Moore’s Law vs. Tech Progress
UNFINISHED
You’ve heard of Silicon Valley, right? It’s an area of California where tons and tons of tech companies (not only related to computers) reside, slowly bringing the future to us. The reason it’s called Silicon Valley is because silicon is one of the main ingredients in computer chips. It also comes from sand. Observe:
Well, in the near future, a new substance could replace silicon (I guess nobody is 100% sure yet)… It’s called Gallium, and, though it has a couple hurdles to get over, it has the capacity to make 10x faster chips. (but “Gallium Valley” isn’t going to sound as good, so they’re going to have to think of something…) If you’re interested, you can read about its promise and challenges here.
So, this sounds pretty hopeful. But then we have “Moore’s Law”…
Most people who know about Moore’s Law understand it as this technological law of nature, where chip speeds somehow, miraculously double in speed every 2 years… (this applies to processors, video cards, and various other things… even things like hard drives, which double in size every two years)
(note for techies: now, technically, it’s not a 2x increase in “speed”, but a 2x increase in transistor counts… but that USED TO be more tied in to speed. However, in this new realm of dual-core, quad-core, etc, only certain programs know how to make full use of all the cores, so they’re the only ones that surge forward in speed)
What I’m here to point out that this every-2-year-speed-doubling should be thought of as a deliberately planned business goal, and not a “how do they keep doing it” miracle. See, even if they could technically advance technology much faster, it doesn’t seem they have any interest in it whatsoever… Instead, it’s like they go after a PLANNED/TARGETED yearly increase, which always seems to be exactly the same (like you also see in video cards, hard drives, camera sensor technology (ISO sensitivity and megapixels), and various other tech fields).
[original chart stolen from Wikipedia]
Even if advancement is as simple as dumping more money into R&D (or, if you can’t afford it, trying to get government/investor funding for that, etc), it seems they wouldn’t do that.
So, what does this mean for gallium? Is the 10x speed increase coming?
Unfortunately, the linearity of “Moore’s Law” suggests to me that the biggest chip-makers probably have no intention whatsoever of speeding up the advancement of processors. (maybe the makers of “mobile” processors might want to (cell phone processors, etc), but unlikely for computers)
See, it couldn’t possibly be the case that, every year, a scientist at Intel stands up, and says, “HOLY CRAP!!! HOLY CRAP, WE’VE DONE IT AGAIN!!! WE’VE FIGURED OUT HOW TO MAKE OUR CHIPS EVEN FASTER!!!!! OH, IT’S A MIRACLE!!!””
It’s unsettling, once you realize that they’ve been making each year’s SAME miraculous breakthrough since the early 70’s. Doesn’t this mean that at any given point along the way, they had the KNOW-HOW to advance CPU’s much further?
I think it’s much more likely that the first gallium chips would be placed perfectly in line after the silicon ones, speed-wise (ignoring transistor-counts), and make the excuse that they’re doing this to save energy, and be green, or something (see, gallium would probably use tons less energy at that speed).
This is just guessing, but something even worse MAY have happened when the first Solid State hard drives came out, a few years ago. See, hard drives were reaching 1TB in size, a point that hard drive manufacturers probably fear, so these new drives, based on this new tech, did a size RESET, and started off painfully small (both the Flash-based and the DRAM-based drives. Two different technologies, both perfectly capable of one day making huge drives)… like around . (Size was key. Why make each year’s hard drives a lot FASTER when you can just make them LARGER; the thing people (wrongly) care about? That’s one less area to put money into)
-
Before we continue, though, there’s something you have to understand: knowing how to do something (like make faster CPU’s) doesn’t necessarily mean you can snap your fingers, and make it happen. When it comes to the tech world, invention is 25% idea, and 75% figuring out how on Earth to manufacture it cheaply enough. See, if it’ll cost $10 to make something, but nobody would pay more than $5 for it in stores, the invention goes to that big, inventioney place in the sky. Unfortunately, you can only sell something at a price that people are willing to pay… and because of this challenge, there are a LOT of great inventions out there just sitting on shelves, waiting for the world’s manufacturing infrastructure to evolve until those things can be made more cheaply (or until we know how to make them at all. See, on PAPER, I might be able to design a molecule-sized nano robot that could fight diseases (and work), but how do you MAKE a robot that small? First, we have to invent the machines that can tackle that). So, consider this: as I understand it, advancing CPU’s is more about inventing a way to MAKE new kinds of chips, and less about how each chip will actually work.
