Moore's Law Dead by 2022, Expert Says

[Renegade]

All good things must come to an end?

http://www.eetimes.com/document.asp?doc_id=1319330

Moore's Law -- the ability to pack twice as many transistors on the same sliver of silicon every two years -- will come to an end as soon as 2020 at the 7nm node, said a keynoter at the Hot Chips conference here.

While many have predicted the end of Moore's Law, few have done it so passionately or convincingly. The predictions are increasing as lithography advances stall and process technology approaches atomic limits.

"For planning horizons, I pick 2020 as the earliest date we could call it dead," said Robert Colwell, who seeks follow-on technologies as director of the microsystems group at the Defense Advanced Research Projects Agency. "You could talk me into 2022, but whether it will come at 7 or 5nm, it's a big deal," said the engineer who once managed a Pentium-class processor design at Intel.

Moore's Law was a rare exponential growth factor that over 30 years brought speed boosts from 1 MHz to 5 GHz, a 3,500-fold increase. By contrast, the best advances in clever architectures delivered about 50x increases over the same period, he said.

Exponentials always come to an end by the very nature of their unsustainably heady growth. Unfortunately, such rides are rare, Colwell said.

"I don't expect to see another 3,500x increase in electronics -- maybe 50x in the next 30 years," he said. Unfortunately, "I don't think the world's going to give us a lot of extra money for 10 percent [annual] benefit increases," he told an audience of processor designers.

Colwell poured cold water on blind faith that engineers will find another exponential growth curve to replace Moore's Law. "We will make a bunch of incremental tweaks, but you can't fix the loss of an exponential," he said.

DARPA tracks a list of as many as 30 possible alternatives to the CMOS technology that has been the workhorse of Moore's Law. "My personal take is there are two or three promising ones and they are not very promising," he said.

More at the link.

On the plus side, it might make it easier to buy ASICs for cryptocurrency mining in a few years.

[eleman]

Yeah, they said the same thing about 28 nm, and then 20 nm.

Moore's second law: Once every two years an "expert" will predict the demise of Moore's Law.

p.s.: Moore's law is long dead from average joe's perspective anyway. My six years old core 2 e7200 has comparable single thread performance to today's i3 somethings (which are curiously sold at the same price level my e7200 sold for in the day). Intel and AMD continue to add new cores but I no longer see real-world performance gains doubling once every two years. Well, at least they had a good run from mid-nineties to late-oughties anyway.

[Renegade]

My current box is 4 years old (about), and still has LOTS of power. It was mid-range when I got it too.

Few people need what many computers can do now. They do email, Facebook, and not much else.

[wraith808]

Yeah, they said the same thing about 28 nm, and then 20 nm.

Moore's second law: Once every two years an "expert" will predict the demise of Moore's Law.

p.s.: Moore's law is long dead from average joe's perspective anyway. My six years old core 2 e7200 has comparable single thread performance to today's i3 somethings (which are curiously sold at the same price level my e7200 sold for in the day). Intel and AMD continue to add new cores but I no longer see real-world performance gains doubling once every two years. Well, at least they had a good run from mid-nineties to late-oughties anyway.

Yeah... wishing and saying isn't going to make it so.  And there was no scientific proof in that article that I could see... just convincing reasoning.

[Vurbal]

Yeah, they said the same thing about 28 nm, and then 20 nm.

Moore's second law: Once every two years an "expert" will predict the demise of Moore's Law.

I think the first time I saw that claim was all the way back when Intel's 180nm process CPUs hit the market.

p.s.: Moore's law is long dead from average joe's perspective anyway. My six years old core 2 e7200 has comparable single thread performance to today's i3 somethings (which are curiously sold at the same price level my e7200 sold for in the day). Intel and AMD continue to add new cores but I no longer see real-world performance gains doubling once every two years. Well, at least they had a good run from mid-nineties to late-oughties anyway.

Yep. We've long since reached the point where the limits on the useful lifetime of a computer were determined primarily by component durability rather than speed or the selection of technical features. For the typical computer user that happened more than a decade ago. For most power users it was probably 5 years back.

