Jump to content

28nm vs 22nm

Gauzl

What's the difference and how would it effect performance?

The difference is 6nm (28nm - 22nm = 6nm it's simple math)

My n00Build http://pcpartpicker.com/user/NameOfNoInterest/saved/ 


Do you really expect me to know things? 


$25.67/$~2500 (Price changes hourly..) I'm 1/100th of the way there!

Link to comment
Share on other sites

Link to post
Share on other sites

more transistors = more better

basically the smaller the die the better, so 22 nm is going to be better if the architecture isn't incompetent 

Link to comment
Share on other sites

Link to post
Share on other sites

28nm vs 22nm is the size or length of a transistor in a silicon chip. The smaller they are the more you can have in a certain area. thus more transistors = more work (or more performance) being done.

 

There's some problems with them getting smaller Physics say we can't get smaller than about 5or 6nm because at that point we won't be able to see where the electrons are.  Que Heisenberg's uncertenty principle.

 

Also- more transistor in one are means more heat. However since they're smaller they can use less voltage (look at haswell using very little voltage compared to 32nm parts) thus producing less heat.

 

Here's a related thread on teksyndicate talking about Maxwell and the jump to 20nms on the GPU side of things: https://teksyndicate.com/users/rsilverblood/blog/2014/02/06/gpu-wars-enter-maxwell-nvidias-successor-kepler

TekSyndicate Forum Moderator: https://forum.teksyndicate.com/users/njm1112

5930K@4.3Ghz | 16GB 4x4 2133Mhz | r9-290 | 6TB Raid5 | 250GB 850Evo | 8.1pro | RM750

Link to comment
Share on other sites

Link to post
Share on other sites

The difference is 6nm (28nm - 22nm = 6nm it's simple math)

Damn... I thought the answer was 2... :P

My Rig :  Case: Cooler Master HAF X ,Motherboard: Gigabyte Z87X-UD3H,PSU: Seasonic SS-750KM3,Processor: Core I7 4770k (overclocked 4.7ghz),Cooler: Corsair H100i, GPU: EVGA GTX 780 with acx cooler, RAM: G.Skill Ripjaws 16gb DDR3 1600 (overclocked to 2000mhz), HDDS  Samsung 840 EVO 250 gb SSD , Western digital  2tb 7200 rpm 64mb cache, Old 1tb laptop drive I had , 320gb for os backup daily, 80gb external for weekly backups,Drives 2x Lg Blu Ray burner WH16MS40,MISC: Tp-Link dual band wireless card, Logitech g510s, Razer Deathadder 2013, Acer G236HLBbd 23" monitor, Old tv I had 23" for secondary monitor, old 32" samsung tv third monitor

Link to comment
Share on other sites

Link to post
Share on other sites

These values are referred to as the litography or the litographic size. They denote the width of a transistor inside an Integrated Circuit (IC or chip). Inherently the litography doesn't affect performance at all, it merely changes the size of the IC die (the piece of silicon onto which the IC is etched).

 

Properties of the IC that are (proportionally) affected by the litography:

  • operating voltage
  • maximum current throughput
  • breakdown voltage
  • heat output
  • physical die size

 

By using a smaller manufacturing process, IC manufacturers are able to put more chips on one wafer. This increases the yield, as little spot errors in the silcon will affect a smaller percentile of chips. This, in turn, decreases the cost of producing one chip. This is mandatory to stay competitive. In order to make their CPU's more powerful, CPU manufacturers need an increasing number of transistors (the basic building stones of IC's). Without shrinking the process, CPU dies would be incredibly big by now, meaning that the yield during manufactering would be dreadfully low.

 

A process shrink is a double-edged sword in terms of IC temperature. Inherently, heat output goes down by shrinking the die, this is because the operating voltage for he transisors goes down. On the other hand, making the die smaller also decreases the surface through which it conducts this heat to your heatsink, making the heat transfer less efficient, thus increasing the temperature of the chip.

Link to comment
Share on other sites

Link to post
Share on other sites

These values are referred to as the litography or the litographic size. They denote the width of a transistor inside an Integrated Circuit (IC or chip). Inherently the litography doesn't affect performance at all, it merely changes the size of the IC die (the piece of silicon onto which the IC is etched).

Properties of the IC that are (proportionally) affected by the litography:

  • operating voltage
  • maximum current throughput
  • breakdown voltage
  • heat output
  • physical die size

By using a smaller manufacturing process, IC manufacturers are able to put more chips on one wafer. This increases the yield, as little spot errors in the silcon will affect a smaller percentile of chips. This, in turn, decreases the cost of producing one chip. This is mandatory to stay competitive. In order to make their CPU's more powerful, CPU manufacturers need an increasing number of transistors (the basic building stones of IC's). Without shrinking the process, CPU dies would be incredibly big by now, meaning that the yield during manufactering would be dreadfully low.

A process shrink is a double-edged sword in terms of IC temperature. Inherently, heat output goes down by shrinking the die, this is because the operating voltage for he transisors goes down. On the other hand, making the die smaller also decreases the surface through which it conducts this heat to your heatsink, making the heat transfer less efficient, thus increasing the temperature of the chip.

How much can it increase performance between GPUs? Apparently maxwell is going to have 28nm and 22nm GPUs.
Link to comment
Share on other sites

Link to post
Share on other sites

How much can it increase performance between GPUs? Apparently maxwell is going to have 28nm and 22nm GPUs.

As I said: inherently the litography doesn't affect performance at all. The only possible performance increase is from either more transistors being jammed in the IC, or the IC design being improved. Since we're talking about a die shrink, it probably will be the latter, in which case, we probably won't see that much of an improvement. If it's the former, permance improvement can be quite huge.

 

Just as a disclaimer: I have no knowledge about specific GPU's and their specification, expecially not about GPU's that aren't out yet. Everything I said in my previous post is purely the scientific side of the story. Anything I said here is based on what we have seen happening for the past years on the CPU side of IC manufactering (tic-toc paradigm from Intel) and may thus be completely wrong.

Edited by MG2R
Link to comment
Share on other sites

Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×