Actually it won't be that at all apparently. Because of the design they can easily switch to just a 3 core die. But I would say it sounds interesting. I can only assume that one less core would definately improve the manufacturing process on many fronts.
Actually it won't be that at all apparently. Because of the design they can easily switch to just a 3 core die. But I would say it sounds interesting. I can only assume that one less core would definately improve the manufacturing process on many fronts.
Keysplayr
Elite Member
- Jan 16, 2003
- 21,219
- 54
- 91
What do you mean "apparently"? What did I miss? You're saying that AMD would turn off a perfectly good core in a quad core GPU to sell it as a tri-core for less money? Am I wrong in saying that Barcelona is massively expensive to produce, failed core or not? And AMD needs all the money it can get right now?
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
Originally posted by: keysplayr2003
What do you mean "apparently"? What did I miss? You're saying that AMD would turn off a perfectly good core in a quad core GPU to sell it as a tri-core for less money? Am I wrong in saying that Barcelona is massively expensive to produce, failed core or not? And AMD needs all the money it can get right now?
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
Can't answer the second one keys, but the first one is actually a marketing question...and the answer is yes, they would.
Look at some hypotheticals here...
Let's say you have a Barcy that bins out as a 2.5 GHz quad core, but if you shut off a core it bins out at 2.8 GHz...what would be the price difference, wouldn't they sell for about the same?
What if you have a 2.5 GHz Barcelona ES again...remembering that HE editions sell for a premium, couldn't you sell it as a 2.2GHz HE tri-core?
JPB
Diamond Member
- Jul 4, 2005
- 4,064
- 89
- 91
Originally posted by: keysplayr2003
What do you mean "apparently"? What did I miss? You're saying that AMD would turn off a perfectly good core in a quad core GPU to sell it as a tri-core for less money? Am I wrong in saying that Barcelona is massively expensive to produce, failed core or not? And AMD needs all the money it can get right now?
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
Its not all about failure or the need for money :roll:
Tri Core
Keysplayr
Elite Member
- Jan 16, 2003
- 21,219
- 54
- 91
Originally posted by: classy
Originally posted by: keysplayr2003
What do you mean "apparently"? What did I miss? You're saying that AMD would turn off a perfectly good core in a quad core GPU to sell it as a tri-core for less money? Am I wrong in saying that Barcelona is massively expensive to produce, failed core or not? And AMD needs all the money it can get right now?
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
Its not all about failure or the need for money :roll:
Tri Core
Thank you for the link, but it explained nothing. I saw a pic of three cores. Not four with 1 disabled. It is my opinion, and I'm entitled to it just as anyone, that tri-core barc's will be quad cores with poor yield (1 failed core). Unless the die's they produce are actually only three cores (which I severely doubt), these are failed quads. Not intentionally disabled. One sure way to tell, is time. If and when AMD releases tri-core, and as time goes on and manufacturing processes improve, the tri-core will become scarce as less cores fail to yield four fully functioning cores. It's going to be a short lived product. Again, this is my opinion and I could be wrong of course. But that's what it is. I don't think that ANYONE is calling for 3 core processors as they say. There is no market for it, but a need to facilitate one has become necessary to utilize a large number of failed cores. AMD PR needs to say there is a demand for such processors, and "we're here to fill that demand". I don't think so.
Key? I didn't pay attention to who wrote the post, I responded to.
The only thing you said that made any sense was "One sure way to tell, is time."
Do we have any need for 4 cores or 3? Ah no.
As far as failures go, I did not see any info released as to the failures or low yeilds being failed cores.
And the design allows them to do such things. Intel doesn't have a cpu capable of being manipulated as such.
Its not always something failed.
Originally posted by: keysplayr2003
Originally posted by: classy
Originally posted by: keysplayr2003
What do you mean "apparently"? What did I miss? You're saying that AMD would turn off a perfectly good core in a quad core GPU to sell it as a tri-core for less money? Am I wrong in saying that Barcelona is massively expensive to produce, failed core or not? And AMD needs all the money it can get right now?
