Sign in with
Sign up | Sign in
Your question

Why does performance increase as the cpu's get smaller?

Last response: in CPUs
Share
a c 186 à CPUs
January 24, 2012 5:11:10 AM

How come a small chip would outperform a larger chip? Shouldn't the larger design be able to cram more cores/things inside of it? Then again I am just a 14 year old boy who hasn't learned about any of this yet.
a c 223 à CPUs
January 24, 2012 5:30:08 AM

Physical size is not really the important part.

Its the size of each individual transistor (e.g 22nm) that is important.

As technology progresses we are able to make each circuit or transistor smaller. Allowing us to fit physically more transistors in a giving space. The decrease in transistor size is faster than the performance increase we require. Giving us a trend where chips get physically smaller and smaller.

A smaller transistor also gives us better efficiency and thus performance because of resistance and other complicated physics.
m
0
l
January 24, 2012 5:58:50 AM

Also, the electrons that do all the processing work travel smaller distances inside the chip, which means they get the work done faster. True story!
m
0
l
Related resources
a c 186 à CPUs
January 24, 2012 7:03:02 AM

Ah I see! So nm are the transistor sizes!!!!
m
0
l
a b à CPUs
January 24, 2012 7:06:21 AM

the difference isn't in the size as in the efficiency of the design and the effectiveness of the transistors.

the chips right now have a lot more transistors than chips 10 years ago even though current chips are smaller.
m
0
l
a c 186 à CPUs
January 24, 2012 7:11:06 AM

So what happens after we get to 1nm transistors!???
m
0
l
a b à CPUs
January 24, 2012 7:16:23 AM

we can only get to 10 nm with the current technology as light cutting into silicon is getting to its limit as well as 10nm being only a handful of atom across.

we will need to find better transistor technology by then or have working quantum computers to advance much further.
m
0
l
January 24, 2012 7:40:08 AM

an easy comparison why smaller ships (or rather smaller transistors) can result in faster processors is this. Think about running laps. If you run a circle that's 400m long or if you run a circle that's 100m long. Which circle could you run more often in 10 minutes.

The same goes for chips. if parts are smaller it takes less energy to flip the switches inside so that means we can turn them on/off faster. (and the same goes for other parts inside the cips).
m
0
l
a c 223 à CPUs
January 24, 2012 7:59:51 AM

nm = nano meter.

I believe that before we hit 1nm we will be using quantum and light based computing.

We can not even comprehend the limitations of that yet.

Transistor nm (meaning performance per watt increased as well) used to decrease exponentially but it has started balancing out due to limitations on what we are able to manufacture cost effectively. Or just be able to produce at all. As esrever said.
m
0
l
a b à CPUs
January 24, 2012 11:37:08 AM

This may be stupid but when we reach the 10nm limit, why not increase the size of the silicone say X2? Technically we could get X2 the amount of 10nm transistors in.
m
0
l
a b à CPUs
January 24, 2012 12:32:09 PM

Because then the chip would be twice as expensive.
m
0
l
a b à CPUs
January 24, 2012 12:46:03 PM

There are materials in the works that are capable of bypassing the law of quantum tunneling.

I personally would prefer quantum computing to stay AWAY. The only thing as powerful as a Quantum Computer is our Brain - thats because it IS one. A Quantum computer has infinite possibilities - Quantum computers would be capable of evolving alone. Scary stuff. Id rather be stuck on the ground and not knowing anything more about the universe than getting destroyed by Skynets IRL big brother.
m
0
l
a c 223 à CPUs
January 24, 2012 1:08:27 PM

silverliquicity said:
This may be stupid but when we reach the 10nm limit, why not increase the size of the silicone say X2? Technically we could get X2 the amount of 10nm transistors in.


That is exactly whats already being done. Socket 1366 and now 2011.

Sandy-bridge extended 2011 is almost twice the transistor count of Sandy-bridge 1155.
m
0
l
a c 223 à CPUs
January 24, 2012 1:10:13 PM

deadjon said:
There are materials in the works that are capable of bypassing the law of quantum tunneling.

I personally would prefer quantum computing to stay AWAY. The only thing as powerful as a Quantum Computer is our Brain - thats because it IS one. A Quantum computer has infinite possibilities - Quantum computers would be capable of evolving alone. Scary stuff. Id rather be stuck on the ground and not knowing anything more about the universe than getting destroyed by Skynets IRL big brother.


Off topic : DAMN! Look at that overclock! 5.02GHz!!! I bow to you!

What cooler are you using?
m
0
l
January 24, 2012 1:28:46 PM

deadjon said:
There are materials in the works that are capable of bypassing the law of quantum tunneling.

I personally would prefer quantum computing to stay AWAY. The only thing as powerful as a Quantum Computer is our Brain - thats because it IS one. A Quantum computer has infinite possibilities - Quantum computers would be capable of evolving alone. Scary stuff. Id rather be stuck on the ground and not knowing anything more about the universe than getting destroyed by Skynets IRL big brother.



It's too late deadjon... Skynet is already online!!

http://www.dwavesys.com/en/products-services.html
m
0
l
a b à CPUs
January 24, 2012 1:44:20 PM

Novuake said:
Off topic : DAMN! Look at that overclock! 5.02GHz!!! I bow to you!

What cooler are you using?


Phanteks PH-TC14PE - I don't have it @ 5.02Ghz 24/7 though - It will cut the life of my processor down like crap.

Its an AMAZING air cooler - I'm using it with the stock thermal paste and getting 50c Max Temp @ 4.6Ghz 1.35v after 4 Hours Prime.
m
0
l
a c 223 à CPUs
January 24, 2012 3:18:51 PM

deadjon said:
Phanteks PH-TC14PE - I don't have it @ 5.02Ghz 24/7 though - It will cut the life of my processor down like crap.

Its an AMAZING air cooler - I'm using it with the stock thermal paste and getting 50c Max Temp @ 4.6Ghz 1.35v after 4 Hours Prime.


Will definitely be taking a look into that cooler. Thanks. Sorry missed the Phanteks part in your signature originally.
m
0
l
!