Ivy Bridge 20 percent faster than Sandy

ghnader hsmithot

Distinguished
some rumour:
not sure if it is true nor do i claim it to be true but here is the rumor
http://www.fudzilla.com/processors/item/22322-ivy-bridge-20-percent-faster-than-sandy

Intel claims
Intel is already talking to its partners about its upcoming 22nm Ivy Bridge CPU, and it is sharing some hints of performance. One of the things that Intel told recently is that this 22nm processor should offer more performance with a similar thermal envelope.

Intel tells its partners that they can expect around 20 percent performance increase from Sandy Bridge, which is roughly the same gain that Sandy Bridge had over Nehalem based CPUs. This is an optical shrink of the existing 32nm Sandy Bridge architectureas Intel traditionally takes a safe approach when moves from one manufacturing process to another. The 22nm part after Ivy Bridge is the one with the new architecture, while Ivy Bridge's first competition are Bulldozer and Llano from AMD.

Intel's Ivy Bridge will work in H67 and P67 boards and it will fit in existing boards for both desktop and mobile parts. We already reported about it here
 

ghnader hsmithot

Distinguished
for me..i doing a little study on cpu architecture this could possibly be true.One is that die size, the smaller the die size the faster it can execute instructions due to the fact travelling of signals.Smaller die size does equate to faster cpu.
My opinion.
 

GunBladeType-T

Distinguished
Jul 8, 2010
553
0
19,010

So basically Posion IVY is 20%Faster than A Sandy Witch! (Batman) Uma Thurman vs. (The Net) Sandra Bullock or Sandy From (GoodFellas)

Another Q is that Peggy Bundy or Niki Cox?
 

Haserath

Distinguished
Apr 13, 2010
1,377
0
19,360
One of the things that Intel told recently is that this 22nm processor should offer more performance with a similar thermal envelope.
Higher clocks due to the smaller transistors causing a lower thermal envelope for the same clockspeed. It might be due to some of the signals reaching certain places faster, but I do believe the 20% "speed" increase will, mostly, be due to an increase in stock/over clocks.

Just my opinion on that. :sol:
 


Possibly or just tweaks too. Core 2 Quad 45nm was 15% faster at the same clock speed due to tweaks, faster cache and larger cache.
 

ElMoIsEviL

Distinguished

The way I read this is that they're saying 20% more performance for the same power enveloppe. What this means, I assume, is that we're not looking at a clock for clock comparaison but rather a performance/watt comparaison.

Also, die shrinks do not equate to improvements in scheduling and speed (at least nothing that could be shown as being statistically relevant). What they do tend to allow for (die shrinks) is a lower power envelope. This is achieved by the fact that a smaller circuitry has a smaller surface area. Smaller area means less distance for a signal to travel. Less distance means less resistance and less resistance means less Amperage and Voltage (a*v=Watts of Power) required to run. Less Power is what we're looking at here in my opinion.

Less power or same power envelope while having a higher operating frequency results in more performance/watt.
 

thats true to a point, but smaller dyes means more chance of leakage, so they have to use higher K materials which in turn increase resistance.
remember the amd switch from 90nm to 65nm. it ended up the brisbane core was 2 percent slower even after the dye shrink, due to them having to use hi K materials to stop leakage, but also ended up with increased resistance, making the chip slower over all.

im not saying it will be the case for intel but it is possible.
 


Last time I checked, AMD hasn't used hi-k in their chips and wont until 32nm and Bulldozer. Theri 65nm was just bad all around and was just SOI.

Intel put HK/MG on 45nm first.