched, will at least be able to be resold as Tri-Core, I Don't get why you say this is bad? sure it's not IDEAL, but its not bad and might end up being better then having a dead single core chip, unless intel can make a single core chip out of their dead dual cores(what is that a celeron I guess?)

so it turns in to
intel 100cpus - 25 defects = 37 quads and 1 dual
amd 50cpus - 25 defects = 25 quads with 15 tri cores and 5 dual cores

Intel 37 quads @ $300 = $11,100
Intel 1 Dual Core @200 = $200
Intel 20 celerons @ 75 = $ 1,500
= $12,800 with 5 non-usable cpus

AMD 25 quads @ $300 $7,500
AMD 15 tri @ $225 $3,375
AMD 5 Duals @ $150 $0,750
= $ 11,625 with 5 non-usable cpus

so in that given scnario, AMD really isn't losing so much in this case, and you have no idea on how good INTEL or AMD has the "yields"

Also AMD chips have far less cache, which is pretty expensive right?

so you're lookin @ 4m per quad vs 8m per quad :|
Click to expand...

Your example is wrong.
You are only looking at Intels budget quadcore here (which in terms of Intel yields is equal to the tricore or even dualcore of AMD... It's made from 'leftover' dies that couldn't be binned higher... Currently Intel is basically limiting its quadcore production because they don't use all their dies for quadcores, they still put out many dualcore processors).
Intel also has quadcores of over $1000.
AMD's yields are so bad that they don't have ANYTHING on this process that is worth more than $300. That is what hurts them most.
The tricores may make up some of that, but not much.

As for cache being expensive... That's all about yields. Intels production technology simply allows them to get away with much more cache. It probably doesn't cost Intel more to put 8 mb on a quadcore than it costs AMD to put 4 mb on a quadcore at this point.
 
duby229 said:
Not so, becouse all you have to do is look at die size. 143^2mm*2 compared with 284^2mm Provided that both companies have the same yeild, they will both be able to produce the same number of quad cores from a wafer or pretty close to it, plus AMD has the added benefit of binning some defects as X3''s.

In the situation where yields are the same AMD's products do in fact cost less.
Click to expand...

You need to start living in reality, rather than make believe wishful thinking land.

Intel is the process king. They are always at least a half a generation ahead on the process side. So at any given time in the same process they will have less defects as defects go down as the process matures. Right now 65nm is near end of cycle for Intel and beginning-middle of cycle for AMD. Intel Will have less defects. That is reality. Get used to it.

Even in your fantasy land of communist fabs with equal defects for all. Intel's way has fabrications advantages.

Intel can speed bin much more easily for high end parts. It needs a 2 out of 2 core matchups. For AMD to build a top bin quad it needs all 4 core to be high speed capable. A 4/4 match. Much less likely.

Also your assumption about AMD benefiting from just being able to disable a core and sell as a Tri misses a few things:

- By the time Intel build quads it already has tested that all the cores work, duals with a dead half can be reused as single cores as well.
- Less than half the chip surface is core anyway, you have half the space devoted to common cache and control logic, many fab defects will kill the whole chip, not just a core. Chip death is twice as expensive for AMD because it is losing double the area.

Finally unless you were asleep you should have noticed that Intel has executed nearly flawlessly on 45nm Penryns, running cooler and at very high clocks on a smaller die, while AMD is struggling to execute it's 65nm quad.

You really need to be a deluded fanboy to see this as other than a complete SNAFU for AMD, and I am no Intel fanboy, I just put together a new system and I went Intel because it was clear that Barcy was pooched. Before now, the last Intel I had was a Pentium Pro. So I previously bought my last Intel in the mid-90's. I don't choose sides. I choose reality. Some of you should try it.
 
Another thing that hasn't been addressed yet is the desirability of tri-cores.
If Intel keeps lowering its quadcore prices, they'll be in the budget-market soon, and then dualcores and tricores will not be very desirable anymore, just like hardly anyone buys a singlecore now... Why would you, when only a few bucks more gives you twice the cores?
The lower prices get, the closer you get to the actual production costs in terms of wafers and packaging and all that.
Then there won't be any savings from turning off broken cores, it will just cost you.
 
duby229 said:
Well, you cant give Intel the best case scenaro, and give AMD the worst.... As this is exactly what your trying do by claiming that the defect ratio is the same across both companies. If AMD and Intel had the same defect ratio AMD would end up with something like a 50% yeild, which you and I both know is BS.

