Rumor: IBM to skip .09 and go direct to .06 process
MacOSXRumors has published a rumor "from very reliable sources" that IBM intends to go direct to .06 nm for the next PPC970 process - by the end of the year. Amazing implications if true! Faster, cheaper, lower power. 3GHz at least in the new year, I would imagine. And the PowerBooks! I know if a rumor is too good to be true then it probably isn't, but blimey...
Link
Link
Comments
Originally posted by Jupiter
MacOSXRumors has published a rumor "from very reliable sources" that IBM intends to go direct to .06 nm for the next PPC970 process - by the end of the year. Amazing implications if true! Faster, cheaper, lower power. 3GHz at least in the new year, I would imagine. And the PowerBooks! I know if a rumor is too good to be true then it probably isn't, but blimey...
Link
I seriously can't see how this is true, unless IBM's silicon jockeys have been keeping something up their sleeve.
However, in the interests of stuffing our Wintel friends, let's keep an open mind and hope for the best.
Surely a 0.06 process will make fast (eg. 2GHz+) G5 Powerbooks more likely than not, and would also tip the scales for the desktop part to exceed the 3GHz that is supposed to be its ceiling.
Also, seeing as you now have a heat budget to play with, can you build a 4-way Powermac.
If this is true, I'm going to keep this Tibook alive for another few years - which I can do thanks to the fact that a) I don't do anything that demanding and b) it's stuffed with 768MB of RAM - and wait and see what silver-skinned miracle is around in the middle of 2005.
I'll welcome it if they can but it seems like too large a leap.
Chipmakers sharpen 65-nanometer focus
By Reuters
August 7, 2003, 6:58 AM PT
Chartered and IBM, two of the world's four biggest contract makers of custom computer chips, together with Infineon, a major maker of memory chips, said in a joint statement that they were working to speed development of 65-nanometer chips.
IBM and Chartered production are among the first to be built on 65-nanometer circuits developed and produced in the new facility, called the Advanced Semiconductor Technology Center, or ASTC 300, which began operations last month.
The companies are also exploring extending the project to 45-nanometer-scale circuits, they said.
Full Story @ News.com.com.com
Google for ASTC 300
Here's hoping they weren't suckered into to this by some ne'er do well, but methinks they may have been fed some bogus information.
from MacOSXRumors 5/26/03:
I will use these domains for a rumor site, that will provide to Mac fans reliable and verified insider information about future releases of Mac OS X.
Yes I know, it looks like I was somehow inspired by our friends at macosrumors.com, but macosXr doesn't want to compete with this site in any way.
I'd also like to add that I will not post any content on the site that may hurt Apple even if it is very interesting information. I and people who help me to write the articles don't want to hurt Apple's interests in any way. We are all big Apple fans here.
from MacOSXRumors 3/29/03:
During the last two months, I've received a lot of mails of readers complaining that I'm not updating macosXrumors.com very often with new articles and cool new rumors. I'd like to explain you why. Actually there are several reasons:
- I'm doing my best to post information that is quite verified, and it sometimes takes time if you want to do that seriously.
- I'm trying to post information that hasn't already been posted elsewhere except for some important subjects or if I have more details than others.
- This site is quite specialized on software related rumors, well... you know... it's not MacRumors.com!!! And I don't think there is need to make a similar site because I believe our friends of MacRumors are just doing their job very well.
- I'm currently student and I don't always have enough time to do all the job I'm supposed to: verifying the reliability of some rumors and writing articles to report them...
Originally posted by FormatC2
Working to speed development of 65-nanometer chips.
....Which began operations last month.
MmMMmm... Full out production of .65 nm G5's?
This would be awesome.
Such a pickle to be in right now, isn't it (I spoke of this a week or two ago): seems we're right on that cusp of "it could go either way". I mean, it's REALLY tough to imagine dropping, say, $2500 bucks on a new 15" PowerBook (again, if/when they ever arrive) and not knowing how far along - if at all - Apple and IBM are on the development of a G5-based PowerBook.
I suppose there is always eBay to quickly unload a regretted purchase, but still...
Would be kinda nice to have an inkling of "okay, G5 PowerBooks: January or September of 2004?"
