Comment by jondubois | Hacker News Reader

Comment by jondubois | Hacker News Reader

Parent Full thread

One thing that I am really curious about is why CPU prices have seemingly stopped dropping? Many years ago, I distinctly remember that as clock speeds doubled (e.g. from 1GHz to 2GHz), the price per GHz would half (keeping prices of CPUs stable)... So why is it that prices of n-core CPUs seem to be growing linearly with n?

Is it because demand is higher for CPUs with low core-counts? I understand this argument in the consumer PC space, but what about in the datacenter?

Maybe companies which operate datacenters are too focused on horizontal scalability and stopped caring about vertical scalability? Surely this will change in the future? There are factors like power consumption and floor realestate to consider - From an electricity and realestate perspective, it is increasingly efficient to run your software on as few machines as possible (each machine having more CPU cores).

Am I right to believe that eventually companies will start writing/using software which leverages multiple CPU cores (when the cost benefits outweigh the hassle)?

Why is it not popular to scale both horizontally AND vertically?

I would be really interested in hearing ideas.

I started thinking about this after noticing that Amazon EC2 instance prices grew linearly as you doubled memory and CPU resources... I expected a sub-linear growth.

Since the 28nm node which happened around 2009, transistor prices didn't go down, and the cheapest chips use 28nm even today. So realistically the main economic interpretation of moore's law has stopped for some time .

NVidia and AMD haven't even utilized processes beyond 28 nm, at all (they are still in the pipeline).

Budget-conscious projects use 180 nm. [1] [2]

[1] http://info.scichina.com/sciFe/EN/10.1360/112010-977

[2] http://www.greenarraychips.com/home/about/

Source? According to Intel, cost per transistor is still falling exponentially: http://www.anandtech.com/show/8367/intels-14nm-technology-in...

source: http://electroiq.com/blog/2014/03/moores-law-has-stopped-at-...

Look for "cost metrics" table, "cost per million transistors".

As or intel: they claim they can make cheaper transistors, but yet their cpu's don't reflect that. Also they've offered manufacturing services for some time, but nobody taken , and that's pretty weird if they are so cheap beyond industry.

So many in the industry don't think it is true.

Cost per transistor is still dropping, but transistor count is skyrocketing as Intel adds cores. Their flagship Xeon CPU has 18 cores and something like 5.5 billion transistors.

New CPUs during the clock speed doubling era had moderate increases in transistor count to accommodate new features. Now that clock speed increases have stalled, adding on more stuff is the only way to differentiate new products.

For those reasons:

* Physics stopped CPUs from advancing any further.

* When CPUs used to advance they almost cost the same, that's why once they reached a new milestone they never looked back to produce slower CPUs.

* Adding a number next to "i" will make costumers pay more for pretty much the same CPU.

* There's only major CPUs manufactures and all CPUs they make are sold anyway so they start by mass production from day one.

They've actually started increasing now. Intel will raise the price of its highest end chip by 50% ($1000 to $1500 by adding 2 more cores), and it has effectively raised the price by multiples for Atom chips by calling them Celerons and Pentiums and removing the real Core-based Celerons and Pentiums from the market.

Because Intel has multiple markets cornered.

One other reason is that current sales pay for future development. Future development is becoming more and more and more expensive, so current prices have to go up (or at least stop decreasing) to pay for it. Since everyone else has to pay for those development costs if they want to stay relevant, no one else can undercut Intel's current-gen processors, either; they'll be left behind when the next process matures.

I work in Intel's Failure Analysis division, and we're expanding pretty aggressively. Same job, three times as many resources, much more money involved. It just keeps getting smaller.

What happened to amd?

Horrifyingly bad decisions with architecture - Bulldozer completely obliterated AMD's desktop market. Intel doesn't even look at AMD's opportunities when deciding on engineer salaries anymore.

They're trying to make a comeback with Zen, but one of the biggest issues with development is that you learn from the successes and failures of your previous product. If your previous product sucked, you're under an enormous amount of pressure to fix those problems before you even get started on your next generation. Otherwise, those problems will persist, and you'll get new problems as well.

They've stayed relevant in the GPU market, though.

I have a Bulldozer CPU in my main desktop machine, it works fine for what I use it for and I'm happy with what I paid for it. I wouldn't have got nearly as many cores in an Intel CPU at the same price point.

I am mainly using it to compile software though, which makes good use of multiple cores, probably closer to a "server" workload than a "desktop" one.

AMD can't compete on single thread performance and performance/watt. It's okay for a personal desktop and likely you'll save some bucks against Intel but for high-end desktops/workstations and servers they just can't compete at the moment.

You save a good amount of money though. Most of the high-end sub-$100 CPUs in the PassMark CPU benchmarks are AMD chips.

I just picked up the AMD Vishera -- 4.2 Ghz, easily overclockable to 4.5 with a proper cooling system, for $90.

For comparison, an Intel Core i3 4350 at 3.6 Ghz runs $205.

> They've stayed relevant in the GPU market, though.

And in APU (or "non-mobile SoC", if you will) market too: Xbox1 _and_ PS4 - that's a huge success by any measure.

These are not huge monetary successes, though.

Amazon deploys huge physical machines sells virtual machines, why would they not charge linearly? Doubling your compute isn't cheaper than selling someone else the same compute as you.

Reply on news.ycombinator.com