[Deozaan]

I remember, perhaps 15 years ago, my older brother telling me that someday we'd reach a point where the CPU's hz wouldn't matter anymore because it would eventually be so high that any further increase wouldn't give us any practical improvement. He guessed that advances in RAM (both amount of RAM and speed of RAM) would be the next deciding factor in PC speeds for the consumer.

It seems we've pretty much met his first prediction. IMO RAM is approaching that point as well. What I mean by that is it seems it won't be long until we can have so much RAM in our systems that we won't do ourselves any good to put more in them. At least for desktop machines. Mobile devices still could use more.

Then again, mobile devices could still use more of just about everything, as it's harder to fit the more powerful things inside the smaller cases and also deal with heat/battery issues. But that's improving as well. (c:

Gotta love the progress of technology.

[x16wda]

Then again, mobile devices could still use more of just about everything, as it's harder to fit the more powerful things inside the smaller cases and also deal with heat/battery issues.

Bingo. I think battery and mobile device power technology is begging for "an exponential" to ride for awhile.  Where's Nikola Tesla when you really need him??

[Renegade]

I remember, perhaps 15 years ago, my older brother telling me that someday we'd reach a point where the CPU's hz wouldn't matter anymore because it would eventually be so high that any further increase wouldn't give us any practical improvement. He guessed that advances in RAM (both amount of RAM and speed of RAM) would be the next deciding factor in PC speeds for the consumer.

It seems we've pretty much met his first prediction. IMO RAM is approaching that point as well. What I mean by that is it seems it won't be long until we can have so much RAM in our systems that we won't do ourselves any good to put more in them. At least for desktop machines. Mobile devices still could use more.

I wonder about that. Developers seem to love soaking up more and more resources as they become available. Take Chrome or Firefox for example... 1 GB for 1 process? Really? I only wish I were kidding... Just for example, I checked right now... not doing anything special and not trying to use memory - just as things are...

[attachimg=#]

1.7 GB!?!

Right now I have 13 processes taking up more than 100 MB of RAM. This is just crazy.

[Deozaan]

Right now I have 13 processes taking up more than 100 MB of RAM. This is just crazy.

That would be nice...

[attachthumb=#]

[SeraphimLabs]

Servers will always benefit from faster CPUs and more RAM.

Improvements in CPU and RAM for servers translates to either more clients served, or a smaller server in general because the growth of the technology has exceeded the growth rate of the transactions being processed.

For desktops though, the resource usage has indeed slowed down considerably, due in no small part to the 32/64 bit changeover holding back a lot of software.

Considering that my 3.4GHz Pentium 4 HT from 2005 is still able to reliably perform all of the basic computing tasks- internet, email, media playing, and storing personal data, I believe technology has reached a plateau where simply increasing the performance is no longer enough to bring about another radial change in how people use this type of equipment.

Now its just a game of making it cheaper and more energy efficient, while the market is saturating because there is far less of an incentive to upgrade all the time than there used to be.

[Vurbal]

Ultimately, it seems to me the future is the continued transformation of components and processors which control peripherals to a pure heirarchy of integrated, but discrete "computers," much like modern networking has been transformed into a heirarchy of contextual peers and servers.

The technology for this is already being used to affect fairly dramatic improvements in server performance. I'm referring to the amazing high end, multiport, PCIe server network cards. They achieve near independence from both the OS and underlying CPU control, even going so far as heuristic (I'm guessing) determination of which physical CPU's PCIe interface to send traffic across.

[40hz]

I think the thing that may ultimately do it is the widespread adoption of "good enough" restricted platform tablet devices by the majority of the public.

And as a result, the future limits of this technology will be increasingly determined by politics rather than genuine "technical" roadblocks.

I think most big companies, and government in general, would be very happy if the personal and open computing platform - as we have heretofore known it -  just went away. Far too much power in the hands of the masses. And far too many opportunities for innovation - which then leads to competition against established mainline firms from 'untrusted' and 'upstart' competitors.

Not building real personal computers negates the need for better and faster chips. Not developing better and faster chips destroys the opportunity to design better and more powerful personal computers. QED.

