PDA

View Full Version : Moore's law coming to end ?



nec207
June 3rd, 2011, 07:22 PM
So we know Moore's law every 18 month processors double in speed and more and more transistors can fit on the IC .

The transistors get smaller and smaller .

But around 2004 processor clock speed hit brick wall of 3.0 GHz with out overheating the the solution to problem dual core!! But how much speed can they get out of a dual core now ? I mean in 2005 and 2006 it shined to see a dual core but 2011 dual core how much better.

I mean look we can look at intel core 2 duo and i3 or i5 but really the i3 or 15 is not going be 2 times faster than a intel core 2 duo .

The is a limit how much better you can make a dual core . I mean dual core has been around for over 5 years now.

Has for new state of art quad core not going to say much about it as I have not read up on quad core .

No idea when a 8 or 10 core will come out and how long we can go with many cores or when a engineering problem where they cannot put any more set number of cores.

No idea when intel will bring out a i9 may be in 2 years from now.

and1bskbl72
June 3rd, 2011, 07:35 PM
Apple Macpro now up to 12 cores processing power.

lykwydchykyn
June 3rd, 2011, 07:36 PM
Not so fast, there's hope yet:

http://www.youtube.com/watch?v=OO7gElMdO6U

Dustin2128
June 3rd, 2011, 07:45 PM
There's 3D transistors which I've heard about, plus graphene, quantum computing, and all else, I doubt it's ending in my lifetime.

and1bskbl72
June 3rd, 2011, 07:48 PM
3d is cool, I wonder how much more information will be sent through this method and what speeds it will reach.

JDShu
June 3rd, 2011, 07:55 PM
Programs are still not trivially parallelizable, so the benefits of more cores is small.

pqwoerituytrueiwoq
June 3rd, 2011, 08:03 PM
Not so fast, there's hope yet:

http://www.youtube.com/watch?v=OO7gElMdO6U

at the end when the shrink ray blows up the smoke particles should be like large bouncing balls when the guy is talking
then again you would suffocate at that size you would need to shrink the air in the room
just saying

sydbat
June 3rd, 2011, 08:23 PM
Not so fast, there's hope yet:

http://www.youtube.com/watch?v=OO7gElMdO6USilly.

Reminds me of this - http://www.imdb.com/title/tt0060397/

lisati
June 3rd, 2011, 08:30 PM
Silly.

Reminds me of this - http://www.imdb.com/title/tt0060397/
Wasn't there a remake in the 1980s?

sydbat
June 3rd, 2011, 08:34 PM
Wasn't there a remake in the 1980s?No. There was supposed to be a CGI remake in 2010, but that has been pushed to 2013, according to IMDB. I really hope they do NOT remake it!

lisati
June 3rd, 2011, 08:41 PM
No. There was supposed to be a CGI remake in 2010, but that has been pushed to 2013, according to IMDB. I really hope they do NOT remake it!

:confused: I've seen the original 1966 version, and I'm sure I've seen something remarkably similar made more recently, possibly with a different title. Actors Martin Short & Rick Moranis come to mind, and there are movies I've seen that aren't listed in IMDB

earthpigg
June 3rd, 2011, 08:42 PM
I think we've reached the point wherein twice as fast doesn't yield proportional benefits, so the incentive isn't what it once was.

I think I'd have been just fine without an i7 on this 2009 build, it is nowhere near "obsolete" two years down the road, and will probably be just fine for another 3 years or so without a single upgrade -- contrast that with how things worked with >$1500 computers in the late 90s and early 2000s.

Furthermore, from tablets to netbooks, software is becoming the defining feature and not hardware.

3Miro
June 3rd, 2011, 08:44 PM
Moore's law is about the size of the transistors only, the speed is a by-product.

In the past, more transistors meant faster clock, today it means more cores. Many apps are still single core only, but this is changing especially in the high-end, where more cores are indeed better. Most games today take advantage of multi-core architecture, video encoding could do that for quite some time, browsers use threading (or in the case of chromium completely separate processes) and scientific applications have been undergoing heavy parallelization for decades.

The only argument to be made is that on average people don't really need that much power.

sydbat
June 3rd, 2011, 08:55 PM
:confused: I've seen the original 1966 version, and I'm sure I've seen something remarkably similar made more recently, possibly with a different title. Actors Martin Short & Rick Moranis come to mind, and there are movies I've seen that aren't listed in IMDBSCTV was famous for making movie parodies, sometimes way more watchable than the originals...even if they were only 10 minutes long. So it is possible that you are remembering that. Or a scene from Spaceballs.

Alternately, there have been a lot of references (http://www.imdb.com/title/tt0060397/movieconnections) to the film and other parodies/ 're-enactments' in various movies and TV shows.

lykwydchykyn
June 3rd, 2011, 09:27 PM
:confused: I've seen the original 1966 version, and I'm sure I've seen something remarkably similar made more recently, possibly with a different title. Actors Martin Short & Rick Moranis come to mind, and there are movies I've seen that aren't listed in IMDB

You're either thinking of "Inner Space" or the "Honey I shrunk the kids" franchise, which were both spins on the "shrink ray" idea, but not remakes of that glorious 1966 awesomeness.

The intel video is horrendously cheesy, I admit, but it's a good explanation of 3D transistor tech for us non-engineers.

mips
June 3rd, 2011, 09:28 PM
:confused: I've seen the original 1966 version, and I'm sure I've seen something remarkably similar made more recently, possibly with a different title. Actors Martin Short & Rick Moranis come to mind, and there are movies I've seen that aren't listed in IMDB

Innerspace http://www.imdb.com/title/tt0093260/

Moore's law ain't going anywhere, if it's not clock speed it's die size, parallel computing etc.

wolfen69
June 3rd, 2011, 09:49 PM
:confused: I've seen the original 1966 version, and I'm sure I've seen something remarkably similar made more recently, possibly with a different title. Actors Martin Short & Rick Moranis come to mind, and there are movies I've seen that aren't listed in IMDB

Honey, I shrunk the kids?

mips
June 3rd, 2011, 10:20 PM
Honey, I shrunk the kids?

