Intel's 45nm Chip due in 2007

SiliconAddict said:
Not really. AMD is having a hard time with their 45nm process. Hell They aren't even shipping 65nm chips in bulk yet. Intel has remained in the lead when it comes to shipping processores at smaller processes.

AMD is usually 9 months behind Intel in terms of process shrinks, but ahead in process features.

All in all they're roughly equivalent. Sure each one pulls ahead now and then, but it isn't like either one is ever that far behind.

People in this thread keep on mentioning the awful 90nm migration, but AMD had a relatively great 90nm, because of the more advanced process that included things like SOI. AMD's processors at 90nm cut power consumption dramatically, and allowed a small speed boost. Intel's ... didn't.

It's good that Intel has verified 45nm manufacturing with simple SRAM chips. I recall them having done the same with 65nm a couple of years ago. Don't expect to see 45nm Intel processors on the market until early 2008 at the earliest.
 
128-bit per se wouldn't be any faster than 32-bit or 64-bit

shyataroo said:
why not just make the chips bigger (the same size transistors just have more of them) or go upto 128 bit computing?
There would be an exceedingly small benefit to 128-bit computing for most application....

64-bit addressing supports up to 17.2 billion GiB of RAM or virtual memory. That's absurdly large for the practical future (imagine a computer with Billions of DIMM slots)...

Some people would like to see hardware 128-bit floating point and integer formats - but not many. You wouldn't see any improvement for most applications.

Note that SSE/2/3 have a few 128-bit operations today. It may make sense, if 128-bit arithmetic (as opposed to 128-bit addressing) is required, to extend SSE again to provide full support for 128-bit floating point and integer formats. (Since the SSE registers are already 128-bits, you avoid compatibility problems that extended the standard registers to 128 would cause.)
 
AidenShaw said:
There would be an exceedingly small benefit to 128-bit computing for most application....

64-bit addressing supports up to 17.2 billion GiB of RAM or virtual memory. That's absurdly large for the practical future (imagine a computer with Billions of DIMM slots)...

Some people would like to see hardware 128-bit floating point and integer formats - but not many. You wouldn't see any improvement for most applications.

Note that SSE/2/3 have a few 128-bit operations today. It may make sense, if 128-bit arithmetic (as opposed to 128-bit addressing) is required, to extend SSE again to provide full support for 128-bit floating point and integer formats. (Since the SSE registers are already 128-bits, you avoid compatibility problems that extended the standard registers to 128 would cause.)

Having been in the industry for more years than I care to count, I now have a wall covered in 'ticks for each time I have heard someone say "That's absurdly large amount of memory". I happen to agree, it sounds absurd now. The idea that I would have 4Gb of memory in my home machine would have seemed just as absurd to me back when I was trying to make things fit into 1K. Scroll forward a few years, and I frequently p***sed that away on a look-up table.

Anyhow, I agree for a CPU 128bit seems unwarranted right now, although there are always arguments for wider buses. Where these things become important are in Vector processing (such as graphics and physics) where the attendant transformations each introduce error and the additional precision afforded by "bigger words" is of benefit for kinematic or texture transformations.

Personally, I'm not expecting a change beyond 64bit for a general purpose CPU (i.e. who knows what the cell might do, but it's really got a narrower target) in the near future.

That said, I've been wrong before.
 
bugfaceuk said:
Could not agree more. For me the advantage of the switch to Intel is that it doesn't even need to stay Intel, AMD is always there as an option. In fact without AMD I'm not sure we would see Intel's focus shifting from Ghz to design efficiency. IMHO AMD shamed intel into doing something better.

AMD is an option, but it would really be a last resort used only if Intel really drops the ball big time. The reason is that Intel is not only making the processor that the Mac's are using but also the entire chip set on the motherboard and I think I heard that they might even be making the motherboard altogether. I also heard that they might use Intel tech in future iPods too. The Intel switch is not just to their CPU, it's a switch to a single vendor for their entire technology platform.
 
That's very exciting news. Hopefully with the 45nm process we can get some pretty small machines that perform at professional level speeds.
 
bugfaceuk said:
Having been in the industry for more years than I care to count, I now have a wall covered in 'ticks for each time I have heard someone say "That's absurdly large amount of memory". I happen to agree, it sounds absurd now.
That's why I said "practical future" :)

To use up 64-bits with RAM you'd need a computer with about 4 billion DIMM slots, you'd need 4 trillion dollars (about the annual GDP of Japan), and over 100 Gigawatts of power.


bugfaceuk said:
The idea that I would have 4Gb of memory in my home machine would have seemed just as absurd to me back when I was trying to make things fit into 1K.
Half a gig isn't that much, some home systems today have 2, 4 or even more. ("b" is bits, "B" is bytes)
 
Good to hear these are comming.....

But since we never got the G5 PB's.....what happend to the low power G5 chips....new Thinkpads :p :confused:
 
I'm so mad at Apple and Intel :mad: - I haven't recieved my MacBook Pro yet, and already it is obsolete!!!