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
Its not all about failure or the need for money :roll:
Tri Core
Thank you for the link, but it explained nothing. I saw a pic of three cores. Not four with 1 disabled. It is my opinion, and I'm entitled to it just as anyone, that tri-core barc's will be quad cores with poor yield (1 failed core).
Nope...give that opinion BACK! (sorry...j/k, don't want to piss off the mod!)
But you didn't answer my question (which I think goes to your point...)
Given a Barcy that bins out as a 2.5 GHz quad core, but if you shut off a core it bins out at 2.8 GHz...what would be the price difference?
If the selling price is about the same, then wouldn't it make sense to convert a perfectly good quad into a tri?
I think we can all agree there!
Keysplayr
Elite Member
- Jan 16, 2003
- 21,219
- 54
- 91
Ahh, sorry. Well, I think if you have a Barcy that bins out at 2.5 for X amount of dollars, but then shut off a core, then it would have to sell at less than X amount of dollars. Then clock the "now" tri-core to 2.8. I think the price would have to be the same as the 2.5 quad core. So it's like it's a pointless trade off. More MHz 3 core, or less MHz 4 core. Break even.
My point is, well, what's the point? There is no purpose to this line of thinking "unless" there was no alternative. Points back again to failed cores. I would guess failed cores happen to all the manufacturers, so it's not uncommon. But if there is an overabundance of failed cores, you have to do something with what you have. Too much money wasted. Too costly to throw away. I'll bet you tri-core, if it does indeed happen, will be VERY short lived. IMHO.
AMD will improve it's process and tri-core will be but a hiccup in the time-space continuum.
Keysplayr
Elite Member
- Jan 16, 2003
- 21,219
- 54
- 91
Originally posted by: classy
Key? I didn't pay attention to who wrote the post, I responded to.
The only thing you said that made any sense was "One sure way to tell, is time."
Do we have any need for 4 cores or 3? Ah no.
As far as failures go, I did not see any info released as to the failures or low yeilds being failed cores.
And the design allows them to do such things. Intel doesn't have a cpu capable of being manipulated as such.
Its not always something failed.
NP. Lots of folks have a great need for as many cores they can get. Scientists the world over require phenomenal processing power for things such as BioInformatics, or even just look at all the grid folding/processing networks worldwide.
For the average user, you are certainly correct in that 3 to four cores is overkill. 1 core will do just fine as it has for over 25 years.
As to failures, you will not ever see accurate info released on yield failures from any semi manufacturer. They have to report some I would imagine, because zero failures just doesn't happen, but bad yields cost money (yes, it's all about money, always was, always will be) and stockholders pay attention to such things i would imagine.
Design: We don't know if an Intel C2Q could function only using three cores. Just as you haven't seen any info released on failed cores from AMD, neither have you seen any information on if a Kentsfield could operate only using 3 cores. That's why I was asking the questions I was asking, hopefully spurring someone who actually knows, would answer.
And why do you have a problem with me using the term "failed"? It happens all the time to every semi-con manufacturer. Some way more than others? Nobody is perfect, and will ever be perfect.
Originally posted by: keysplayr2003
Ahh, sorry. Well, I think if you have a Barcy that bins out at 2.5 for X amount of dollars, but then shut off a core, then it would have to sell at less than X amount of dollars. Then clock the "now" tri-core to 2.8. I think the price would have to be the same as the 2.5 quad core. So it's like it's a pointless trade off. More MHz 3 core, or less MHz 4 core. Break even.
My point is, well, what's the point? There is no purpose to this line of thinking "unless" there was no alternative. Points back again to failed cores. I would guess failed cores happen to all the manufacturers, so it's not uncommon. But if there is an overabundance of failed cores, you have to do something with what you have. Too much money wasted. Too costly to throw away. I'll bet you tri-core, if it does indeed happen, will be VERY short lived. IMHO.
AMD will improve it's process and tri-core will be but a hiccup in the time-space continuum.
