Sign in with
Sign up | Sign in
Your question

difference in die shrink

Last response: in CPUs
Share
June 28, 2013 11:44:46 PM

what are the core difference in terms of performance being changing the micro architecture of the processor?
sandy bridge was 32 nm & ivy bridge was minimize to 22 nm & haswell is the same as ivy bridge

More about : difference die shrink

a b à CPUs
June 29, 2013 5:44:08 AM

Mfzb0910,

I think that the reduction in lithography size is not the reason for performance improvements as much as is the ability to have a higher transistor density and therefore the calculation density increases. The smaller lithography allows more cores and the inclusion of and improvements to features like onboard graphics. The penalty of the higher density of Ivy Bridge 22nm is that they ran hotter, so for Haswell, the power required was reduced which also benefits mobile (battery) use. Also, in Haswell, the HD4000 was improved to the HD4600.

They're not common as they are so costly- $3,500 and more- but there are 10-core E7 Xeons ( some of these can be used in 8-CPU configurations) and Intel is said to be working on twelve and fifteen core CPU's that won't be larger than a current LGA 2011 as they will be 14nm. I've read that 14nm may be at or near the limit with current materials as beyond that there would be proximity effects.

Cheers,

BambiBoom
m
0
l
July 1, 2013 1:56:08 AM

Yeah I too felt the same as much as the lithography size gets reduced it does not make any huge difference in terms of performance as long as you have more transistors . as far as my experience is concerned 2011 socket processors are much more better than the current and previous 1155 &1150 series ones ..
i appreciate your attention and prompt reply in this regard.
cheers,
fayaz
m
0
l
Related resources
a c 478 à CPUs
July 1, 2013 11:19:49 AM

There is no performance increase that is directly related to a die shrink. Unless you include lower power consumption as a performance characteristic.

Overall, the amount of heat produced is decreased as well. However, as a CPU becomes smaller, the surface area also shrinks which means less contact area with the heatsink. While the CPU may produce less heat, the small surface area can also conduct less heat as well. So even after a die shrink, the overall temperature as measured by CPU's temperature sensor could be the same or higher than before the die shrink.

Higher clockspeed can be achieved with a die shrink since on the one hand higher clockspeed means more voltage, a die shrink means less voltage. That is assuming heat is effectively being conducted from the CPU's surface to the heatsink.
m
0
l
a b à CPUs
July 1, 2013 1:17:28 PM

bambiboom said:
Mfzb0910,

I think that the reduction in lithography size is not the reason for performance improvements as much as is the ability to have a higher transistor density and therefore the calculation density increases. The smaller lithography allows more cores and the inclusion of and improvements to features like onboard graphics. The penalty of the higher density of Ivy Bridge 22nm is that they ran hotter, so for Haswell, the power required was reduced which also benefits mobile (battery) use. Also, in Haswell, the HD4000 was improved to the HD4600.

They're not common as they are so costly- $3,500 and more- but there are 10-core E7 Xeons ( some of these can be used in 8-CPU configurations) and Intel is said to be working on twelve and fifteen core CPU's that won't be larger than a current LGA 2011 as they will be 14nm. I've read that 14nm may be at or near the limit with current materials as beyond that there would be proximity effects.

Cheers,

BambiBoom


jaguarskx said:
There is no performance increase that is directly related to a die shrink. Unless you include lower power consumption as a performance characteristic.

Overall, the amount of heat produced is decreased as well. However, as a CPU becomes smaller, the surface area also shrinks which means less contact area with the heatsink. While the CPU may produce less heat, the small surface area can also conduct less heat as well. So even after a die shrink, the overall temperature as measured by CPU's temperature sensor could be the same or higher than before the die shrink.

Higher clockspeed can be achieved with a die shrink since on the one hand higher clockspeed means more voltage, a die shrink means less voltage. That is assuming heat is effectively being conducted from the CPU's surface to the heatsink.


Nice writeup. Skylake and 14nm is going to get very interested when it comes to cooling
m
0
l
!