And it was said that they shall sloweth...

[zidane1strife]

And the model may even thrive if the industry opts to abandon Moore's Law.

Only Intel can afford a 300-mm fab, though that hasn't stopped the foundry industry from building such plants, he said.

In the last 10 years, the cost of building a fab has increased sevenfold while the semiconductor industry's revenues have grown fivefold, according to Goldman Sachs. Building a 300-mm, 90-nanometer fab is expected to cost a whopping $4 billion, and each plant must generate $6 billion to $9 billion in sales per year to get the required return on investment (ROI), foundry executives said. No pure-play foundry provider today realizes that much revenue. The world's largest foundry, TSMC, reported $4.66 billion in sales last year.

"The cost of a new fab is absolutely enormous," said Edward Ross, president of TSMC North America. Ross, who cited $6 billion as an acceptable annual sales figure for a 300-mm, 90-nm fab, noted that only five companies in the chip industry have such sales now, and none are pure-play foundries.

http://www.eetimes.com/semi/news/OEG20030317S0012

It seems... like this that was said is... but innevitable to occur...

EDIT

"Moore's Law is not driving the industry anymore," he said. "It's economics."

 

[London Geezer]

And the model may even thrive if the industry opts to abandon Moore's Law.

Only Intel can afford a 300-mm fab, though that hasn't stopped the foundry industry from building such plants, he said.

In the last 10 years, the cost of building a fab has increased sevenfold while the semiconductor industry's revenues have grown fivefold, according to Goldman Sachs. Building a 300-mm, 90-nanometer fab is expected to cost a whopping $4 billion, and each plant must generate $6 billion to $9 billion in sales per year to get the required return on investment (ROI), foundry executives said. No pure-play foundry provider today realizes that much revenue. The world's largest foundry, TSMC, reported $4.66 billion in sales last year.

"The cost of a new fab is absolutely enormous," said Edward Ross, president of TSMC North America. Ross, who cited $6 billion as an acceptable annual sales figure for a 300-mm, 90-nm fab, noted that only five companies in the chip industry have such sales now, and none are pure-play foundries.

http://www.eetimes.com/semi/news/OEG20030317S0012

It seems... like this that was said is... but innevitable to occur...

is that supposed to be a good thing or a bad thing? and for who?.... :? am i missing something?

 

[zidane1strife]

Well if it occurs, it will BE VERY GOOD for me, I'm planning on building a PC over the next couple of years... If the speed sloweth it will make my pc remain decent for quite some time.

EDIT:

And yes there's an industry with ever growing profits that won't have to be slowed down by this.

 

[London Geezer]

Well if it occurs, it will BE VERY GOOD for me, I'm planning on making a PC for the next couple of years... If the speed sloweth it will make my pc remain decent for quite some time.

EDIT:

And yes there's an industry with ever growing profits that won't have to be slowed down by this.

well i stll don't get it... i thought it would be a bad thing... :?

 

[Magnum PI]

i thing that the problem is also with the demand.

the performance of any low-entry desktop model is sufficient for needs of most users, and it will remain so for some time.

if you want to do some internet, wordprocessor, some gaming the slowest athlon XP is enough..

there is no incentive to upgrade, no mainstream killer-app that would need a faster processor.
and even worse, people have not much money in this period of bad economics.

low demand combined with strong competition b/w AMD and intel makes for bad ASP, low margins, and less money to reinvest.

 

[London Geezer]

i thing that the problem is also with the demand.

the performance of any low-entry desktop model is sufficient for needs of most users, and it will remain so for some time.

if you want to do some internet, wordprocessor, some gaming the slowest athlon XP is enough..

there is no incentive to upgrade, no mainstream killer-app that would need a faster processor.
and even worse, people have not much money in this period of bad economics.

low demand combined with strong competition b/w AMD and intel makes for bad ASP, low margins, and less money to reinvest.

ON a side note, what is the size limit for silicon again? i mean, when will they just not be able to go any smaller in transistor size? and what will we do then?

 

[zidane1strife]

Well, I've heard it gets far tougher below 50nm, and clearly they must stop somewhere... Ibm has shown 6nm transistors(not sure if in silicon, though), those or some sligthly smaller will probably be the limit.

After that molecular computing will likely take place, and someday in the very very distant future... computers with infinite bandwith, processing speed, and memory will arise, using the knowledge... the true knowledge...

 

[London Geezer]

Well, I've heard it gets far tougher below 50nm, and clearly they must stop somewhere... Ibm has shown 6nm transistors(not sure if in silicon, though), those or some sligthly smaller will probably be the limit.

After that molecular computing will likely take place, and someday in the very very distant future... computers with infinite bandwith, processing speed, and memory will arise, using the knowledge... the true knowledge...

:?

whats with the *knowledge* thing...
ok i'll stop now.
this is a serious matter.

 

[Thowllly]

After that molecular computing will likely take place, and someday in the very very distant future... computers with infinite bandwith, processing speed, and memory will arise, using the knowledge... the true knowledge...

There is an upper limit to computing speed. Max megahertz is limited by the Planck time (10^-43 seconds), and the maximum amount of data that can be calculated in that time is limited by Heisenberg uncertainty principle.