It's a marketing point...many clients would be better served with a higher clocked tri-core than a slower quad core. Gamers for one...
It's just a question of choices and what's best for you.
As for AMD's cost, it means that they can deliver a broader range of choices for no real cost difference to them...and as you say, it also allows for a market with defective quad cores as well. I'd call it a win-win myself.
TuxDave
Lifer
- Oct 8, 2002
- 10,571
- 3
- 71
Originally posted by: coldpower27
Originally posted by: AmberClad
Yes, you're right that it shouldn't really matter. The lack of elegance just doesn't sit well with me. Just like the idea of a non-binary-based number of cores doesn't sit well with me
There is no law that says it must be 2^n where N is a Integer value.
I declare it law. Discuss...
Now official.
Say hello to Toliman and Heka.
http://www.extremetech.com/art.../0,1697,2184275,00.asp
Say hello to Toliman and Heka.
http://www.extremetech.com/art.../0,1697,2184275,00.asp
Originally posted by: keysplayr2003
Another question. Related. If a Core2Duo Core was created with one of the cores bad, does intel sell it as a Celeron or Core 2 Solo? And in a Core2Quad, if one of the cores suddenly craps out, does the whole processor cease to function, or do the other three cores carry on? OR does 1/2 of the C2Q CPU shut down leaving you with essentially a C2D. Only answer if you know for sure. Where is Dmens when you need him.
The Xeon 7200 series is Clovertown with one core on each side disabled, leaving each core with 4 MB. Triple-core really should not be a problem if Intel needs to produce an SKU.
In case the story i was reading, and the picture i was looking at, were fake. And they were.
Someone had rebranded a die shot of IBM's Xbox 360 CPU as the new triple core from AMD, which made it look like a specifically made native tri core chip, instead of a quad core with a deactivated core.
Phynaz
Lifer
- Mar 13, 2006
- 10,140
- 819
- 126
Originally posted by: Cygni
In case the story i was reading, and the picture i was looking at, were fake. And they were.
Someone had rebranded a die shot of IBM's Xbox 360 CPU as the new triple core from AMD, which made it look like a specifically made native tri core chip, instead of a quad core with a deactivated core.
Hehe...I saw your original post...And yeah, I was going to jump on it
A tri core processor actually makes a great deal of sense. AMD could either throw them away ...... or sell them as tri-cores when one core doesn't operate.
There is little doubt that a tri-core K10 processor, even at 2.5Ghz, would beat a 2.93Ghz dual core Core 2 in highly threaded applications. Right now, that 2.93Ghz dual core is going for around $1000.00.
At a bare minimum, having an X3 in AMD's lineup is going to cause a real stir in marketing. Surely a consumer would consider an X2 as being a competitor to a Core 2 duo so an X3 must be better than Core 2 duo, but maybe not as good as a Core 2 quad.
As already pointed out, AMD could also use tri-core to yield faster clocks. 3 of the 4 cores may yield to 2.8Ghz while one core only yields out at 2.4Ghz..... AMD may well be able to get more for a 2.8Ghz X3 than they can for a 2.4Ghz X4.
And let me take a minute to discuss clock speeds for AMD.
AMD scaled K8 from 2.6Ghz on .13um to 3.2Ghz on 90nm ..... and they doubled the number of cores at the same time.
Should AMD repeat this trick, they would be capable of producing a 4.0Ghz quad core on 65nm.
Now I am not saying that they can do this, but keep in mind that IBM is churning out 5.0Ghz Power6 processors on 65nm SOI. Sure it is a totally different architecture, but it shows that 65nm SOI can scale.
There is little doubt that a tri-core K10 processor, even at 2.5Ghz, would beat a 2.93Ghz dual core Core 2 in highly threaded applications. Right now, that 2.93Ghz dual core is going for around $1000.00.
At a bare minimum, having an X3 in AMD's lineup is going to cause a real stir in marketing. Surely a consumer would consider an X2 as being a competitor to a Core 2 duo so an X3 must be better than Core 2 duo, but maybe not as good as a Core 2 quad.
