People complain about G5 progress, BUT...

[DakotaGuy]

You must give them a little credit. It took from April 1976 to January 2002 to move from 1Mhz to 1Ghz. That is just shy of 26 years. It took from January 2002 to April 2005 to move from 1Ghz to 2.7Ghz. That is just a little over 3 years. 26 years for 1Ghz and 3 years for 1.7Ghz. That doesn't seem way out of line. Now I will get bashed by people that will say, but look at what Intel has done in that amount of time, and that is fine, but what I am saying is technology in general is moving much faster these days then we give it credit sometimes.

 

[Daveway]

Good point, but we were promised 3Ghz.....

 

[Lacero]

Sorry but I was expecting we hit 22 THz by now.

 

[Dr. Dastardly]

Sorry but I was expecting we hit 22 THz by now.

At the very least 20, 18 if it has 27 cores.

 

[James Philp]

Sorry but I was expecting we hit 22 THz by now.

If you have silly money (a few mill) you can!
Fancy running Doom 3 on one of those super-computer x-serve clusters?

 

[crap freakboy]

If you have silly money (a few mill) you can!
Fancy running Doom 3 on one of those super-computer x-serve clusters?

I'd still be underwhelmed with the gameplay.

 

[wrldwzrd89]

If you have silly money (a few mill) you can!
Fancy running Doom 3 on one of those super-computer x-serve clusters?

Actually, I'd run Doom 4 Mac OS X edition, optimized for Xgrid...that way I'd get the best possible performance from my cluster.

Anyway, getting back on topic...

I agree with the point of the topic starter. However, it seems that every processor maker has "hit the wall", so to speak, when it comes to CPU speed. Advancements there are coming slower than they have in the past. My personal opinion as to why this is is that someone (I don't remember who it was) said that chip making would run up against some physical limits at around the 100 nm (nanometer) mark. Sure enough, that's exactly what happened when the 90 nm parts were released (just look at what happened to both Intel and IBM).

 

[James Philp]

I'd still be underwhelmed with the gameplay.

That's what I thought! Played Brothers In Arms though? What a game!

 

[salmon]

My personal opinion as to why this is is that someone (I don't remember who it was) said that chip making would run up against some physical limits at around the 100 nm (nanometer) mark. Sure enough, that's exactly what happened when the 90 nm parts were released (just look at what happened to both Intel and IBM).

The problem with using these quotes as a predictor in the past has been that they have been notoriously inaccurate, and Moore's Law has been much more accurate. I remember reading years ago that they thought CPU speeds would top out at 100Mhz - not quite as bad as the "only ever see the need for 5 computers" and the "640KB is all anyone will ever need" predictions.

I do agree that it seems like this one is finally coming true though, everybody is having trouble with it.

 

[PlaceofDis]

i totally agree, things have certainly come a long way, and yes the Mhz speeds seem to have hit a wall, it will be interesting to see where the chip makers go from there, besides dual core of course

 

[GFLPraxis]

You must give them a little credit. It took from April 1976 to January 2002 to move from 1Mhz to 1Ghz. That is just shy of 26 years. It took from January 2002 to April 2005 to move from 1Ghz to 2.7Ghz. That is just a little over 3 years. 26 years for 1Ghz and 3 years for 1.7Ghz. That doesn't seem way out of line. Now I will get bashed by people that will say, but look at what Intel has done in that amount of time, and that is fine, but what I am saying is technology in general is moving much faster these days then we give it credit sometimes.

Quick math.


1 MHz to 1 GHz = 100,000% increase. Took 26 years for that.
1 GHz to 2.7 GHz = not even 200% increase.


I'm not complaining, just pointing out a flaw in your reasoning. I'm fine with the G5 processor now, I don't feel its especially slow, since the others are having similar problems.

 

[nrd]

The problem with using these quotes as a predictor in the past has been that they have been notoriously inaccurate, and Moore's Law has been much more accurate. I remember reading years ago that they thought CPU speeds would top out at 100Mhz - not quite as bad as the "only ever see the need for 5 computers" and the "640KB is all anyone will ever need" predictions.

I do agree that it seems like this one is finally coming true though, everybody is having trouble with it.

The limit (I believe it's 90nm) is due to the fact that copper wire can't be reduced any further without being stable. It's not a prediction, it's a fact of life. Chips aren't going to get much faster until that hurdle is cleared, hence the shift to alternative technologies like dual-core processors.

If I'm off on something, correct me.

 

[GFLPraxis]

The limit (I believe it's 90nm) is due to the fact that copper wire can't be reduced any further without being stable. It's not a prediction, it's a fact of life. Chips aren't going to get much faster until that hurdle is cleared, hence the shift to alternative technologies like dual-core processors.

If I'm off on something, correct me.

Or they find an alternate material to use instead of copper.

But no, its not 90nm, we have 65nm processors out now.

 

[wrldwzrd89]

Or they find an alternate material to use instead of copper.

But no, its not 90nm, we have 65nm processors out now.

I'd love to know which processor maker is shipping these tiny processors, if only because I'm a technology follower and like to know about these sorts of things.

 

[revenuee]

http://www.newsfactor.com/perl/story/22353.html


This is the technology that i'm guessing everyone will get exited over

 

[crap freakboy]

That's what I thought! Played Brothers In Arms though? What a game!

Yeah cracking game, lost most of my team on the later farm mission, now I've gotta take out gun placements with 3 men.

 

[R.Youden]

Copper track based CPU's are on the way out. The main problem is that it takes too long for the signal to pass from one end of the copper to the other. I can not remember which company it is but they are developing fiber optic based CPU's. This means that data can travel literally at light speed throughout a CPU. They are now developing a way to maintain siganl strength so that informtaion does not get blurred!

 

[slipper]

Good point, but we were promised 3Ghz.....

...a year ago

 

[840quadra]

we just need to use that super computer to help Macrumors climb to the top of the ladder at Folding@home

 

[iBunny]

I am going to throw this out there.

The Pentium 4's fastest resides at 3.8Ghz. I dont want to even talk about performance and Intel and windows, because I know all you guys are mac guys. But just as a general thing in computing... That is our Fastest *In MHz* CPU right now. which is only a 380% increase from 2002.

I still was expecting more all around the table from IBM/Apple, Intel and AMD. But I really do think 10Ghz is going to happen sooner than later. Maybe 2010. Thats not so bad if you think about it. Progress is comming along nicely.

Also, Sure our G5 is only at 2.7Ghz. However look at this. Think back to the days of the first 1Mhz Processor. Weather it be a x86 or whatever CPU. if you could theoretically clock that to 2.7Ghz, it still wouldnt be as fast as the G5. Mhz isnt always the focus factor, Prrocessors have been tweaked thru countless revisions to reach the complexety of the G5. Which allows more features and performance than processors of old.

So I think we are doing ok.

 

[redeye be]

http://www.newsfactor.com/perl/story/22353.html
This is the technology that i'm guessing everyone will get exited over

interesting
10Ghz, no cache plz.

Developing this main board replacement sure looks like it's going to be a pain in the @$$. Going 3d to make them all 'connect' probably. Cool tech nontheless.