Celebrating 50 Years of Moore's Law (IEEE Spectrum)

[Mortius]

As I wrote, it was a rough estimate. Your considerations above are valid. But do they compensate enough for the fact that the price per transistor to the consumer is about 6-7x higher than it should be if the cost reduction progress from Moore's law was passed on to the customer? I have a hard time seeing how that could be the case...

Intel R&D Expenses

Lets assume that it takes two years to develop both the process and the cpu.
Expenses in the lead up to Haswell: 20.169 Billion USD
Expenses in the lead up to Penryn Quads: 11.628 Billion USD

Research and Development expenses have approximately doubled, (assuming they haven't started supplying champagne, caviar and harlots for their engineers). Did the number of processors shipped during that time also double to maintain a similar cost per CPU?

Lets assume for the moment that they did. We have then demonstrated a significant cost (22% of gross sales in 2013) that doesn't scale with the improvements of production cost that Moore's law brings.

But lets suppose that Intel do as you said and reduce their prices by a factor of 6 (to 16%). Compare the two percentages. Intel would be paying more per chip for R&D than they would gain in revenue, even if you assume sales & marketing, customer support and even production costs were free.

You might argue that they would sell more chips than would otherwise have, but would there be enough to overcome the other costs that I mentioned?

Yes Intel make money. No, they do not gouge as badly as you suggest. If you want a company that operates on a revenue-neutral basis, try AMD.

 

[Fjodor2001]

Why make assumptions when you can quite easily read up on it and enlighten yourself?

Thanks! I was 80% sure of it when I wrote the post, but did not have time to check up on it to get 100% sure because I had stayed up too late that night anyway and wanted to get the post completed. :$

But luckily the assumption and point I was trying to make turned out to be correct in the end anyway...

 

[witeken]

2. The cost reduction in 1) has not been passed on to the consumer at the same rate.
3. Intel is making more profit per transistor than before?

Yes, because Intel isn't shipping more chips (~flat revenue). Volume stays roughly flat, so if prices would go down at the same rate, Intel's revenue would decline every node.

 

[witeken]

Graphic: Transistor Production Has Reached Astronomical Scales

The rate of growth has also been extraordinary. More transistors were made in 2014 than in all the years prior to 2011. Even the recent great recession had little effect. Transistor production in 2009—a year of deep recession for the semiconductor industry—was more than the cumulative total prior to 2007.

If my rough estimation and math are correct, the red and blue lines are equivalent to a doubling about every 18 months (in the last 10 years):

2^x = 100
x = 6.65
t = 10 years
10 / 6.65 = 1.5

This, people, is Moore's Law at work for you.

 

[Fjodor2001]

Too bad:

* It isn't being passed on to the consumer.
* It doesn't bring performance improvements like it used to any longer.

 

[Fjodor2001]

Yes, because Intel isn't shipping more chips (~flat revenue). Volume stays roughly flat, so if prices would go down at the same rate, Intel's revenue would decline every node.

True, but you're forgetting that the die size of the Intel chips could have remained the same if Intel wanted to (but instead it has decreased over time). If Intel would have kept the same price then, their revenue could have remained the same. And the price per transistor could be lower for both Intel and the consumer.

I.e. we could e.g. have seen 8 core CPUs instead of 4 cores CPUs by now, at the same price as when 4 core CPUs were introduced. Then Moore's law would have been passed on to the customer.

 

[Hulk]

I want robots and holodecks by now.

Hey come on, you did get tablets and they're thinner than the ones on Star Trek!