"If you don't build it, they can't come." to paraphrase Shoeless Joe.

[Renegade]

Right now I have 13 processes taking up more than 100 MB of RAM. This is just crazy.

That would be nice...
(see attachment in previous post)

Hahaha!

It's called 'closing tabs'. Try it sometime!

I used to be really bad at that, but finally managed to discipline myself to start closing tabs and windows. 15 Chrome windows open with 10, 20, 50 tabs open? Nuts. But, but... I might NEED to open one of those again! And... I found myself searching through countless windows & tabs to actually find something when it would have been faster to just open a new tab and navigate back to the page. I still have 30 tabs open right now, but I'm a heck of a lot better than I used to be.

Chrome will suck up every bit of memory that it can. It's just a greedy memory hog.

[Renegade]

And as a result, the future limits of this technology will be increasingly determined by politics rather than genuine "technical" roadblocks.

Sadly, that is likely to happen.

I think most big companies, and government in general, would be very happy if the personal and open computing platform - as we have heretofore known it -  just went away. Far too much power in the hands of the masses. And far too many opportunities for innovation - which then leads to competition against established mainline firms from 'untrusted' and 'upstart' competitors.

It will likely go the same way as the automobile with incremental regulation a drop at a time. It will resemble one of those "pen dot" drawings where the artist merely makes dots on the paper. Each dot is nothing more than a dot until you zoom out far enough to see the entire picture.

Any for those of us that scream and kick and fuss about each dot, we'll be ridiculed because "it's only a dot" and we're "conspiracy theorists"... until the picture is made full and people finally wake up and realise that they are looking at their own prison funeral.

Death by a thousand pin pricks.

[eleman]

Considering that my 3.4GHz Pentium 4 HT from 2005 is still able to reliably perform all of the basic computing tasks- internet, email, media playing, and storing personal data, I believe technology has reached a plateau where simply increasing the performance is no longer enough to bring about another radial change in how people use this type of equipment.

Now its just a game of making it cheaper and more energy efficient, while the market is saturating because there is far less of an incentive to upgrade all the time than there used to be.

Assuming you had replaced your 90 watt prescott p4 with a 45 watt core 2 sometime in 2009, and assuming you use the computer on an average of 4 hours a day, you'd have recouped the cost of the new processor by now through the electricity bill (here we pay something like .2 dollars per kW/h). You really need to replace that intel branded space heater

Right now I have 13 processes taking up more than 100 MB of RAM. This is just crazy.

That would be nice...
(see attachment in previous post)

Hahaha!

It's called 'closing tabs'. Try it sometime!

And experiment with outsourcing the gmail tab to a dedicated browser such as ie. My firefox would grow and grow to 1.5 GB ram use, at which point it would be unusable. Now I use ie for gmail, and firefox for all the rest. Neither one will grow to unmanageable sizes. Gmail is one of the worst offenders in ram use.

[Deozaan]

Right now I have 13 processes taking up more than 100 MB of RAM. This is just crazy.

That would be nice...
(see attachment in previous post)

Hahaha!

It's called 'closing tabs'. Try it sometime!

They close?!

[TaoPhoenix]

My current box is 4 years old (about), and still has LOTS of power. It was mid-range when I got it too.

Few people need what many computers can do now. They do email, Fac

My current box is 4 years old (about), and still has LOTS of power. It was mid-range when I got it too.

Few people need what many computers can do now. They do email, Facebook, and not much else.

I'll update this a hair but I basically agree with it.

My box was quasi-high-end on purpose when I had a buddy build it cold.

I'm not just a "Facebook-email Newb", but if ya think about it, posting on like seven more sites and running a few new progs, I nailed the sweet spot too, and don't need much more beyond silly "end of life for XP" concerns.

I don't do hardcore gaming, so there's very little else that I could even want to do that requires more than (the original) quad core engine and twin Terabyte drives really needs. Yeah, I have a couple of system maintenance things I've been putting off, but performance wise, I thought long and far ahead and did it right and basically don't need much more.