4 ms·
It's a benchmark for transistor density, nothing else. Of course a single company could grow faster than transistor density can scale. The only possible insight
by positr0n 9y ago
It's a benchmark for transistor density, nothing else. Of course a single company could grow faster than transistor density can scale. The only possible insight, if you can call it that, is that facebook owns more square centimeters of transistors now than they did in 2012.
- jxramos 9y agoThat's the original inspiration, but I thought people adopted the exponential rate from the original context as a bench mark for all sorts of stuff. https://en.wikipedia.org/wiki/Moore%27s_law#Other_formulations_and_similar_observations https://en.wikipedia.org/wiki/Moore%27s_law#Other_formulatio.... At best I can figure the problem with the auhtor's usage is that it's not relative to a per unit performance but rather a per enterprise performance, ie not a faster server but more of them over a unit of time. I'm not entirely convinced the restriction to a per unit basis is valid, but maybe that's a more meaningful definition than just arbitrary exponential growth.
- jxramos 9y agoI'm a bit mystified by all the downvotes, am I mistaken in understanding that Moore's law can be understood in a generalized context with respect to the rate not the quantity the rate was originally measured from? Basically he observed a particular rate constant or exponential growth profile. The quoted wiki page states: "Several measures of digital technology are improving at exponential rates related to Moore's law, including the size, cost, density, and speed of components. Moore wrote only about the density of components, "a component being a transistor, resistor, diode or capacitor,"[102] at minimum cost."". So essentially when comparing the growth of one quantity vs another it's not the quantities that are being compared but the rates.