Or something like that

If we reach that limit the only way to speed up computers will be to make them bigger, but we are again “limitedâ€￾ by the amount of mass/energy available in the universe.

 

[BoddoZerg]

Don't worry, the whole Universe is already a quantum computer. God and Satan are probably playing "Command and Conquer Generals" on it right now.

 

[Sonic]

Nanotechnology! That's probably a pipedream today though.

 

[zidane1strife]

After that molecular computing will likely take place, and someday in the very very distant future... computers with infinite bandwith, processing speed, and memory will arise, using the knowledge... the true knowledge...

There is an upper limit to computing speed. Max megahertz is limited by the Planck time (10^-43 seconds), and the maximum amount of data that can be calculated in that time is limited by Heisenberg uncertainty principle.

Or something like that

If we reach that limit the only way to speed up computers will be to make them bigger, but we are again “limitedâ€￾ by the amount of mass/energy available in the universe.

Who said anything about this universe, or its laws...

 

[London Geezer]

After that molecular computing will likely take place, and someday in the very very distant future... computers with infinite bandwith, processing speed, and memory will arise, using the knowledge... the true knowledge...

There is an upper limit to computing speed. Max megahertz is limited by the Planck time (10^-43 seconds), and the maximum amount of data that can be calculated in that time is limited by Heisenberg uncertainty principle.

Or something like that

If we reach that limit the only way to speed up computers will be to make them bigger, but we are again “limited” by the amount of mass/energy available in the universe.

Who said anything about this universe, or its laws...

shall we really go into this discussion right here?

we still know nothing about how this universe works.... (let alone knowing how other universes work.....)

thinking about the universe as a huge computer is in itself flawed. a computer was invented as a tool for humans to use at their discretion to facilitate their lives. we cannot use the universe to run programs on. it doesn't make sense. still, looking at the bigger picture, you could say that the universe is a computer that someone else is using right now. leave God or Satan alone for the moment... of course technology will progress and one day (if we are still alive, as a species i mean) we will be able to create a new universe in a parallel dimension. the thing is, that must have happened to our universe as well, so who's playing with our universe...
i know these theories are a bit extreme, but they are real theories from very acclaimed cosmologists. some of these theories are also being taken VERY seriously by the scientific community.
like the String theory, which explains the universe as having 10 dimensions, our 4 normal ones and the rest *squashed* to Planck size connected to each other by a series of *strings*...
but why are we talking about this in the console forum?

 

[zidane1strife]

You're right we've deviated from the topic.... That is Intel, Ibm and the big boys will continue foward at speeds beyond moore, while most of the rest will sloweth... those with the most advanced technology will be in the position to do something interesting...

 

[Panajev2001a]

this is bad news for fabless chip manufacturers like ATI, nVIDIA and IMGtech...

As far as next-generation consoles, whoever does the GPU for the Xbox 2 better be partered with a guy like Intel as the Sony, IBM and Toshiba alliance and Nintendo-NEC ( not as good as Sony+IBM+Toshiba, but NEC is still a very good semiconductor company ) will leave behind guys like TMSC...

Sony, IBM and Toshiba are planning 65nm ( the Cell design seems to be targeted for 65 nm, wether the first PS3 prototypes will use large and expensive 90 nm chips it is not known ) for PS3 ( launched probably in 2005, but the process should available earlier ).. and 45-50 nm after that...

TMSC is not ready to do very large scale 90 nm chips and by the time intel and Sony, IBM and Toshiba get to the 65 nm node I do not think TMSC and UMC will be close...

Intel and Toshiba currently hold at least a 6 months advantage... I do not know if they will slow down as TMSC slows down ( maintaining the 6+ Months advantage ), but I do not think they will slow down as much ( if they even do slow down considerably... few years ago Sony wanted to use 100 nm for PS3's new CPU, their roadmap was updated and it wasn't slowed down, but instead accelerated to include 50 nm in the picture and later on we had the announcement of the completed 65 nm process with Toshiba and the research for sub 50 nm... )...

 

[Lazy8s]

zidane1strife:

And yes there's an industry with ever growing profits that won't have to be slowed down by this.

All the more reason for powerful PC chip makers and companies to seek a big piece of the console industry pie. They need a strong, growing market with enough assured demand to make sense of the investment in these new fabs.

 

[overclocked]

Can they pull off a 65nm design in early 2005(?), probably in late 2005/early 2006. I think it´s the most "realistic" that it is then we vill see PS3 IMO.

It´s IBM/Toshiba/Sony so there´s plenty of talent to suprise us i guess.

 

[jvd]

Well they can allways move to multi chip configs. Like the old rampage from 3dfx. INstead of having the tnl done on the chip make a new chip to work with it doing just tnl stuff. With amd and intel they can go to multi core chips. Dual systems. I doubt things will slow down all the need to do is get a little creative.

 

[overclocked]

Well they can allways move to multi chip configs. Like the old rampage from 3dfx.

Let it rest in peace

 

[jvd]

Well they can allways move to multi chip configs. Like the old rampage from 3dfx.

Let it rest in peace

Its resting in peace. It was sad that it never came to fully be. It was a great idea at the time and that should never be taken away from rampage and sage.