Y doesn't Motorola let Intel make them a 3GHZ G4???

[peter2002]

Ok, this question has been bothering me for a very long time. With all the specualtion and hype about new or possible CPUs for future PowerMacs, the obvious solution seems right in front of everyone, but nobody is steping up to the plate to do something about it.

Most high tech compainies don't actually make anything anymore. Most gear is farmed out to Taiwan, Malayasia, or China because of cheap $1 an hour skilled labor. Since Motorola is cash poor and can't afford a $2 Billion state of the art fab, I don't see why they couldn't hire Intel to make a 3GHZ G4 for them. It wouldn't be difficult nor expensive since the actual cost of producing a 3GHZ P4 is only $50 per unit.

All this talk about reinventing the wheel with this hyped up vaporware IBM 970 is giving me gas in the lower abdomen.

If Motorola has anybody left with a half a brain they would let Intel make their G4. The profit margin would be a little thinner, but they would save $ by laying off those folks in their fabs that are behind the times. This idea would benefit the customer, and the developers because they wouldn't have to poor millions to convert their software to be optimized for an AMD or 970.

Peter :)

[FattyMembrane]

well, if moto can't make a faster g4, intel can't make a faster one with the same instructions. if my internal combustion engine car can only reach about 250mph and i hire you with your bigger factory to make the same car, yours will not magically be able to break 300mph. if moto could make faster g4s, they would. the problem is in the design, not the production. 2+2=4 regardless of the calculator you use.

[Nipsy]

First and foremost, the cost os producing a 3GHz P4 has nothing to do with the speed of cost of producing a 1, 2, 3, or 99GHz G4.

The reason Intel can clock higher and higher is that they do less and less per cycle. A 1GHz PIII is as fast as a 1.4 GHz P4. So, even if magically Intel could produce a G4, we would get a 2GHz chip, which performs like a 1.25 Moto G4...great...

Second, "reinventing the wheel with this hyped up vaporware IBM 970" is inaccurate in every way. The Power4 is the best chip money can buy at the moment. Apple/IBM are taking the biggest, fastest wheel, and scaling it back to from the server to the desktop.

Vaporware implies Copland, OSX on x86, Quark for OSX. The 970 will be produced (certain). The 970 will be a great performer (certain). The 970 may be the top desktop dog when introduced (50% certain), but it will certainly be on par with the best from Intel & AMD (certain) Apple will use the 970 in the top end machines (95% certain).

[ffakr]

So, any idea why Intel would be interested in making a PowerPC processor... even in IBM, Motorola, and Apple granted them license?
Would it be so that they could capture the enormous PPC desktop market share?
Would it be because they have a huge RISC development team waiting for something to do?

It won't ever happen... If only because Intel wouldn't want to do it. Where are the advantages to Intel?
Dump tons of R&D to design a new PPC for such a small market?
Make the competition better? (You'd have to assume Intel PPC would generate a smaller profit because of licensing, and that Intel PPCs would not generate secondary sales of intel chipsets)

Also, would IBM and Motorola hand Intel the keys to the PPC imbedded market? The PowerPC is a bigger player in embedded circles than it is in the desktop market.

Nope, such a situation doesn't benefit anyone in the right way.

[yosoyjay]

Originally posted by Nipsy


The reason Intel can clock higher and higher is that they do less and less per cycle. A 1GHz PIII is as fast as a 1.4 GHz P4.

This is kinda true. The work done per clock cycle did take a hit when Intel moved from P3 to P4. And it is true that the last versions of the P3 did out perfrom the earliest P4s, but that is very, very far from the truth now due to much, much faster clock speeds. Also, he newest, fastest P4s generally do more work per clock cycle than their P4 predecessors thanks to technology Intel inhereted from Digital. Thus, the inital performance snag that arose due to a change in chip design has been overcome and far surpassed due to insanely higher clock speeds and newly implemented technology.

So, even if magically Intel could produce a G4, we would get a 2GHz chip, which performs like a 1.25 Moto G4...great...

Why, if we pretend Intel did produce a G4 that ran at 2GHz, would it perform like a 1.25 Motorola G4? Are you assuming that an increase in clock speed automatically implies less work per clock cycle? If Intel did produce said chip wouldn't it behave exactly as a Motorola produced chip? So, if they did produce a G4 ran @ 2GHz wouldn't it far outperform any Motorola produced chip? Also, why would Intel be able to scale the same chip to 2GHz while Motorola couldn't?

The 970 will be produced (certain). The 970 will be a great performer (certain). The 970 may be the top desktop dog when introduced (50% certain), but it will certainly be on par with the best from Intel & AMD (certain) Apple will use the 970 in the top end machines (95% certain).

