3/4 of quad core.

[monkeymanuk]

Does anybody know whether AMD's K8L implementation will allow for 3 of 4 cores to be used if for example a one of the four cores fails during production? Thus providing a cheaper chip but still a great performer?

 

[joset]

Does anybody know whether AMD's K8L implementation will allow for 3 of 4 cores to be used if for example a one of the four cores fails during production? Thus providing a cheaper chip but still a great performer?

Other than intentionally no, I don't; and, that's still a mere possibility since K8L is not out yet... but, you know that, don't you?


Cheers!

 

[endyen]

My opinion (strictly conjecture at this point) is that chips with one broken core, will have a second core deactivated as well, and will be sold as dual core. This way they can hide any yield problems. Latter they may intro a tri-core chip.

 

[joset]

My opinion (strictly conjecture at this point) is that chips with one broken core, will have a second core deactivated as well, and will be sold as dual core. This way they can hide any yield problems. Latter they may intro a tri-core chip.

I'm not sure why I agree with you ; could you elaborate on that, in theory, that is?


Cheers!

 

[TabrisDarkPeace]

:?: So far K8L hasn't had any quad-core on a single die chips, has it ?

:? So my 2nd question is "How could 1/4 of 4 cores fail, if they procude no quad-core dies ?"

:!: So just when are (65 nm) quad-core, Socket-F (LGA 1207 pins) CPUs coming out by AMD ?

If several quad-core CPUs had 'full cores' failing, and lowering clock speed didn't help out then I'd just lower clock speed + power consumption, and package it as an embedded processor for specific markets (to be sold in pairs, for 6 core low power embedded servers).

It avoids the 'embarrassment' in western markets, and may still be classed as 'a winner' in its own right.

They should've just paired 2 x dual-core dies using 90 nm if they have not got 65 nm ready for mass production of 'large'* die sizes.

But they can't do that after their 'Quad-Core for Dummies' stunt. AMD really shot themselves in the foot. As they have the interconnect speed to pull off 2 x dual-cores per socket even better than Intel has (with a several month lead). It won't help performance 'that much' having all 4 cores on one die, it'll only help share L2 (per core) and L3 cache (per socket).

* - As in good yields on 8 x 8 mm or larger

With a weaker US$ on the way, China will be their best bet of a market IMHO (April 2007 - Nov 2007).

 

[monkeymanuk]

1) I was talking about the future!!!!

2) see q1

3) I believe the release date is sometime in the future :lol:


I posed the question because I think 3 cores is the perfect number right now and for the next few years (unless there is a major leap in programming practices).

Dual core is great and really improves response times in Windows but I think that when you are running background processes like dvd ripping, then a third core would ensure that the responsiveness remains.

A try-core chip would come in quite a bit cheaper (I would hope) than a full quad.