As covered in the ExtremeTech article here, the semiconductor industry is living the reality that Moore’s Law is dead. Much like a marriage that has ended (not necessarily through divorce), it is appropriate to reflect upon the relationship and what it meant. In this case, was Moore’s Law the solid bond that the software industry could rely upon forever? The answer is most certainly no.
The software industry has been able to comfortably rely on the hardware industry to deliver results for decades. Surely most technologists will agree that semiconductors are subject to the laws of physics, which is why the end was inevitable. Until new materials or approaches to using current materials become available (I don’t want to rule that out as a possibility), Moore’s Law is dead.
An Out of Balance Relationship
Like the spouse that manages all of the finances, the kids, works to bring home the bacon and does all of the chores, there is extreme imbalance in the hardware and software marriage. It turns out, the hardware industry has been pulling off enormous feats to obtain what will prove to be rather meager returns on investments. It costs billions of dollars to produce new computer processors.
Concerns raised by the likes of Intel (the largest chip manufacturer in the world) were predicted by many who are knowledgeable about the computer industry. A Forbes article quoted a DARPA (Defense Advanced Research Programs Agency) senior who is very concerned about the demise of Moore’s law last year.
According to Colwell, who was Intel INTC +2.05%‘s chief chip architect from 1990 to 2001 and an Intel Fellow, there’s absolutely no doubt that Moore’s Law will eventually be repealed. Colwell believes that 2020 will be the earliest date for the Law’s demise.
“Let’s at least face the fact that [Moore’s Law] is an exponential, and there cannot be an exponential that doesn’t end,” he said. “You can’t have it.”
“That’s only seven years away,” he reminded his audience. “I’m thinking seven nanometers. You could talk me into 2022 – you might even be able to talk me into five nanometers, I don’t know, but you’re not going to talk me into one nanometer, you’re not going to be able to talk me into femtometers or whatever.”
Read the full article here.
In predicting the end of Moore’s Law in 2020, Colwell is assuming that there isn’t a delay in the investment in fabrication facilities. It appears that his assumption was wrong. This more recent (2014) extract from an article in the Wall Street Journal sees the situation as more dire.
“The long-debated shift to fabricating chips on larger silicon wafers is the latest cycle in a transition the industry has repeated every decade or so to reduce the cost of producing each chip.”
“To realize these savings, companies have to make massive upfront outlays for plants and equipment. The latest change could boost the cost of a single high-volume factory to as much as $10 billion from around $4 billion, industry executives say.”
“Amid questions about the future strength of chip demand and how development costs will be shared, some companies have been reining in their investments, raising fears the equipment needed to produce the new chips might be delayed for a year or more.”
Read the full article here.
A Classic Case of Trench Mentality
The end has been forecasted for years. Since around 2003, multi-core processors have been the solution to the demand for greater speed, without expending too much energy and creating high error rates. So why is software largely unable to leverage multi-core? Perhaps the reliable delivery by the hardware industry of a 2X speed up every 18 months has been too hypnotic for the software industry to break free from its grip. A more likely answer is, programming applications for multi-core parallel processing is very, very complex.
So do we just accept that the rate of processing speed increases will diminish or is there an approach that efficiently exceeds the 2X speedup result? To find the answer, it is analogous to climbing out of a trench that is very deep and very wide. This trench is so wide, that the walls are not visible and so deep that there is no direct light from the sun. One must climb out of the computing trench to experience pure light.
Life Outside of the Trench
At mindaptiv™, we are about to launch our new software platform, Essence™, that is truly remarkable. Essence™ is based on an approach that we call “Semantic Intelligence”. As stated (read it here) recently in a column by a prominent Futurist Speaker, Thomas Frey, “on a zero to ten scale for rating tectonic shifts on the Richter Scale of computing, Semantic Intelligence (SI) is drawing lines on parts of the chart that haven’t ever been written on before.”
Our approach to turning existing and future computing devices into SI Machines is nothing less than a paradigm shift in computing. We climb out of the computing trench and enable capabilities that are generally viewed as impossible with the small amount of resources involved.