Hey Jeff, thanks for reminding me that on December 16th, 1947 William Shockley, John Bardeen & Walter Brattain created the first working transistor, the basic building block that helped build some nations and a few trillion dollar fortunes.
Six decades later, the computer business is facing a brand new set of challenges. Moore’s Law as we have known it is facing a ceiling, argues Associated Press.
Gordon Moore, the Intel Corp. co-founder who famously predicted in 1965 that the number of transistors on a chip should double every two years, sees that the end is fast approaching — an outcome the chip industry is scrambling to avoid. “I can see (it lasting) another decade or so,” he said of the axiom now known as Moore’s Law. “Beyond that, things look tough. But that’s been the case many times in the past.”
The ever smaller and always faster chips are beginning to run into physical limits, including the problems created by generating too much heat, and excessive power consumption. This has the industry scrambling for new technologies and radical new thinking.
Intel for instance is experimenting with different ways to keep it going, indulging in highly speculative technologies that include everything from optics to quantum computing. Others like Microsoft’s Craig Mundie are talking about multiple core chips, where each core is assigned a specific task. He compares them the future multi core chips to orchestras. The only problem, as The New York Times points out is the software that makes these chips do what they are supposed to ultimately do.
And while we wait for these big ideas come to fruition, the real future means packing these transistors tighter and closer in ways previously not thought, and ultimately putting them into mobile phones — the real computers of the future. Moore’s Law has always been equated with PCs.
The PC centric approach overshadows the fact PC represents the past of our industry, the future is mobile — not just mobile phones but a whole gaggle of mobile devices that haven’t been dreamed off yet.
In my Business 2.0 column, Moore’s Law 2.0, I quoted Drew Lanza, general partner with Morgenthaler Ventures, who pointed out that “Moore’s original research paper didn’t say anything about processor clock speed. It said you could, with every generation of chips, cram more transistors into the same space.”
As phone makers cram more features into cell phones – FM radios, TV tuners, Wi-Max, and ultra-wideband – chip designers will have to pack them intelligently into ever smaller circuits. Moore’s Law in the 21st century is about building these supercombo chips, not the fastest chip for your desktop. With nearly 1 billion mobile phones sold every year, this is an opportunity much larger than the PC market.
Here is to next 60 years!
7 thoughts on “Hey Transistor, Happy 60th. Now Move Closer”
Storage follows a similar law. It’s been called Kryder’s Law, and states that the number of bits per area doubles every year. Continued growth in storage may be even more important that in processors because the value of content stored there is growing on the same trajectory (or faster) than either physical device.
Good point Pete. I think there are some fundamental problems in the “physics” of technology as we know it that need to be tackled. From an innovation standpoint, it is perhaps the most exciting time – but I often wonder when it is going to happen. Last I checked the R&D investments by the US Government and US companies were shrinking.
storage is doing fine… Fujitsu Eagle in mid/late 80s
470 MB in 10-1/2 inches (6U) of 19-inch rack space, at a retail price of about US$ 10,000.
to TeraByte drives in a twentieth of the volume and a tenth of the price (if not less)
The ability to stuff more processors or more transistors onto a chip pales in comparison to the ability to create software to run on these megachips. That’s the real problem and it’s only being addressed now by companies investigating new standards and the use of virtualization to design software and hardware concurrently. You can get the performance of a multicore chip by using an accelerator, usually on a small FPGA, paired with an off the shelf processor. It costs a lot less, gives you what you need and takes weeks to deliver a design, rather than years.
Just because it can be done doesn’t mean you have to do it.