Monday, February 15th 2010

Phenom II X6 to Get C-State Performance Boost Feature

AMD's upcoming six-core desktop processor, the Phenom II X6, will introduce a new feature to the Phenom II series, currently known as "C-state performance boost". The feature gives the processor control over individual cores' power states, and the ability to gate power down completely a core. The remaining active core(s), will then be overclocked beyond the normal clock speeds, so that low-power operation with reduced parallel computing load goes on with much lesser energy consumption.

Features such as C-state performance boost and 'power-gating' is relatively new for AMD processors. The two were originally expected to be introduced with AMD's 32 nm "Llano" Fusion APUs. AMD's Phenom II X6 is expected to be released in May 2010, its architecture is derived from the six-core "Istanbul" Opteron processors, albeit in the socket AM3 package that supports dual-channel DDR3 memory. There are as many as four models in the works for an initial release. Known details of these can be found here.

Source: X-bit Labs
Add your own comment

36 Comments on Phenom II X6 to Get C-State Performance Boost Feature

#1
Zubasa
BababooeyHTJ said:
In apps that can take advantage of more than four threads. I just don't get the point of this chip for the vast majority of desktop pcs. So with most apps it's no faster than my (what three year old?) Q9650. Without a doubt useless for gaming.
It is not like the main use of a PC is gamming anyways. :cool:
A Hexa-Core is good for any thing that needs that CPU power, like video transcoding and etc.

BTW, the Yorkfield was released on March 08, so there is no way that the Q9650 is 3 years old.
Posted on Reply
#2
BababooeyHTJ
Zubasa said:
It is not like the main use of a PC is gamming anyways. :cool:
A Hexa-Core is good for any thing that needs multi-threaded that CPU power, like video transcoding and etc.
So, it's good for a workstation. Isn't that what Operton is for? A 965 will out perform this in a vast majority of desktop apps.
BTW, the Yorkfield was released on March 08, so there is no way that the Q9650 is 3 years old.
Close enough, my point still stands. If it makes you feel better replace that with kentsfield.
Posted on Reply
#3
Zubasa
BababooeyHTJ said:
So, it's good for a workstation. Isn't that what Operton is for? A 965 will out perform this in a vast majority of desktop apps.
So for some reason as a software dev, you will code a program that is not multi-thread so that you can get people screwed?
We are living in year 2010 here, and this proc is due for May.

Also, the whole point to this "Turbo Boost" type technique is so that the 965 does not "out perform this in a vast majority of desktop apps".

BababooeyHTJ said:

Close enough, my point still stands. If it makes you feel better replace that with kentsfield.
Both the Deneb and the Yorkfield are superior chips to the Kentsfield, so no.
Posted on Reply
#4
BababooeyHTJ
Zubasa said:
So for some reason as a software dev, you will code a program that is not multi-thread so that you can get people screwed?
We are living in year 2010 here, and this proc is due for May.

Also, the whole point to this "Turbo Boost" type technique is so that the 965 does not "out perform this in a vast majority of desktop apps".
I'm not even sure what you are trying to say. Multi-threading is not always that easy. Devs aren't going out of their way to not make things multi-threaded.

This thing will be slower in most desktop apps than an AMD chip that is already on the market. I just don't understand what segment they are going for with this. Non ECC workstation?

Have you seen the specs as to what AMD's "turbo boost" is going to put out? How can you comment?
Both the Deneb and the Yorkfield are superior chips to the Kentsfield, so no.
Not by very much and in the case of Deneb that isn't always the case. I'm talking clock for clock here.


If this retails for under $300 at a decent clock speed and is somewhat overclockable it would be a pretty decent chip.
Posted on Reply
#5
Zubasa
BababooeyHTJ said:

This thing will be slower in most desktop apps than an AMD chip that is already on the market. I just don't understand what segment they are going for with this. Non ECC workstation?

Have you seen the specs as to what AMD's "turbo boost" is going to put out? How can you comment?
You are also commenting on AMD's new chip without knowing exactly how it will perform.
We are both speculating here.
Posted on Reply
#6
BababooeyHTJ
Zubasa said:
You are also commenting on AMD's new chip without knowing exactly how it will perform.
We are both speculating here.
True, I can't wait to see some benchmarks and if it retails at a reasonable price it could be a solid performer for some people.
Posted on Reply
#7
devguy
Wile E said:
Considering that Intel has the same number of physical cores as AMD, but also has HyperThreading, that statement never made any sense whatsoever. Intel has 6 core cpus as well, but they can do twice as many threads per core. I fail to see how that's a bad thing.
I never said I agree with AMD on that statement, lol. However, I don't think the context of it applied to Thuban. I think it was meant to be applied to the fight between Magny-Cours and the octal-core (16 thread) Nehalem-EX. (A fight that may be closer than one thinks, btw). Although, my issue with this is that they are putting down Intel for doing something "real men don't do". Yet, for a considerable amount of time up to and after the release of the Agena/Barcelona, AMD touted the "native" quad core approach (what "real men do") vs Intel's "duct-tape" approach on both Pentium D and on Kentsfields (Magny-Cours will adopt the "duct-tape approach).
Posted on Reply
#8
Wile E
Power User
devguy said:
I never said I agree with AMD on that statement, lol. However, I don't think the context of it applied to Thuban. I think it was meant to be applied to the fight between Magny-Cours and the octal-core (16 thread) Nehalem-EX. (A fight that may be closer than one thinks, btw). Although, my issue with this is that they are putting down Intel for doing something "real men don't do". Yet, for a considerable amount of time up to and after the release of the Agena/Barcelona, AMD touted the "native" quad core approach (what "real men do") vs Intel's "duct-tape" approach on both Pentium D and on Kentsfields (Magny-Cours will adopt the "duct-tape approach).
Perhaps it was in reference to Magny-Cours, but if that's the case, the statement doesn't really apply when speaking of these desktop cpus anyway.

And I wonder what the power consumption and clock speeds are gonna be like on Magny-Cours. It can't be all that pretty, that's for sure.
Posted on Reply
#9
saikamaldoss
EarlZ said:
About time AMD will release something that can outperform a i920
The first reason for that is 3channel ram that intel is using and the NB is more powerful than of AMD. if AMD had 3channel DDR3 ram support and a 7MT/s bus, they will kick every intel chip. :) but the price will be up on the sky...

but i think AMD learned something from ATI this time.

performance per watt is not good as performance for price :)

I hope to see AMD answering every intel processor like they are doing with Nvidia. :cool:
Posted on Reply
#10
Roph
I'll wait for the/an Athlon II X6, I think I'd go with that + a 5770 for a new efficient build =o
Posted on Reply
#11
newtekie1
Semi-Retired Folder
xrealm20 said:
Just a side note, but wouldn't the guy with the 502 and the blower have more displacement than a 454? :D
Not to derail the thread too much, but yes that is the point.

AMD's comment didn't make any sense because at the time they made it both Intel and AMD had 4 physical cores, but Intel's were more powerful.

Just like someone with a 454 commenting about the guy with a supercharger by making the "There's no replacement for displacement" comment, when even without the supercharger the guy with the 502 has more displacement.
Posted on Reply
Add your own comment