As already pointed out, AMD could also use tri-core to yield faster clocks. 3 of the 4 cores may yield to 2.8Ghz while one core only yields out at 2.4Ghz..... AMD may well be able to get more for a 2.8Ghz X3 than they can for a 2.4Ghz X4.
And let me take a minute to discuss clock speeds for AMD.
AMD scaled K8 from 2.6Ghz on .13um to 3.2Ghz on 90nm ..... and they doubled the number of cores at the same time.
Should AMD repeat this trick, they would be capable of producing a 4.0Ghz quad core on 65nm.
Now I am not saying that they can do this, but keep in mind that IBM is churning out 5.0Ghz Power6 processors on 65nm SOI. Sure it is a totally different architecture, but it shows that 65nm SOI can scale.
The problem is the X6800 is a obsolete SKU and there is a better 3GHZ Dual Core for only $266 from Intel. For all Intents and purposes the best Dual Core is no pricier then that, obsolete SKU's don't count.
Intel has Quad Core at $266 it will be interesting if AMD makes any Tri-Core above that price.
It took AMD 2.5 Yrs give or take to scale K8 that far, so by the time they reach 4.0GHZ assuming that is even possible for AMD's particular architecture it may not even matter as it will be going against Intel's 32nm processors.
IBM has Power6 processors at 4.7GHZ in Dual Core form last time I checked and as we know architecture is the stronger factor in determining how high a clockspeed it can attain rather then the process.
Honestly, I don't see how AMD expects to compete using triple-core.
Q6600 is currently $266. With Penryn/Yorkfield set to arrive soon, Q6600 is bound to drop in price, probably to the $224 slot (or less).
If $224 isn't cheap enough, Intel can probably package together a pair of 2.13GHz Conroe-2M dice for $183. Still not cheap enough? How about using a pair of Allendale cores (say, 2GHz E2180) for $133?
Q6600 is currently $266. With Penryn/Yorkfield set to arrive soon, Q6600 is bound to drop in price, probably to the $224 slot (or less).
If $224 isn't cheap enough, Intel can probably package together a pair of 2.13GHz Conroe-2M dice for $183. Still not cheap enough? How about using a pair of Allendale cores (say, 2GHz E2180) for $133?
what ibm can do with their arch on 65nm has zero relevance to what amd can do with it on their arch, in either current or future frequency bins.
Here is an idea. Maybe they are trying to maximize their yields as best they can.
3 core die designed from ground up, as some sites are claiming. Then you have a real good set of fail safes built into the process,
Good x4 dies -> phenom x4
1 bad x4 core -> x3
Good x3 dies -> x3
1 bad x3 core -> x2
Good x2 dies -> x2
1 bad x2 core -> low end, low power mobile market for x86?s that growing rapidly
Would there be a capacity issue with running 3 separate dies (x2, x3, x4) as opposed to 2?
3 core die designed from ground up, as some sites are claiming. Then you have a real good set of fail safes built into the process,
Good x4 dies -> phenom x4
1 bad x4 core -> x3
Good x3 dies -> x3
1 bad x3 core -> x2
Good x2 dies -> x2
1 bad x2 core -> low end, low power mobile market for x86?s that growing rapidly
Would there be a capacity issue with running 3 separate dies (x2, x3, x4) as opposed to 2?
TRENDING THREADS
-
Discussion Zen 5 Speculation (EPYC Turin and Strix Point/Granite Ridge - Ryzen 9000)- Started by DisEnchantment
- Replies: 25K
-
Discussion Intel Meteor, Arrow, Lunar & Panther Lakes + WCL Discussion Threads
- Started by Tigerick
- Replies: 22K
-
Discussion Intel current and future Lakes & Rapids thread- Started by TheF34RChannel
- Replies: 23K
-
News NVIDIA and Intel to Develop AI Infrastructure and Personal Computing Products- Started by poke01
- Replies: 403
-
Facebook
Twitter