Probably Martin Short & Dennis Quaid in Innerspace (http://en.wikipedia.org/wiki/Innerspace).

Old_Grey_Wolf
June 3rd, 2011, 10:43 PM
Moore's Law is an average over a longer period of time that 18 months. There have been times in the past when there seemed to be no progress then there was a sudden jump.

In the current economy, companies are struggling to stay profitable. They have cut their research and development budgets in order to reduce the price of their products. I think when the economy improves we will start to see progress again.

sydbat
June 3rd, 2011, 11:05 PM
In the current economy, companies are struggling to stay profitable.This completely explains the quality of the Intel video...:p

The rest of what you said is pretty well on point.

Bandit
June 3rd, 2011, 11:59 PM
Although there are still some areas to tweak out more performance of current X86 CPU design, I think we have pretty much come to the end of the modern silicon microprocessors as we know them. There were talks a few years back about going with Quartz and Diamond based chips, but I havent seem much about it recently (prob due to recession/funding).

nec207
June 4th, 2011, 03:51 AM
Just to be clear Moore's law has nothing to do with clock speed.It just getting more and more transistors on IC .

I think that may be do to video cards have good GPU and lots of VRAM for games and video editting.

It is it is mostly video editing and games where you need a good video card or CPU.

nec207
June 4th, 2011, 10:57 PM
No. There was supposed to be a CGI remake in 2010, but that has been pushed to 2013, according to IMDB. I really hope they do NOT remake it!

What was is so wrong with that movie you do not want a remake that is HD and new special effects.

Anyone read up on AMD bulldozer processor 12 core?

I know AMD has some Athlon 2 and phantom 2 that are 4 cores and some 6 cores.

intel i7 is 4 core with hyper threading that are like 8 cores.

beew
June 4th, 2011, 11:09 PM
Why is it even called a "law"?? In the real sciences such as physics there are a lot more stringent criteria to be called a "law" than to make a soundbite by pulling some numbers based on some broad observations and snapshots.

Dustin2128
June 5th, 2011, 12:00 AM
What was is so wrong with that movie you do not want a remake that is HD and new special effects.

Anyone read up on AMD bulldozer processor 12 core?

I know AMD has some Athlon 2 and phantom 2 that are 4 cores and some 6 cores.

intel i7 is 4 core with hyper threading that are like 8 cores.
I'm waiting until bulldozer comes out before I build my next 'puter actually. 12 cores... =P~

PhillyPhil
June 5th, 2011, 01:51 AM
I know AMD has some Athlon 2 and phantom 2 that are 4 cores and some 6 cores.

intel i7 is 4 core with hyper threading that are like 8 cores.

Hyperthreading is just one core pretending to be two. In some cases (depends on what you're doing) it can give a small increase in performance, compared to a plain single core. In other cases it will give a decrease in performance.

In no way can a single hyperthreading physical core be compared to two physical cores (in terms of performance). So much better in a comparison like the one above to just say ''intel i7 is 4 core.'' and stop there - hyperthreading is irrelevant in a core count comparison.

Simon17
June 5th, 2011, 07:22 AM
Why is it even called a "law"?? In the real sciences such as physics there are a lot more stringent criteria to be called a "law" than to make a soundbite by pulling some numbers based on some broad observations and snapshots.

No there aren't. A physical law is something describes some behavior. There's nothing more to it than that.

swoll1980
June 5th, 2011, 07:49 AM
I don't think there is a need for that type of rapid anymore. The processors that are out there now go well beyond what 99% of users would ever need.

PhillyPhil
June 5th, 2011, 08:50 AM
I don't think there is a need for that type of rapid anymore. The processors that are out there now go well beyond what 99% of users would ever need.

Surely history has taught us statements like this are not a good idea? :P

''640K ought to be enough for anybody''

''There's only a market for 5 computers in the world''

(I'm paraphrasing, btw. Can't remember the exact quotes)

I would say until a computer can simulate the universe, or at very least the galaxy, i.e. never, we will always want and be able to use more computing power.

ssam
June 5th, 2011, 01:01 PM
12 core CPUs have been around for a while (and you can put 4 of them in a high end motherboard).
http://en.wikipedia.org/wiki/Opteron#Models

transistors can't get much smaller, they are already only a few hundred atoms across.

lykwydchykyn
June 5th, 2011, 10:00 PM
transistors can't get much smaller, they are already only a few hundred atoms across.

http://www.geekosystem.com/7-atom-transistor/

This seems like we're getting to the lower limit, methinks.

nec207
June 6th, 2011, 08:19 PM
12 core CPUs have been around for a while (and you can put 4 of them in a high end motherboard).
http://en.wikipedia.org/wiki/Opteron#Models

transistors can't get much smaller, they are already only a few hundred atoms across.


Not for home use.

forrestcupp
June 6th, 2011, 10:52 PM
Moore's law is about the size of the transistors only, the speed is a by-product.
Exactly what I was going to say. It's not about clock speed; it's about die size. Instead of doubling clock speed, they're shrinking die sizes which enables them to run more efficiently and add more features.

They're still very much on track with Moore's Law, and the trend is expected to continue until sometime between 2015-2020. The new 3D transistors will probably give that a boost.