One small step.
One giant leap.
On July 20, 1969 Neil Armstrong set foot on the moon.
It was awesome.
It still is.
The picture above shows part of the instrument panel of the Apollo lunar command module. The person in the picture is interacting with the Apollo Guidance Computer (AGC) through its interface which can be seen in more detail in the picture on the right. The AGC was a rectangular unit that was roughly 6 inches thick, 1 foot wide, and 2 feet long. It weighed 70 pounds. The lunar lander also had its own AGC. One in the command module and one in the lander put a man on the moon. How much processing power did the AGC have?
In 1985, 16 years after Armstrong landed on the moon, Nintendo released the Nintendo Entertainment System (NES) in the United States. It introduced the world to Super Mario and was the first place many people met Zelda and Link. It was a sensation selling almost 62 million units worldwide before it was overtaken by more powerful gaming consoles. The whole unit was 3.5 inches high, 10 inches wide and 8 inches deep. It weighed approximately 2.75 pounds. How much processing power did it have?
The AGC had roughly the same processing power as 2 NESs. In 1985 four high-school juniors with an NES were playing games with the same amount of processing power that controlled the command module and the lunar lander that put Neil Armstrong on the moon in the year they were born. The four NESs combined weighed less than 12 pounds. The 2 AGCs in the lander and the command module had a combined weight of 140 pounds.
1985 also saw the release of the Cray-2 supercomputer. It was the fastest computer in the world until it was overtaken by the ETA 10G in 1990. The Cray-2 was housed in a cylinder that was 53 inches in diameter and 45 inches high. It weighed 5500 pounds. How much processing power did it have?
In 2010, 25 years after Cray introduced the Cray-2 and Nintendo launched the NES, Apple released the iPhone 4. Although it was a big deal among Apple fans in the US, by 2010 the iPhone was just another mobile phone to the world at large. At the end of 2010 Apple held only 4% of the mobile phone market and 15.7% (compared to Android’s 22.7% and Symbian’s 37.6%) of the smartphone market. The iPhone 4 was 4.54 inches long, 2.31 inches wide, and 0.37 of an inch thick. It weighed 4.8 ounces, a little more than a quarter pound. In terms of processing power the iPhone 4 was nothing special. How much power did it have?
The iPhone 4 had roughly the same processing power as the Cray-2. The Cray-2 took up 16 square feet of floor space and weighed 2.75 tons. The iPhone 4 weighed about a quarter of a pound and fit in your pocket. The Sony Ericsson Xperia X10 released in the same year as the iPhone 4 had 25% more processing power. The Samsung Galaxy S6 released in 2015 has the processing power of a bit more than 20 iPhone 4s/Cray-2s.
What made these enormous increases in processing power coupled with decreases in size possible? Moore’s Law.
“Moore’s Law” refers to Intel co-founder Gordon Moore’s prediction that the number of transistors on an integrated circuit will double approximately every two years. Moore had originally predicted a doubling every year in a paper published in 1965. He revised his prediction in a speech he gave in 1975. The two-year prediction has largely held true for the past 40 years and it is the primary reason why we now have so much computing power contained in such small devices.
Looking back to the original paper published in 1965, many tech-oriented websites published 50-year anniversary articles about Moore’s Law earlier this year. One of the best I saw was a fascinating infographic put together by Expert’s Exchange titled Processing Power Compared.
Expert’s Exchange compared a selection of supercomputers, smartphones, video game consoles and smart watches in terms of their processing power measured in FLOPS (FLoating-point Operations per Second). Processing Power Compared illustrates a number of these comparisons in addition to the AGC/NES and Cray-2/iPhone 4 cases mentioned above.
The infographic also contains a very interesting section that sequences devices based on processing power. The sequence starts with 1954’s 20K+ FLOP IBM 704, the first commercially produced computer capable of floating point calculations.
It ends with today’s fastest supercomputer, the roughly 34 quadrillion FLOP Tianhe-2 which is located at the National University of Defence Technology in Changsha China.
A number of interesting tidbits can be gleaned from the information presented in Processing Power Compared. For example, Apple’s iPhones are consistently outpowered by almost every comparable mobile phone on the market. At the time of this writing Samsung’s Galaxy S6 and Apple’s iPhone 6 are their respective company’s flagship smartphones. The Galaxy S6 has approximately 5 times the processing power measured in FLOPS as the iPhone 6.
Expert’s Exchange measured the processing power of game consoles in terms of their GPU (graphics processing unit) rather than their CPU (central processing unit) based on the idea that console performance depends more on GPU than CPU. In terms of GPU FLOPS, the Playstation 4 (PS4) is roughy 25% more powerful than the Xbox One. The PS4 is the most powerful device included in Processing Power Compared that isn’t a supercomputer. Comparing the PS4 to the world’s fastest supercomputer is like comparing a World Series champion to a Little League team. It takes 18,400 PS4s to equal the processing power of the Tianhe-2. That’s a lot of PS4s but to put that number into perspective, Sony has, on average, sold roughly 18,400 PS4s every 10.25 hours since the console launched in mid-November of 2013.
One last treat. Processing Power Compared includes this gif which captures how data storage capacities have grown while the hardware needed to store the data has shrunk over time. Moore’s Law in action right before your eyes.
The devices included in Processing Power Compared are only the tip of the iceberg. The power and miniaturization made possible as Moore’s Law has held up over decades has enabled most of the digital technologies that are everyday elements in many people’s lives. Health and fitness tech, digital books, streaming music, digital home security systems, automotive electronics, the internet of things, all of these things and more are common and ordinary because Moore’s Law has held true.
Imagine where we’ll be if Moore’s Law remains accurate for another 10 or 20 years. How long will it be before we have the processing power of a Tianhe-2 implanted in our bodies and hooked up to our central nervous systems?