End of Moore's Law and future processor stagnation *spawn

[Albuquerque]

Worrying about Moore's law may become academic before it becomes physically and economically impossible.

Truly, this should be the point of this thread. Is a divergence from Moore's law any sort of concern? A law that Moore himself really didn't specifically intend to dictate or create, and further an observation that he himself did not think would continue in perpetuity?

There are ways that we can continue doubling transistor count -- we can make bigger chips, we can stack them, we can make them smaller. But all of these growth dimensions have physical limitations, some of which are more pressing (at this time) than others.

Maybe the better question is: do we really need it to continue forever? We can tackle problems in different ways now. We do not live in a world where only the primary CPU can do useful work; multi-core, multi-socket, multi-node-clustered, distributed, or otherwise "cloudified" compute can do a lot more of that work and potentially at a lot less cost than a monolithic humongous CPU.

I think it's time that we stop worrying about how to address the continuation of Moore's unsustainable law, and work towards solving the problems we actually care about.

 

[Orion]

Truly, this should be the point of this thread. Is a divergence from Moore's law any sort of concern? A law that Moore himself really didn't specifically intend to dictate or create, and further an observation that he himself did not think would continue in perpetuity?

There are ways that we can continue doubling transistor count -- we can make bigger chips, we can stack them, we can make them smaller. But all of these growth dimensions have physical limitations, some of which are more pressing (at this time) than others.

Maybe the better question is: do we really need it to continue forever? We can tackle problems in different ways now. We do not live in a world where only the primary CPU can do useful work; multi-core, multi-socket, multi-node-clustered, distributed, or otherwise "cloudified" compute can do a lot more of that work and potentially at a lot less cost than a monolithic humongous CPU.

I think it's time that we stop worrying about how to address the continuation of Moore's unsustainable law, and work towards solving the problems we actually care about.

The end of miniaturization will definitely have an impact on the market. I mean what's going to happen to things like smartphones? The improvements from this years model to next years model will vanish.

It is also likely that unless power consumption becomes negligible, things like gpus will also hit a limit and be unable to increase in performance.

While it is true that we can make do with whatever we have when we finally hit the limit, the disappearance of improvements will be a bit of shock for most consumers who've grown accustomed to them.

 

[rpg.314]

Feel free to quote Moore where he talked about shrinking in his "law". Also feel free to point out the fixed cost portion of it too, while you're there.

The statement was doubling of transistors, that's it. The rest of the baggage that you describe is not part of it at all, in fact the cost (for a very long time) was absolutely NOT static, nor was the die size, and yet a lot of it happened on the same lithography size.

This is exactly why I felt like a reminder was in order, as a lot of things get attributed to Moore's law that really aren't part of it at all. Some folks say a doubling of performance, some folks equate it to equal capability at half cost, some others equate it to half power. Some of those are things that have come to fruition over the years, but they aren't part of Moore's law.

As 3d pointed out, lowest transistor cost is the invariant involved. Doubling size on the same node is not part of Moore's law.

I am not aware of doubling transistors on the same node without doubling your costs, and how it might have happened in the past. Simply, defect density would drive up costs for larger sizes non linearly, beyond the increase in cost due to fewer die candidates. I don't see which trends in the past might have precluded that.

 

[Exophase]

In the past there had to have been stricter limits on the largest dies you could make due to smaller reticles. Also scaling was probably worse since they had smaller wafers, and poorer yields due to larger features w/less room for redundancy or even the mechanisms to handle redundancy.

These are just guesses though, my knowledge of the history of IC manufacturing from 30+ years ago is zero.. as opposed to my knowledge of manufacturing today which is merely poor

 

[eastmen]

The end of miniaturization will definitely have an impact on the market. I mean what's going to happen to things like smartphones? The improvements from this years model to next years model will vanish.

It is also likely that unless power consumption becomes negligible, things like gpus will also hit a limit and be unable to increase in performance.

While it is true that we can make do with whatever we have when we finally hit the limit, the disappearance of improvements will be a bit of shock for most consumers who've grown accustomed to them.

Things get bigger again. I have no problem with large cases for fast cpus / gpu and what not.

Of course then you run into other problems but things can be improved that way also.

 

[Orion]

Things get bigger again. I have no problem with large cases for fast cpus / gpu and what not.

Of course then you run into other problems but things can be improved that way also.

This strategy can't last long, you can't be doubling and doubling performance by increasing size for long. Power consumption, costs, etc will make it prohibitive.

 

[gamervivek]

 

[Davros]

Id love to interview him
"So jen do you realise than if the whole pc industry adopted nvidia's practices the industry would be destroyed"

 

[eastmen]

This strategy can't last long, you can't be doubling and doubling performance by increasing size for long. Power consumption, costs, etc will make it prohibitive.

Depends on when it happens I would wager. I don't think anyone would care if we built Magrathea on Jupiter or Uranus