By the way guys, happy thanksgiving... Enjoy the turkey feast. :D
Click to expand...


Happy Thanksgiving to you to duby ! :) Oh,and everyone else here thats celebrating !


When Paul puts the word out to cut some of the quadcore pricing to 150,what will that to do to quad core asp's,nevermind tri core ASP's ? Send them into the toilet thats what.

What will a Phenom's go for next Summer when a Q9300 is under 150$ ? Plentiful,cool running and extremely overclockable to boot ??

AMD needs to ramp 45nm NOW,but does not have the cash to tool up the fab fast enough.They need real engineering miracles right now,and a sugar daddy.

http://www.fabtech.org/content/view/3868/
 
Scali2 said:
Yes, but reality is that MCM gives better yields, so AMDs don't cost less, they cost more.
That's the part you don't want to get.
The same number of defects per wafer gives AMD less working quadcore than Intel, as has already been established. Hence, the yields will NOT be the same, they'll be in Intels favour.
So you are hanging up your story of "AMD costs less" on the assumption that they have equal yields, which they don't.
Which is not saying that Intel has better production technology... Not at all, even with the same number of defects per wafer, this would be true. Intel just has smarter production technology. With the same number of working transistors per surface area, Intel gets more working quadcore CPUs. As simple as that.

What you're saying is "Intel must have more defects per wafer, else it's not fair to AMD", which is ofcourse a load of BS.
Click to expand...


No, I'm sayng that defect ratio --can not-- be the same across both companies. If they were AMD would end up with about a 50% yield, and we both know that is BS. You base your entire argument that both companies have the same defect ratio, when in fact that simply isnt possible. A 50% yield is not sustainable./ AMD would have chosen to scrap this project then go through with that suicide. And what I said is that it isnt fair for --you-- to claim the same defect ratio, when --you-- know it is BS. It's the same as lying...

Besides you --once again-- choose to ignore APM. jeez. It's really amazing how much you can twist the truth by ignoring facts.
 
Intel is turning a profit, and AMD isn't. That fact should make everything clear.
 
Dan_D said:
Intel is turning a profit, and AMD isn't. That fact should make everything clear.
Click to expand...

Yep because AMD's debt is exactly how you determine the cost of it's products.....jeez if only we had all thought of that.
 
Scali2 said:
Another thing that hasn't been addressed yet is the desirability of tri-cores.
If Intel keeps lowering its quadcore prices, they'll be in the budget-market soon, and then dualcores and tricores will not be very desirable anymore, just like hardly anyone buys a singlecore now... Why would you, when only a few bucks more gives you twice the cores?
The lower prices get, the closer you get to the actual production costs in terms of wafers and packaging and all that.
Then there won't be any savings from turning off broken cores, it will just cost you.
Click to expand...

I think tri-cores will sell really well in the low-end market for a simple reason. For one, it is not "only a few bucks more" for a quad core vs. a dual core. Cheapest Intel Core 2 Quad (at newegg): Q6600: $265, cheapest Intel Core 2 Duo (at newegg): E4500 $130. That is a *HUGE* price difference. A tri-core for, say, $150 would fill the gap nicely. Sure, it may get outperformed by a Core 2 Duo, but average joe will think 3 cores > 2 cores, and companies like Dell, HP, etc.. all know that very well. Of course, AMD is still "winning" on price alone as the Phenom 9500 is $5 cheaper than the Q6600 and the X2 3600 is a mere $50. So if we compare cheapest dual vs. cheapest quad there is a staggering 520% increase in price. Only a "few bucks more"? I don't think so.
 
kllrnohj said:
I think tri-cores will sell really well in the low-end market for a simple reason. For one, it is not "only a few bucks more" for a quad core vs. a dual core. Cheapest Intel Core 2 Quad (at newegg): Q6600: $265, cheapest Intel Core 2 Duo (at newegg): E4500 $130. That is a *HUGE* price difference. A tri-core for, say, $150 would fill the gap nicely. Sure, it may get outperformed by a Core 2 Duo, but average joe will think 3 cores > 2 cores, and companies like Dell, HP, etc.. all know that very well. Of course, AMD is still "winning" on price alone as the Phenom 9500 is $5 cheaper than the Q6600 and the X2 3600 is a mere $50. So if we compare cheapest dual vs. cheapest quad there is a staggering 520% increase in price. Only a "few bucks more"? I don't think so.
Click to expand...