IBM will go to 90nm this year or next and to 65nm in 12 month at the earliest.
What exactly does a smaller process do? What does it mean? I really have no idea. I was talking to my geek friend about the G5 (he's a PC user but indifferent towards Macs) and he was interested that it's a 64 bit computer, but when he asked about laptops using the chip I said they would once IBM moves to a .09 micron process. He said, "Why would that be? Doesn't making it smaller heat it up, because the smaller surface area makes it harder to cool?" That kinda stumped me... see, I've heard people say that making it smaller will help with power and heat, but I don't know exactly how it works. My only explanation was that being smaller, the electrical impulses wouldn't have to move as fast to achieve the same speed, therefore reducing the power requirements.
So could someone explain this to me?
Originally posted by Luca Rescigno
What exactly does a smaller process do? What does it mean? I really have no idea. I was talking to my geek friend about the G5 (he's a PC user but indifferent towards Macs) and he was interested that it's a 64 bit computer, but when he asked about laptops using the chip I said they would once IBM moves to a .09 micron process. He said, "Why would that be? Doesn't making it smaller heat it up, because the smaller surface area makes it harder to cool?" That kinda stumped me... see, I've heard people say that making it smaller will help with power and heat, but I don't know exactly how it works. My only explanation was that being smaller, the electrical impulses wouldn't have to move as fast to achieve the same speed, therefore reducing the power requirements.
So could someone explain this to me?
This article on CNet covers your questions in a non-technical way. Would be interested to hear what the AI board experts have to say though. One thing's for sure, the smaller process enables smaller transistor gates which draw less power to move the electrons across. It's not a question of chip surface area as the chip will probably end up the same size, just with more transistors, such as a bigger L2, to fill the gap - as the chip needs to be a certain size to make the hundreds of connections to the motherboard.
Originally posted by Luca Rescigno
I thought I'd post a reply rather than creating a new thread for this question I had.
What exactly does a smaller process do? What does it mean? I really have no idea. I was talking to my geek friend about the G5 (he's a PC user but indifferent towards Macs) and he was interested that it's a 64 bit computer, but when he asked about laptops using the chip I said they would once IBM moves to a .09 micron process. He said, "Why would that be? Doesn't making it smaller heat it up, because the smaller surface area makes it harder to cool?" That kinda stumped me... see, I've heard people say that making it smaller will help with power and heat, but I don't know exactly how it works. My only explanation was that being smaller, the electrical impulses wouldn't have to move as fast to achieve the same speed, therefore reducing the power requirements.
So could someone explain this to me?
A smaller process means that the smallest dimensions of elements on the
chip become smaller. Generally, the smallest dimension is the physical
gate length of the transistor, other dimensions, such as interconnects are
much larger, but tend to be scaled with the minimum dimension.
As for the power use.
Simple answer:
Moving to a smaller process means that the transistors switch faster, which
in turn means that they draw power for a shorter time when they switch,
and hence the energy dissipated per switch is less. For a given frequency,
then, a faster transistor uses less power.
Slightly less simple answer:
There are two main sources of power dissipation in a modern CMOS chip,
static dissipation from leakage current, which is independent of the
switching speed, and dynamic dissipation from the transistors as they
switch.
Dynamic dissipation comes about as the transistor's gate capacitors are
charged up from the power supply, and then discharged to earth. Reducing
the size of the transistor reduces the capacitance, and hence the energy
dissipated each time it switches. In addition, as transistors become
smaller, the power supply voltage can be reduced, whilst keeping the same
electric field values within the transistor, meaning the power use goes
down even further. Added to this is the power required to charge up the
capacitance of the iterconnecting wires, though this is currently not
quite so important.
Static dissipation is caused by current flowing continuously through
transistors when they are not switching, largely leaking through the gate
capacitor into the body of the transistor. As the transistor becomes
smaller the insulation in the gate capacitor becomes thinner and the
leakage current goes up. Other leakage also goes up because of increasing
electric field intensities. Reducing the power voltage also seriously
reduces the leakage.
At the 90nm level, the static power (leakage )is becoming quite important
for many devices (especially at Intel, I hear), and will become even more
important going to even smaller processes.
