IBM's Push for 65-nm Chip Development

[Snowy_River]

Re: Communism

Originally posted by jaedreth
Either way, no, it wouldn't have happened then either.

Then why bring up the distinction?

Communism ...

Democracy ...

Uh... you're describing two different types of systems. 'Pure' communism (aka Marxism) is an economic system, whereas democracy (or republicanism) is a governmental system. Comparing them is like comparing apples and oranges. Soviet Communism was a governmental system that tried to also implement a version of Marxism. The US is a representation of republicanism (not referring to the political party) implementing capitalism with some socialist touches. (Personally I'd like to see more socialism incorporated into our economic system, but that's just my opinion...)

I'm merely describing how economic factors as well as social tend to hamper or dampen technological progress deployment. Not development, but deployment.

And my point was that if it weren't dampened, i.e. if all tech were pushed onto the market as soon as it could, the economy would go bankrupt, and we'd all starve to death...

 

[Phil Of Mac]

Originally posted by DrGonzo
What do you mean "something entirely different"? Different from the PPC line of cpus?

Different than silicon CPU's. Optical.

Originally posted by jaedreth
Do you have any clue how much research and technology is based on science fiction? In fact, that's a favorite game among scientists, they pick their favorite scifi and they work on trying to make it reality...

"I'm working on that." --Stephen Hawking on the set of Star Trek: The Next Generation as he passed the warp core

Originally posted by jaedreth
Actually, this is kinda counter-intuitive.

Think of a copper wire like a garden hose.

You're treating electrons as if they were particles. They aren't always

(Just kidding. I didn't want to get into that, really.)

Oh, and capitalism doesn't deal with scarcity or abundance, it causes abundance. Economics is rather off-topic here, though, so I'll try to explain briefly.

First off, what we have in the U.S. isn't capitalism. The closest thing to capitalism is Hong Kong. By "capitalism", I mean the economic system resulting from the lack of government involvement (except for preventing fraud or outright violence). The necessity of the existence of government makes true capitalism very difficult to attain.

Capitalism, makes people as productive as possible by tying their personal productivity into their self-interest. Want a new house? Gotta make enough money to earn it.

Communism, on the other hand, does the opposite. Under communism, you "produce according to your ability and take according to your need". Now it's in your self-interest to have more need and less ability.

In other words, self-interest causes less productivity (and more consumption) under communism, but more productivity under capitalism. Capitalism, in tying consumption to production, makes us produce more, if we want to consume more.

The communists try to make communism work by repealing self-interest. But you know what? It doesn't work. Self-interest, properly understood and applied, is a perfectly legitimate part of human nature. Under capitalism, my self-interest results in benefits for others.

That is, of course, a simplified explanation. Economists spend all their lives arguing about the details and exceptions. I don't want to do that, which is why I'm not an econ major.

 

[themadchemist]

Originally posted by Tequila Grandma
What could this lead to? PowerBook G5s maybe?

First of all, if you notice, the article cites a multi-year agreement. The products of this research will not be seen for quite a while.

Moreover, 65nm and 45nm chips won't yield G5s. This is a size we have NEVER seen before, and I would assume that that would be your garden variety G7s and G8s (assuming the G6 is 90nm).

It may be the case, though, that with such a small process, they could tone down the frequency enough to resolve heat issues while still having a very high frequency, by comparison to anything Motorola could dish out.

 

[idea_hamster]

Re: Chip basics

Originally posted by jaedreth

If you had a hose with half the diameter, then you could turn the faucet down half as much to get the same flow.
...Make sense?

Ahh...so it seems to have more to do with having enough current to complete the circuit (i.e., overcome the natural resistance of the conductor) rather than actually "operating" anything (like a light bulb filament).

I was, in fact, thinking in terms of the water and hose originally, but in my mental experiment I was thinking that we needed a certain volume of water -- so that if we halve the diameter of the hose, we would have to actually raise the rate of flow to keep out-flow equal.

After your explanation, it occurs to me that registers are either on or off -- there's no fretting about how dim something is. (Leave it to a designer to worry about alpha channels when there aren't any!) So if the register can only be on or off, then the circuit just has to be completed, not "powered" in any way.

There should be an emoticon for "Ooooh! -- Now I get it!"

Thanks much -- this was great! (I feel so...technical!)

 

[jaedreth]

Subatomic particles

Yeah, subatomic particles technically only have tendencies towards existence. *heh*

Pwave? Warticle? Whatever.

Was just thinking...

Steve said G5'd hit 3GHz by December. Well, not on camera.

Power5 won't hit 3GHz, but Power5+ will.

And Power5 isn't coming all that soon. It might be this time next year or later before Power5+ hits.

By that time, Intel will have exceeded 3GHz by a good margin.

However the throughput on Power5 and its derivative will be far more awesome.