AMD cannot survive long term by trying to be the Walmart of the X86 space though.45nm fab equipment is tremendously expensive,as will 32nm be even more expensive in turn.

Making lemonade out of lemons is fine,but when each tree costs billions,and only last a few years before it whithers and dies,and you need another one,at the cost of several hundered million more...

Not to mention your main competitor has a grove full of healthy trees.Not the best analogy,but certainly you see what I mean.
 
madgun said:
Now if AMD decides to buy AGEA, where is the cash going to come from?
Click to expand...

I think AMD would like to buy that company, but I dont think they have the money to do it, and even if they could scrounge it up, I doubt very much the board would allow it.
 
madgun said:
Now if AMD decides to buy AGEA, where is the cash going to come from?
Click to expand...

They'll take out another loan, or sell something to get the money, or offer some kind of trade for them. AMD will no doubt be able to convince some bank of some company to loan them the cash for the promise of turning AMD around based on technology they'll aquire from Ageia.

duby229 said:
As does your, clearly by the intention of the post I responded to...
Click to expand...

You always side with AMD/ATI no matter what the discussion is about. You twist facts and slant everything to support your favorite company and you can't or won't see the truth if it goes against them in some way. I've seen too many threads where you've posted nothing but fanboi comments. You outright attack others anytime they point out a failing of AMD/ATI. If you think AMD doesn't have a higher defect rate than Intel in regard to manufacturing you are incredibly naive. Intel's processors may or may not cost more to make, but they aren't losing money due to poor yields like AMD probably is. Intel's 65nm process is solid and from the looks of things they have the 45nm process nailed down pretty well too.

duby229 said:
I think AMD would like to buy that company, but I dont think they have the money to do it, and even if they could scrounge it up, I doubt very much the board would allow it.
Click to expand...

AMD is good at accumulating debt. I have no doubt they can find a way to do it.
 
Manny Calavera said:
AMD cannot survive long term by trying to be the Walmart of the X86 space though.45nm fab equipment is tremendously expensive,as will 32nm be even more expensive in turn.

Making lemonade out of lemons is fine,but when each tree costs billions,and only last a few years before it whithers and dies,and you need another one,at the cost of several hundered million more...

Not to mention your main competitor has a grove full of healthy trees.Not the best analogy,but certainly you see what I mean.
Click to expand...

Oh, no doubt, I totally agree, and I would love to see AMD turn profits and be able to maintain constant competition in all market segments. I was more trying to remind Scali that $260 for a quad core is definitely NOT only "a few bucks more" than a dual core, and that tri-core, if executed correctly, could be well positioned to score AMD some "easy" money by exploiting the more mid-range market segment.
 
kllrnohj said:
Oh, no doubt, I totally agree, and I would love to see AMD turn profits and be able to maintain constant competition in all market segments. I was more trying to remind Scali that $260 for a quad core is definitely NOT only "a few bucks more" than a dual core, and that tri-core, if executed correctly, could be well positioned to score AMD some "easy" money by exploiting the more mid-range market segment.
Click to expand...



I'll agree that on the very low end,AMD has some truly excellent chips (from a gamers perspective.)
 
kllrnohj said:
Oh, no doubt, I totally agree, and I would love to see AMD turn profits and be able to maintain constant competition in all market segments. I was more trying to remind Scali that $260 for a quad core is definitely NOT only "a few bucks more" than a dual core, and that tri-core, if executed correctly, could be well positioned to score AMD some "easy" money by exploiting the more mid-range market segment.
Click to expand...

I have to agree on this.

Really AMD is doing tri-cores so they don't have to throw away failed quad core CPUs. It's much better for them to sell these processors rather than throw them away.

As I understand it the reason they are selling triple core processors instead of castrating the third usable core and selling them as duals is because they can sell it with three cores without much of an impact to overclocking or power usage and it will provide extra performance in some applications. It's a win for AMD and a win for the customers.
 
duby229 said:
No, I'm sayng that defect ratio --can not-- be the same across both companies. If they were AMD would end up with about a 50% yield, and we both know that is BS. You base your entire argument that both companies have the same defect ratio, when in fact that simply isnt possible.
Click to expand...