Just kidding, I couldn't resist! :D
 
maya said:
To actually make decent speed and operational usage of the 2006 and up released applications you will require a minimal G5 by early 2007. If you last till early-mid 2008 you are most likely using your Mac for surfing, email and basic word processing. ;) :)

Hmm, let's see, that covers a bit of it... I'd also add on that list using Office v.X in general (not just Word), Sound Studio, Final Draft for my screenplay writing, MactheRipper/Handbrake/Toast 7 for DVD stuff, Audio Hijack Pro, Azureus, the entire iLife suite sans GarageBand, hmm, there's actually quite a few more apps as well... :p But I get your point... ;)
 
~Shard~ said:
My decision to hold off buying a new Mac until 2007 is looking better and better.... :cool:


With that logic why not delay until 2008. The MacBooks that year will be even faster than the ones the come out in 2007.
 
boombashi said:
I'm so mad at Apple and Intel :mad: - I haven't recieved my MacBook Pro yet, and already it is obsolete!!!

Just kidding, I couldn't resist! :D

twak.gif
;)
 
motulist said:
AMD is an option, but it would really be a last resort used only if Intel really drops the ball big time. The reason is that Intel is not only making the processor that the Mac's are using but also the entire chip set on the motherboard and I think I heard that they might even be making the motherboard altogether. I also heard that they might use Intel tech in future iPods too. The Intel switch is not just to their CPU, it's a switch to a single vendor for their entire technology platform.


Exactly. Intel would really have to screw the pooch for Apple to switch to AMD. Intel is more than a supplier to Apple, it's a partner. Apple is co-developing the next gen chips with Intel, of which Apple will get first AND ALSO royalties on every Win PC sold that uses the chips.

The reason Apple went with Intel over AMD in the first place is because Intel is the world's most effiecient producer of chips. So, unless Intel suddenly turns into a Moto/IBM and can't deliver (highly unlikely) Apple won't be going to AMD anytime soon, regardless of whether AMD makes a better product or not.
 
motulist said:
AMD is an option, but it would really be a last resort used only if Intel really drops the ball big time. The reason is that Intel is not only making the processor that the Mac's are using but also the entire chip set on the motherboard and I think I heard that they might even be making the motherboard altogether. I also heard that they might use Intel tech in future iPods too. The Intel switch is not just to their CPU, it's a switch to a single vendor for their entire technology platform.

Oh I agree completely, but the important factor for me remains that they at least have an option. Your partnership argument is strong, and I would imagine was a key factor in the selection.
 
AidenShaw said:
That's why I said "practical future" :)

To use up 64-bits with RAM you'd need a computer with about 4 billion DIMM slots, you'd need 4 trillion dollars (about the annual GDP of Japan), and over 100 Gigawatts of power.



Half a gig isn't that much, some home systems today have 2, 4 or even more. ("b" is bits, "B" is bytes)

I think you are missing the point somewhat. If memory chips were the same as they were in 1980, we would also need millions of slots. However technologies do not develop in isolation. It's a bit like saying that a PC with a 4Gb maximum needs twice as many slots as a 10 year old PC with a 2Gb maximum (made up numbers, but then again, this is all hyper hypothetical).

As for your second point, you are clearly a much more thorough and intelligent individual than I am, and I cannot thank you enough for pointing out my absolute ignorance as your correction greatly contributed to the discussion.
 
Chupa Chupa said:
With that logic why not delay until 2008. The MacBooks that year will be even faster than the ones the come out in 2007.

But then by the time 2008 rolls around, there'll be something cool coming in 2009 to wait for, then 2010, then it'll be OS 11 in 2011, and then.... :p :D Nah, if you play the waiting game, all you'll do is wait. I have made a committment to myself to buy a new Mac sometime in 2007, and I won't regret that decision no matter what. ;) Buy what you need, when you need it, and you'll never be disappointed. :cool:
 
~Shard~ said:
My decision to hold off buying a new Mac until 2007 is looking better and better.... :cool:

The thing is that they won't be commercially available until Q3 2007 at the earliest. So you might be in for some waiting until then.
I for one am going to buy mine at the first revision after Meron ships with Leopard. This time next year.
 
Diatribe said:
The thing is that they won't be commercially available until Q3 2007 at the earliest. So you might be in for some waiting until then.
I for one am going to buy mine at the first revision after Meron ships with Leopard. This time next year.

Yeah, you're right - I'll probably in the same boat as you. Merom with Leopard pre-installed sounds good enough to me! :cool:
 
a new factory to make 45nM chips

Don't you need an entire new generation of frabrication machinaries to make chips with transistor 45nm apart? If this is true, don't you need an entire new factory to do this? I wonder how much this new factory will cost (3 to 5 billion?).

Cinch
 
bugfaceuk said:
I think you are missing the point somewhat. If memory chips were the same as they were in 1980, we would also need millions of slots.
Actually, it would be about 16,000 DIMM slots (give or take a power of two or so), nowhere close to millions.

Memory chips have increased in capacity only 8000 fold since 1980.

And I see that <shift>b is still broken on your keyboard... :p
 
Fiveos22 said:
not many nanometers left, I better start saving mine.

That's okay. I've got plenty of picometers to spare; I'd be happy to share them (when the time is right).;)
 
AidenShaw said:
Half a gig isn't that much, some home systems today have 2, 4 or even more. ("b" is bits, "B" is bytes)
The term "gig" is rather imprecise, considering it comes from someone who regularly gives us lectures on why we should be using the term "GiB" instead of "GB".
 
jbh001 said:
Thanks okay. I've got plenty of picometers to spare; I'd be happy to share them (when the time is right).;)


I think a width of a typical atom (C or H) is measured in Angstrom (-10), whereas picometer is -12 meters. Can't have a transistor smaller than an atom.

Cinch:D
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Back
Top