So does that mean we won't have 3GHz by this time next year? (Steve's actual on camera words, but back at wwdc)

Does this mean that the Power5 derivative won't be G6? Since G5 was stated to reach 3GHz.

Also, Apple's use of G5 is the entire architecture, not just the ibm chip.

It looks like Apple is not going to make a correlation between G5 and 970/Power4 but G5 and PowerX derivatives itself. We might not see G6 until the architecture changes so much (in the form of a slow evolution) to the actual Power chips. Just theorizing. Then IBM wouldn't have to produce two sets of chips. And Apple once migrated would be using very powerful chips indeed.

Just my opinion. I know many hold tight to the idea "IBM's 970/Power4 is Apple's G5, thus IBM's 980/Power5 *has* to be Apple's G6" and there has been absolutely no proof or even circumstantial evidence for such a claim. Those who feel this is the truth, end of story, it's their conviction. It's not mind. I'm not intending to start another flame war, I'm simply expressing my own opinion. Do names matter though? Not really. We're still gonna get the awesome IBM chips by whatever name.

Jaedreth

 

[Phil Of Mac]

Re: Re: Communism

Originally posted by Snowy_River
Uh... you're describing two different types of systems. 'Pure' communism (aka Marxism) is an economic system, whereas democracy (or republicanism) is a governmental system. Comparing them is like comparing apples and oranges. Soviet Communism was a governmental system that tried to also implement a version of Marxism. The US is a representation of republicanism (not referring to the political party) implementing capitalism with some socialist touches.

Actually, you're missing the distinction between communism and capitalism.

Marxist communism is also a political system because it is unnatural, it must be implemented and forced through government, specifically "a dictatorship of the proletariat". Capitalism, on the other hand, is not implemented, it is a "spontaneous order" that arises naturally from the lack of government interference (other than preventing fraud and violence).

The US system closely resembles capitalism. The regular occurrence of entrepeneurship and small businesses that are common in capitalism are also common in America. However, there are a number of ways that America's economy is not capitalist due to government interference.

1. Money. The US uses "fiat money", i.e. money without any actual backing other than the government itself. A central bank under government control produces this money at rates that it decides. In many countries there have been instances where money has been made very easily availible, causing immediate economic growth followed by an inflationary hangover. There have also been instances where the central bank suddenly cuts off the money supply, causing deflation. If inflation is a hangover, then deflation is withdrawal. The result is often a total economic collapse. Many economists believe this is what caused the Great Depression.
2. Taxes. Fair taxes are theoretically impossible and in practice, have never occurred. Taxation overall redistributes wealth in directions it would not normally have gone. Some people consider this a good thing.
3. Wage and price monkeying. Minimum and maximum wages, price supports and controls, etc. Wage controls can cause unemployment while price controls can cause shortages. Price controls can also in some causes cause gigantic surpluses.
4. Subsidies, etc.

There are more things than this, but the US is, in short, a combination of capitalism (there's a lack of control in some places, no maximum wages, relative ease in starting businesses), corporatism (state-established corporations and inter-involvement, subsidies and favors to some businesses), fascism (regulation), social democracy (welfare in its various forms), outright nationalized industry (government ownership of utilities, postal service, Amtrak, etc.), and economic protectionism (tariffs to protect domestic interests, etc.)

I'm not talking about whether government involvement is good or not, or when it's good or not, but it's good to understand how these things work.

 

[Phil Of Mac]

Re: Subatomic particles

Originally posted by jaedreth
Yeah, subatomic particles technically only have tendencies towards existence. *heh*

Well, as far as we can tell. I think they're ducking in and out dimensions we can't see

Originally posted by jaedreth
Steve said G5'd hit 3GHz by December. Well, not on camera.

Power5 won't hit 3GHz, but Power5+ will.

I'm guessing the 9xx is faster in clockspeed maybe? Power needs more reliability, and reliability is always inversely proportional to clockspeed.

 

[jaedreth]

Reliability is inversely proportional to clock speed

You make much sense, oh wise one.

So, NYAH! all you PC users with your 3.0GHz running around in your antes... You don't frighten me you tiny brained wipers of other peoples bottoms...

*ahem*

I mean, Nyah! Your unstable pieces of...

Nevermind...

Jaedreth

 

[Phil Of Mac]

What I mean by "stability is inversely proportional to clockspeed" is that, on any given piece of silicon, if you clock it higher it's less stable. I could overclock my processor from 400 MHz to, say, 600 MHz, but it would be very unstable, so I don't do it.

On a PC, stability isn't as essential as with a POWER-based server or workstation. So any given piece of silicon could be certified by IBM as 3 GHz within the stability tolerance of a PC, but if that same core was going into a POWER server, the higher tolerance means you can only clock it at 2 GHz or something.

 

[daveL]

Phil Of Mac

It's all good. Nice posts.