No, your 50% yield figure is just complete nonsense. What the heck are you basing that 50% on? And again, what do you mean EXACTLY when you say 'yield'?
It is highly likely that the defect ratios between Intel and AMD are not that far apart. After all, they use very similar production technologies. It is also not impossible at all that they are equal. It's probably never exactly equal, because it depends on many factors, but in theory it could be.

Basically because you cling to that 50% figure of yours, you have to conclude that AMD has a lower defect ratio than Intel, which we all know is BS :p

Also, even if AMD does have a 50% defect ratio, they wouldn't scrap the project. They can't, at this time. They need quadcores. As you said yourself, some of the defects can be sold as X2 and X3. Other than that, the yields will gradually go up to more acceptable levels as the production matures.
They probably DID have a very high defect ratio anyway, if we see that they had to keep dropping clockspeeds, and eventually even pull the 2.4 GHz because of stability problems.
 
kllrnohj said:
Oh, no doubt, I totally agree, and I would love to see AMD turn profits and be able to maintain constant competition in all market segments. I was more trying to remind Scali that $260 for a quad core is definitely NOT only "a few bucks more" than a dual core, and that tri-core, if executed correctly, could be well positioned to score AMD some "easy" money by exploiting the more mid-range market segment.
Click to expand...

I was talking about Intel cutting prices so quadcores would also become sub-$200 parts, and THEN tricores will be hard to sell.
They have their place NOW, but they aren't available NOW, and by the time they are, Intel may have even cheaper quadcores on offer. 45 nm will most probably mean pricecuts. Perhaps also lower clocks than 2.4 GHz, since Intel now sees that AMD can't even reach those.
 
Scali2 said:
I was talking about Intel cutting prices so quadcores would also become sub-$200 parts, and THEN tricores will be hard to sell.
They have their place NOW, but they aren't available NOW, and by the time they are, Intel may have even cheaper quadcores on offer. 45 nm will most probably mean pricecuts. Perhaps also lower clocks than 2.4 GHz, since Intel now sees that AMD can't even reach those.
Click to expand...



I can see the Q9300 going for 150 USD @ the egg in 6 months time easily.You want to talk about putting the hurt on AMD ? I wouldnt put it past Intel,Ottelini has openly talked about such deep cuts recently.It seems to be Intels plan,to slowly wear them down.First
the 266$ Q6600 well before Phenom ships,then this as another powerful body blow.
 
duby229 said:
Not that bad though, what your saying is that AMD has a yield right around 50%. You and I both know that is BS. If it was that low they would have scrapped it, and wouldnt have even bothered trying. Yields that low are simply not sustainable. Shit anything less the 70% is hard to sustain. The truth is that it's probably alot closer to 85%. But nobody knows for sure. The best we can do is put both companies on equal footing and say they have the same yield. In this case AMD's monolithic core costs less.
Click to expand...

You're out for lunch as always.Please allow me to educate you a bit :

-1. AMD's yields were around ~50% for their very mature 90nm process node ( from the looks of it , 65nm has yet to reach those yields )
-2. AMD published a DD number of <0.5/cm^2 , which is bad considering that Intel claims world class ( DD of 0.22-0.24 )

Links :
1. http://www.fabtech.org/content/view/1390
2. http://investorshub.advfn.com/boards/read_msg.asp?message_id=23441379
http://www.planet3dnow.de/vbulletin/showthread.php?t=314639&garpg=16#content_start
http://siliconinvestor.advfn.com/readmsg.aspx?msgid=23745158
 
Manny Calavera said:
I can see the Q9300 going for 150 USD @ the egg in 6 months time easily.You want to talk about putting the hurt on AMD ? I wouldnt put it past Intel,Ottelini has openly talked about such deep cuts recently.It seems to be Intels plan,to slowly wear them down.First
the 266$ Q6600 well before Phenom ships,then this as another powerful body blow.
Click to expand...