Your friend is correct in that removing the same amount of heat from
increasingly smaller chips is becoming a bit of a problem as well.
michael
Originally posted by FormatC2
Not rumor, fact:
Full Story @ News.com.com.com
Google for ASTC 300
That article is about a joint R&D project that's started. At one point the article mentions that one of the goals is to figure out what machinery will be needed, which strongly implies that they don't have the machinery to make anything at that size yet. It doesn't mean that IBM is going to skip the 90nm node. Together with IBM's earlier announcements it strongly implies that IBM is gearing up for an imminent switch to 90nm as it does the research necessary for the next step down to 65nm.
By the tone of that article we'll get to 65nm in good time, but not within months, and not by jumping past 90nm.
i really hope this rumour is true,if anyone can do it ibm can,but im a little worried.
unless apple gets it act together in regards to the "year of the laptop" i see BIG trouble coming ahead.
lets face it,motorola has shafted apple over and over again.
need i remind those with short memories that motorola is suppose to be rolling out its .65um process soon now(alleged).
the proof is in the pudding.
apple also HAS to try to compete in the low-priced space that its currently ignoring.
this is the only way its gonna get market share.
im on a rant here.
and what about the so-called 64-bit operating system?
panther wont be 64-bit.
apple would have had to create an entirely new o.s that required old legacy apps to be recompiled.
apple is gonna need to do this eventually.
what apple needs right now is speed,speed,speed.
i read somewhere that ibm pushed the current 970 chip up to 4.2 ghz on a 0.13um process!
apple needs low power powerful chips right now,THEN they can kick motorola to the curb.
of a frequency for a .13um process.
Originally posted by geekmeet
unless apple gets it act together in regards to the "year of the laptop" i see BIG trouble coming ahead.
Once more, with feeling: The "Year of the Laptop" specifically referred to laptops eclipsing desktops in terms of units sold.
This year, laptops eclipsed desktops in terms of (Apple) units sold. PowerBooks and iBooks continue to be strong sellers. So the phrase, as defined, is correct.
The only people disappointed are people who extrapolated Steve's phrase beyond the given definition to include quad-POWER5 23" PowerBook HDs, or whatever.
The only people disappointed at the rumor that new PBs won't ship till October have forgotten that we knew all the way back in January from Motorola itself that the MPC7457 wouldn't ship until Q4 this year. Q4 starts in October.
right now people are clamouring for a powerbook update,it is way past due.
the significance of the ibm 0.06um rumour is that if true could allow apple to catch intel processors by the end of next year!
thats right.
if those rumours are true ibm is more than capable of producing desktop processors that clock at way more than 3 ghz.
but the real excitement could be the mythical "mojave" processor.
this could be a replacement processor for the ibooks.
i believe even with a die shrink to 0.09um the 970 will still run a little too hot for powerbooks and the battery performance of said product will be unacceptable.
remember,apples portable products achieved some of the best battery life times under motorola.
people want MORE battery life,not less.
the industry is trending towards smaller,faster,lighter.
when ibm comes out with a product(shipping) that will continue this trend,then i will be happy.
if ibm is indeed already moving towards 0.06um process.........i believe that wait is over.
And perhaps, by then, the iBook will be running something a little more OS X- and iApp-friendly than its current chip? Gobi, Mojave, whatever.
We've got the OS. We've got the iApps and all of Apple's other cool software. We've got things like AirPort, the iPod, iSight, Bluetooth, FireWire 800, the best hardware design on the planet.
Just capping all that off with blistering, no-bullcrap processors will just make for a damn near perfect experience (even better than it is now). IMO, that one - albeit important - area is the only place Apple is coming up a bit short on: the horsepower/processing end of things. And happily, that seems to be on an optimistic, positive road to being addressed.
Think about it: all we have now in the way of the whole "Mac experience", married to ridiculously-speced processors! It'll be beautiful, man.
I've always said that this is God's way of keeping us humble and in check. We have the best of everything else, so we have to suffer with the occasional lag in raw speed and power because we lucked out on all the other stuff I mention above (OS, hardware, etc.). Otherwise we'd be REALLY overbearing and obnoxious if we also ruled in the area of speed and sheer power.
Can you imagine the unbridled arrogance of the Mac community if we simply had it all right now? Eek!