So, does this affect Intel? Is it simply too expensive to say, “Hey, instead of working on next year’s chip right now, why don’t we just skip one?” (I also wonder how much cost could be saved by making only 1-3 different chips each year, rather than 10+… Why not just make: the fast one (i7), the one for servers (Xeon), and the one for small, “mobile devices” (Atom)?)
Can that be done without slipping into the red? Let’s say the answer is “no” (because it could be). That’d be where I say that the world’s governments should all pitch in to fund top-speed CPU development. See next:
-
A 2x speed-increase every 2 years may seem fast to some, but I’ll explain why I think this progression should be ridiculously faster (aside from the simple fact that the entire world’s technological progress appears to be “capped” (given a hard limit), for the sake of making money… in other words, maintaining the slowness and limitations of technology for EVERYONE, so that 0.000001% of the world’s population – the rich – can make dough).
[move paragraph up higher]
Smaller chip-makers – if they could – would LAUNCH upon the opportunity to speed up CPU development, and thus make billions while the larger companies crumbled for not playing catch-up… but it seems to me that those who control the industry have long known that THEIR best interests lie in trying their doggone hardest to not allow speeds to advance faster than is most profitable (the inch-up method). Or else you get the “Ahhh, finally, I have a comfy computer speed, so I don’t have to upgrade in 3 years,” effect… = less dough.
I personally see it as SUPREMELY important to the advancement of humanity to always advance technology at the max possible rate, so that programmers can get started on the programs needed to:
- Control and think for complicated physics-driven robots
- Run “augmented-reality”apps, which process live HD video feeds all around you, to show you more info about your surroundings… such as while shopping, working, driving, etc)
- Do the thinking for self-driving cars, that would similarly crunch feeds from multiple HD cameras, to watch the roads, while also communicating with thousands of other cars at once. Imagine living in a world where nobody crashes, and you can be in bumper-to-bumper traffic, cruising at 200mph. With computer-controlled cars that communicate with each other, you could do something like that)
- Do the thinking for in-house cameras that watch/listen to your gestures/voice, and project screens on all the walls around you (in properly massive resolution). (a fluid system of this sort would require more than a dozen small HD cameras, and enough CPU power to track a crowd of people within the home)
- Run incredibly dense 3d simulations, such as games, but also allowing the introduction of detailed 3d schools/universities/etc that could be attended digitally. For maximum effect, it would require enough CPU power to show accurate representations of very many students – and their projects – at once. Similarly, the idea of “digital locations” would finally become widespread, because they could finally not look like junk, like in Second Life.
- Power a much, much more powerful internet, where you can do things like:
- Digitally attend conventions and trade-shows (imagine being at “Tech Convention 2020” (made-up convention), with hundreds of thousands of tiny cameras swarming the ceilings, each controlled by a person on the internet, rented for $2). If technology was made into a supremely-attractive UNFINISHED
I personally believe there is FAR more innovation possible for the world of CPU’s than the CPU-makers seem to realize. Trouble is, no programmers (or companies) even THINK of making these things, because the CPU power required is too far out of reach to even CONSIDER. I personally think the future will hold more CPU’s in the walls and devices around us than in our computers, so, in my view, what CPU-makers need in order to get out of the current stagnation they’re facing is something to flatten our current CPU’s, and start a new generation of not-just-desktop-computers computing.
Trouble is, at THIS rate, we won’t reach some of these things within the next 25 years, if we don’t develop WAY more powerful CPU’s… such that we could have pushed for decades ago, just that the ring-leaders have likely always been AFRAID of having them designed. (and we’re finally reaching their dreaded stagnation point, from which there is no return… if you don’t press for mass innovation). Let’s start with cars… we’ve almost reached a billion of them. (now, would all car makers WANT cars that don’t crash, reducing sales? Not sure (it could vary from company to company), but, since 1 in 5 people have been in some kind of accident in the last 5 years, there’s at least the incentive there for a number of companies to want this crashing to continue. Every wreck is a car sale.)
- All Alphaila articles can be re-posted and sent around freely, without permission. See FAQ page.
- Get my free wallpaper pack | About the author