Yes, and the Q6600 worked both ways.
People who didn't want to wait, bought Intel instead of AMD.
People who did want to wait, now feel like AMD has flatout lied to them about their upcoming products (claiming 40% higher speeds than Clovertown in general, while you can't even keep up with the SLOWEST quadcore Intel offers... well that's the biggest lie I've ever heard in the computer industry). I think they've alienated many of their customers by that move. Not just consumers... what about Cray? Their orders have been delayed, delayed, delayed... and they're not getting the performance they anticipated.
 
savantu said:
You're out for lunch as always.Please allow me to educate you a bit :

-1. AMD's yields were around ~50% for their very mature 90nm process node ( from the looks of it , 65nm has yet to reach those yields )
-2. AMD published a DD number of <0.5/cm^2 , which is bad considering that Intel claims world class ( DD of 0.22-0.24 )

Links :
1. http://www.fabtech.org/content/view/1390
2. http://investorshub.advfn.com/boards/read_msg.asp?message_id=23441379
http://www.planet3dnow.de/vbulletin/showthread.php?t=314639&garpg=16#content_start
http://siliconinvestor.advfn.com/readmsg.aspx?msgid=23745158
Click to expand...

Oh dear, actual numbers! From reputable sources! (eg AMD itself)
I wonder how duby will brush those off :)
 
duby229 said:
It's been a long time since the last time I got to see a paid for rant like this. How much did you make? Did they pay you per word? Or per topic?
Click to expand...

Wow! Twisted fanboy logic: "that scary reality comments must be paid for by Intel". Geez man, I think you need some Lithium or something. Why not try addressing the points instead of saying Intel is paying me to say this stuff (which is insane paranoia BTW).

I watch this thread because of the ongoing train wreck of Duby's ranting fanboy paranoia. It actually amazes me when I see people so divorced from reality.

Do you have a big AMD stock holding or something? Why are you working so hard to convince the world that AMD is peachy keen land.

Wow!
 
Scali2 said:
Your example is wrong.
You are only looking at Intels budget quadcore here (which in terms of Intel yields is equal to the tricore or even dualcore of AMD... It's made from 'leftover' dies that couldn't be binned higher... Currently Intel is basically limiting its quadcore production because they don't use all their dies for quadcores, they still put out many dualcore processors).
Intel also has quadcores of over $1000.
AMD's yields are so bad that they don't have ANYTHING on this process that is worth more than $300. That is what hurts them most.
The tricores may make up some of that, but not much.

As for cache being expensive... That's all about yields. Intels production technology simply allows them to get away with much more cache. It probably doesn't cost Intel more to put 8 mb on a quadcore than it costs AMD to put 4 mb on a quadcore at this point.
Click to expand...


I wasn't posting an example of actual $ and #s, It was a scnario to show that AMD really isn't doing that bad with the "Native" approach, nothing to do with speed of any chip, just the fact that a native approach doesn't give them any disadvantage really,

and yea AMD isn't making profit now, after years and $ of R&D they still arn't able to sell a quad core @ a good price :p like intel did by releasing it first ^^
 
Digital Viper-X- said:
I wasn't posting an example of actual $ and #s, It was a scnario to show that AMD really isn't doing that bad with the "Native" approach, nothing to do with speed of any chip, just the fact that a native approach doesn't give them any disadvantage really,
Click to expand...

It's not the price, the thing is that Intel has various quadcore parts, at a variety of speed ranges (and indeed price), where AMD can only create two parts, which are only 100 MHz apart. So apparently their yields aren't that good (the poor overclockability and extreme power dissipation are also writings on the wall), else they would offer a more varied product range. Basically it shows that 2.2-2.3 GHz is the worst, and also the best they can do.
Since they can clock their dualcore parts much higher on the same process, it is very likely that they would be doing better with MCM technology.
 
Scali2 said:
It's not the price, the thing is that Intel has various quadcore parts, at a variety of speed ranges (and indeed price), where AMD can only create two parts, which are only 100 MHz apart. So apparently their yields aren't that good (the poor overclockability and extreme power dissipation are also writings on the wall), else they would offer a more varied product range. Basically it shows that 2.2-2.3 GHz is the worst, and also the best they can do.
Since they can clock their dualcore parts much higher on the same process, it is very likely that they would be doing better with MCM technology.
Click to expand...

It is a combination of design issues coupled with process issues that are holding K10 back.Which means that dual core K10s will pretty much suffer from the same problems as the QC version.
 
savantu said:
It is a combination of design issues coupled with process issues that are holding K10 back.Which means that dual core K10s will pretty much suffer from the same problems as the QC version.
Click to expand...

