back to article Intel pops $4.1bn to save Moore's Law from repeal

Intel will invest $4.1bn in ASML, widely regarded as the world's leading semiconductor-lithography equipment manufacturer, in an effort to keep Moore's Law alive and kicking for the foreseeable future. The deal includes $1bn in cash to support ASML's research and development efforts on extreme ultraviolet lithography (EUV), and …

COMMENTS

This topic is closed for new posts.
Silver badge
Facepalm

once bitten

The tool manufacturers learned after taking a bath on the 300mm upgrade that its not milk and honey to upgrade wafer sizes (granted a lot of other things besides wafer size also changed). After taking so many years to see the ROI I correctly guessed this time the IC makers were going to have to front a lot of the money for the transition to 450mm.

0
0
Silver badge

Re: once bitten

That is after it took so many years for the tool makers to see the ROI on the upgrade to 300mm the IC makers would have to front cash for 450mm. The IC makers love the upgrades as they see their ROI much quicker than the tool makers.

0
0

Re: once bitten

It's unclear which is better for the tool suppliers in the long run.

If the suppliers foot the development bill, many chip makers can remain afford to buy the process equipments and stay in business. If the chipmakers are forced to pay for R&D up front, only the biggiest of the chipmakers can afford the equipment. Also they will likely ask for exclusivity, and small players will have to fold.

You think Samsung and Intel have strong leverage over the equipment suppliers today with 300mm technology, wait until these 2 have funded development of 450mm technology...

0
0
Silver badge

Re: once bitten

Good points. The thing is the the industry is already consolidating as more and more of the smaller players disappear (Infineon, Elpida, etc). More proof eventually OCP will eventually be contracted to provide our police service (Robocop future for joo).

0
0
Anonymous Coward

Re: OCP

Monopolies are legal in Japan. Isn't is about 5 companies that own everything there? Mitsubishi is the country's bank. That has been working out just fine, hasn't it? What will AMD do after this? I would think the only way this can be allowed is to AT LEAST give AMD access to the same tools. No?

0
1

This post has been deleted by its author

Silver badge

Re: OCP

Worst description of zaibatsu ever. Besides which, zaibatsu don't even exist now, at most they are keiretsu.

0
1
WTF?

Re: OCP

Keiretsu, meaning "series" or "subsidiary" - A subsidiary of the still existing Mitsubishi conglomerate is still part of a Monopoly IMO. Did I miss something, or did you forget that the leaders of all these subsidiaries still regularly meet up and talk shop in ways that would be considered questionable, if not illegal in the US?

http://en.wikipedia.org/wiki/Mitsubishi

http://en.wikipedia.org/wiki/Keiretsu

0
0
WTF?

Smaller still means hotter right ?

If smaller means hotter, and we can already get 4x >1Ghz CPUs into a *mobile phone* why do we need to pointless keep up with The Law ? What's the real benefit we'll see ?

0
0

Re: Smaller still means hotter right ?

Smaller does not mean hotter?

One of the advantages to buying smaller nm CPU is less current leakage, so less power is required, so less heat is generated.

1
2
Silver badge

Re: Smaller still means hotter right ?

What's the real benefit we'll see ?

There are plenty of real world problem that happily lap up any excess computing power we have available (curing diseases with protein folding, energy exploration, etc). You are correct though that the benefit to consumers of more computer power is only somewhat of use as evidenced by the fact my 2006 Mac Pro 1,1 still plays all the games I want and is still a perfectly fine desktop. Still less and less power use of the same computing power as mention is darn useful for mobile, etc.

0
0
Alert

Re: Smaller still means hotter right ?

It's the opposite. The smaller the process, the highest the leakage. But then, the lowest the operating voltage. So by going to a smaller process, you reduce the dynamic power consumption (due to lower operating voltage) per transistor, at the price of increased leakage while the transistor is powered. But the leakage part can be controlled by power gating (= cutting the power) parts of the chips when they're not needed.

1
0

Re: Smaller still means hotter right ?

Thanks for correcting me.

every days a school day eh :)

0
0
Silver badge
Pint

Coin shrinking

Warm the wafers in an oven to just the right temperature and then zap them with a huge pulsed magnetic field (à la coin shrinking). Shrink the wafer radius by half. Don't tell anyone how you did it. After the competition figure out a method to replicate the same size features, then steal their more practical method.

0
0
Trollface

Re: Coin shrinking

Is blowing all the etched circuitry with the rather effective EM pulse part of the plan? Or maybe losing all the deposited traces and components as the die underneath shrinks, but they don't?

I gotta say, I did wonder at first, but I do belive you've used the wrong icon. The joke one is the on to the left... ;)

0
0
Silver badge
Boffin

450mm is immense

I'm still trying to wrap my imagination around a single nearly-perfect monocrystalline silicon ingot almost half a meter thick and maybe two meters long - to be made into processors and memory and Flash worth many billions of dollars.

The long-term answer is still in dimension Z of course, but the more they get out of X and Y the better too.

If near-zero G environments turn out to be ideal for growing ever-larger crystals, and even better at very high resolution lithography then maybe Planetary Resources is on to something. Asteroids have a lot of silicon.

1
0
Boffin

Moore's [not a] Law [at all]

The "law" has changed since he wrote about it, adusting it to meet the facts, hardly a law! originally it was doubling every year, then it was doubling very two years, besides, it's a nonsense anyway, the original prediction was "the number of components per integrated circuit for minimum cost" whereas it's often been taken to mean the maximum we can cram in a chip, when some of these multi core behmouth chips come out they cost well over $1000 a unit, hardly a minimum cost!

Moore could see (historically) that there was an obvious exponential growth, he made a broadly accurate prediction that growth would continue to be exponential (he got the factor wrong, so he changed it), and of course it's still wrong, for example, in 1975 when he changed his "prediction" to doubling every two years, the 6502 had about 3.5K transistors, which means by 2006 it should have been 40M, but actually it was 1.7Bn transistors (itanium 2).

Don't get me wrong, I'm not criticising Moore, it's not his fault that he's being held up as some kind of "uncannily accurate"magical fortune teller, it's all a bit "Life of Brian", Intel (et al) like to perpetuate such things, if you want a prediction that is genuinely looking at the future then in 1950 Alan Turning said computers would have at least 128Mb of memory by the year 2000, which, considering computer programs were a novel concept, and it was the physical configuration of vacuum tubes that was generally the "program" he wasn't just extrapolating history.

0
0
Anonymous Coward

Just one question- why was El Reg's front page link to this story illustrated with a photo of a Shreddie?

http://regmedia.co.uk/2009/12/02/intel_scc_teaser.jpg

Are these new 450nm wafer chips destined for the lucrative breakfast cereal market?

1
0

This is a yield thing, not Moore's

Unless we are making incredibly huge chips, the wafer size won't have much impact on the density of the resultant IC. This is yield .. and one suspects it would be difficult in the early years to get 450mm wafers not to have so many defects it was counter-productive to yield.

the real interesting figure for speed, power consumption and transistor density is the 14nm mentioned. IF that is a problem, then we will enjoy a gradual slowing of the exponential ramp we have been on. Bigger wafers will eventually mean more yield, which may result in cheaper product but overall the effect of decreasing line widths was much more dramatic on density and costs. Without that we have some paradigms to shift.

0
0
This topic is closed for new posts.

Forums