This plot reveals the similar observation from the Moore's Law, which is telling us the number of transistors per square inch on integrated circuits doubles every 18 months. It's first proposed by Gordon Moore, one of the founders of Intel, in 1965. During 60s, the transistor density was an important indicator of the chip performance. However, things have changed a lot after 50 years. Nowadays Intel can manufacture transistors as small as 10nm, so there might only be little improvement left for them to do. The fact is that the shrinkage of transistors does come to a stagnation in this decade. (the horizontal axis ends in 2003...) That's might be a reason that parallel computing seems to be more important today.
ggm8
@yeq I agree with your point about us seemingly reaching a kind of plateau as far as the number of transistors on a single chip is concerned. I think when we look at the numbers for CPU processors we can see them underperforming Moore's Law in the recent years. However, it is interesting when looking at the speeds and/or processing powers of chips on mobile devices, where we actually see them outperforming Moore's Law in the few years after where the graph above ends.
Moore's Law
This plot reveals the similar observation from the Moore's Law, which is telling us the number of transistors per square inch on integrated circuits doubles every 18 months. It's first proposed by Gordon Moore, one of the founders of Intel, in 1965. During 60s, the transistor density was an important indicator of the chip performance. However, things have changed a lot after 50 years. Nowadays Intel can manufacture transistors as small as 10nm, so there might only be little improvement left for them to do. The fact is that the shrinkage of transistors does come to a stagnation in this decade. (the horizontal axis ends in 2003...) That's might be a reason that parallel computing seems to be more important today.
@yeq I agree with your point about us seemingly reaching a kind of plateau as far as the number of transistors on a single chip is concerned. I think when we look at the numbers for CPU processors we can see them underperforming Moore's Law in the recent years. However, it is interesting when looking at the speeds and/or processing powers of chips on mobile devices, where we actually see them outperforming Moore's Law in the few years after where the graph above ends.