Probably not. A big part of the design issues are probably related to the sheer size of the die. Getting the power and timing signals to every part of the chip properly is a tough issue, and AMD has no experience with dies of anywhere near that size, unlike Intel.
Dualcores are much smaller and hence less complex to design and manufacture.
Which is not to say that AMD won't still have process issues, because they had (have) those with the K8 on 65 nm aswell.
 
Scali2 said:
Probably not. A big part of the design issues are probably related to the sheer size of the die. Getting the power and timing signals to every part of the chip properly is a tough issue, and AMD has no experience with dies of anywhere near that size, unlike Intel.
Dualcores are much smaller and hence less complex to design and manufacture.
Click to expand...


When I say design issues I refer to proper validation.As late as early 2006 AMD had engineers doing high level uarch work on K10.You need about 30-36 months from then on to launch a viable product.
AMD shipped the K10 only 20 months later : a heroic and probably insane effort by AMD engineers.
The results ? Bugs , speed path issues , low yields , no OEM support ( they did not have the chips to validate the servers ) in other words a complete mess.

K10 as now is in a remarkably good condition for such a premature chip.

Furthermore , DC aren't smaller or less complex to design.They have all the features of the QC devices since the level of integration isn't as advanced ( no shared units or links between L1s , stuff like that ).Like the QC , the DC also has a crossbar , L3 cache , voltage controllers , etc.Since design issues ( like bugs ) are core centric ( not die centric ) the DC will suffer from the same problems.
 
savantu said:
When I say design issues I refer to proper validation.
Click to expand...

Well, I was talking about problems with manufacturing due to the chip design, not the validation of the logic design itself.
The logic hasn't changed all that drastically from the K8 design.
 
Seeing as how this thread has degraded to a fanboy lovefest... Fanboys linking to fanboys.... It's kind of a dirty rotten shame really.....
 
well, even though it might not have been the best option strategically, you have to give AMD credit for simply being able to release a monolithic quad core like that.... i mean the die is huge! that really is an achievement in itself... and it amazes me that such a small company with limited manufacturing is able to make quads that are monolithic

despite it's performance, the K10s really are a marvel of engineering
 
brucedeluxe169 said:
well, even though it might not have been the best option strategically, you have to give AMD credit for simply being able to release a monolithic quad core like that.... i mean the die is huge! that really is an achievement in itself... and it amazes me that such a small company with limited manufacturing is able to make quads that are monolithic

despite it's performance, the K10s really are a marvel of engineering
Click to expand...

I would, but since the same pretty much goes for Pentium 4 and D (original 8x0 series was also a single die), and they got nothing but flack, especially from the AMD camp, I'd feel out of place if I were to praise such characteristics when the performance isn't there.
 
Scali2 said:
I would, but since the same pretty much goes for Pentium 4 and D (original 8x0 series was also a single die), and they got nothing but flack, especially from the AMD camp, I'd feel out of place if I were to praise such characteristics when the performance isn't there.
Click to expand...

?? I thought that ALL pentium 4 Ds were MCM? which models were single die?
 
brucedeluxe169 said:
well, even though it might not have been the best option strategically, you have to give AMD credit for simply being able to release a monolithic quad core like that.... i mean the die is huge! that really is an achievement in itself... and it amazes me that such a small company with limited manufacturing is able to make quads that are monolithic
Click to expand...

K10 ( late 07 65nm ) is 286mm^2 and 463m transistors.
Let's compare that with :
Madison 6M ( mid 03 130nm ) is 374mm^2 and 410m transistors.
Madison 9M ( mid 04 130nm ) is 432mm^2 and 592m transistors.
Montecito ( mid 06 90nm) is 596 mm^2 and 1720m transistors.
Tulsa (mid 06 65nm ) is 435mm^2 and 1300m transistors.

Tukwilla ( late 08 65nm ) will be >650mm^2 and over 2B transistors.

These are achievements wrt to manufacturing prowess.It must be noted though that K10 has the highest logic to cache share among those cpus.

despite it's performance, the K10s really are a marvel of engineering
Click to expand...

A bit sorrow feat isn't it ?
 
Status
Not open for further replies.
Back
Top