[AMD] K12 will be on 28nm

[Phynaz]

Hey, don't shoot the messenger. I'm just reading off the AMD spec sheet. Well, Wikipedia actually.

http://en.wikipedia.org/wiki/List_of_AMD_FX_microprocessors

 

[AtenRa]

Performance wise AMD got beaten so badly.
http://www.anandtech.com/show/5553/the-xeon-e52600-dual-sandybridge-for-servers

Bulldozer failed everywhere for everything.

Get your facts straight,

Intel 32nm Gulftown XEON release date March 2010

AMD 32nm Interlagos (Opteron 62xx) release date November 2011

Intel 32nm Sandybridge XEON release date March 2012

AMD 32nm Delhi (Opteron 63xx) release date December 2012

Intel 22nm Ivybridge XEON release date September 2013

 

[Abwx]

Full load clocks are 100Mhz higher. Who cares about clocks for an idle CPU? AMD tried tried that one when they intentionally lowered the base clocks to be able to say they had more turbo. Nobody fell for that marketing tactic.

Full load clocks with 8 cores are 3.6 and 4.0 respectively, if there was only 100MHZ the Mthreaded benches wouldnt look like this, although you are free to keep believing your own lies :

 

[AtenRa]

Hey, don't shoot the messenger. I'm just reading off the AMD spec sheet. Well, Wikipedia actually.

http://en.wikipedia.org/wiki/List_of_AMD_FX_microprocessors

FX8150 and FX8350 have the same Vcore, 1.425V

 

[ShintaiDK]

Get your facts straight,

Intel 32nm Gulftown XEON release date March 2010

AMD 32nm Interlagos (Opteron 62xx) release date November 2011

Intel 32nm Sandybridge XEON release date March 2012

AMD 32nm Delhi (Opteron 63xx) release date December 2012

Intel 22nm Ivybridge XEON release date September 2013

And? What part is wrong. SB utterly destroyed Faildozer. And AMD admitted it was an epic failure.

 

[Intervenator]

 

[Intervenator]

Get your facts straight,

Intel 32nm Gulftown XEON release date March 2010

AMD 32nm Interlagos (Opteron 62xx) release date November 2011

Intel 32nm Sandybridge XEON release date March 2012

AMD 32nm Delhi (Opteron 63xx) release date December 2012

Intel 22nm Ivybridge XEON release date September 2013

3.5 years after Intel mass produces 32nm, 2.5 after 22nm... No tri-gate... And that's assuming K12 comes out Q1 2016? Ugh.

 

[Kallogan]

they better have a killer new arch if they stay on 28 nm

 

[Phynaz]

FX8150 and FX8350 have the same Vcore, 1.425V

Will you be updating Wikipedia?

 

[Phynaz]

Full load clocks with 8 cores are 3.6 and 4.0 respectively, if there was only 100MHZ the Mthreaded benches wouldnt look like this, although you are free to keep believing your own lies :

Nevermind, not getting drawn into this anymore.

 

[Abwx]

Okay, then CPU design can increase efficiency by 10% or more?

I talked of a 10% voltage disadvantage, this doesnt translate by 10% more losses but 21%, i said that at equal competences it s a gap that is impossible to close, you ll have to rely on the competition neglecting their design, wich is not a good management model, the exemple of Kabini vs Baytrail is not totaly relevant, not only the CPUs do not work at comparable frequencies for a given performance level but they are both at a point where increasing IPC could still provide a positive return, not so much for high throughput designs like the FX8xxx or the i7, the former has some margin left by virtue of the smaller cores while the latter is already beyond the knee of the curve.

 

[Fjodor2001]

I think what's more interesting is what node Zen (next AMD x86 uArch) will be on. My guess is it'll be targeting GF/Samsung 14 nm, since they are designing from scratch.

 

[NostaSeronx]

If there is any foundation to the 28-nm statement. It sort of follows both TOP-PIM and EUV technologies.

We consider two potential PIM design points geared towards high-performance applications, one in 22nm technology and another in 16nm technology.

GloFo EUV nodes are 22-nm/16-nm/10-nm. This is not to be confused with half pitches.

28/22-nm @ GlobalFoundries+IBM = 90-nm M1 Pitch to 80-nm M1 Pitch (45/40-nm HP)