If you are 50% certain that the 970 will be the best desktop chip and certain that it will outperform future unreleased Intel & Amd chips, where is the uncertainty based? Do you think Motorola may possibly release the rumored G5 by then?

[daveg5]

how much does moto charge for there 1GHZ chips in Quanity anyway?
AMD?ANTHLON
INTEL? P4
TRANSMEDIA?
IBM? G3-970
DOES ANYONE KNOW?

[Nipsy]

Originally posted by yosoyjay


This is kinda true. The work done per clock cycle did take a hit when Intel moved from P3 to P4. And it is true that the last versions of the P3 did out perfrom the earliest P4s, but that is very, very far from the truth now due to much, much faster clock speeds. Also, he newest, fastest P4s generally do more work per clock cycle than their P4 predecessors thanks to technology Intel inhereted from Digital. Thus, the inital performance snag that arose due to a change in chip design has been overcome and far surpassed due to insanely higher clock speeds and newly implemented technology.



But the point is they are sacrificing elegant design for higher numbers. The work/clock ratio is still lower than the PIII. Feeding the 'bigger is better mentality' preys on the uneducated masses, who think 3000 cans of Coors is better than 2000 bottles of Bass. The higher numbers are going to give more performance, of course, but they do so at the cost of much wasted energy (heat). This will change with Itanium 2. I would sure rather have a 1GHz Itanium with a lower wattage, and higher performance. I would most certainly rather have a 2GHz PIII, with consumed less energy, and outperformed a 3GHz P4...


Why, if we pretend Intel did produce a G4 that ran at 2GHz, would it perform like a 1.25 Motorola G4? Are you assuming that an increase in clock speed automatically implies less work per clock cycle? If Intel did produce said chip wouldn't it behave exactly as a Motorola produced chip? So, if they did produce a G4 ran @ 2GHz wouldn't it far outperform any Motorola produced chip? Also, why would Intel be able to scale the same chip to 2GHz while Motorola couldn't?


Because the first thing they would do is increase the pipeline to gain MHz (like the P4). Longer pipeline == slower processing when branching occurs. Intel marketing dictates higher is better. A 3GHz Pentium sounds faster than a 1GHz Power 4, no?


If you are 50% certain that the 970 will be the best desktop chip and certain that it will outperform future unreleased Intel & Amd chips, where is the uncertainty based? Do you think Motorola may possibly release the rumored G5 by then?

Lemme quote myself, and then I'll explain:
The 970 may be the top desktop dog when introduced (50% certain), but it will certainly be on par with the best from Intel & AMD (certain)


By this I am saying that I think there is about a 50% chance that the 970 will debut as the fastest desktop chip (in processing horsepower, not in GHz) when it debuts. I am certain that it will be comparable in horsepower to the fastest x86 desktop chips at that time.

I think that the Moto G5 may be a goner. Based on the MR front page, the 970, and to a lesser degree the AMD hoopla, I am pretty confident we will never see a Motorola G5. IBM would have to fail to deliver (not something IBM is known for), additionally, even the wild speculation of x86 OSX boxes being used as a last resort makes me more sure that a Moto G5 will never be.

[tpjunkie]

When it comes down to it, comparing a P4 chip, which is a CISC chip with a G4, a RISC chip, is like comparing apples to oranges. (heehee, a little pun intended...who had out there an Orange PC card in their 68k mac?) You cannot compare the Mhz rating of one to the other and expect any kind of meaningful result.

[topicolo]

Originally posted by tpjunkie
When it comes down to it, comparing a P4 chip, which is a CISC chip with a G4, a RISC chip, is like comparing apples to oranges. (heehee, a little pun intended...who had out there an Orange PC card in their 68k mac?) You cannot compare the Mhz rating of one to the other and expect any kind of meaningful result.

Not only is that blind Apple propaganda, it's OLD Apple propaganda (1994 era). Except for a few quirky chip designs (itanium), CISC chips have become so RISC-like and vice versa that there isn't really any distinctive line between the two designs. RISC is supposed to stand for "Reduced Instruction Set Computing," but with Altivec and all the other extras they've put into G4s, they might as well be considered as CISC.

[jefhatfield]

stranger things have happened

ie) office for mac:)

[ffakr]

Originally posted by jefhatfield
stranger things have happened

ie) office for mac:)

That wasn't strange at all... Microsoft makes a lot of money off Apple users.
They didn't do it because they love Apple... They did it years ago because of the agreement in the stock deal... and they continue to produce Office because they make money on it... lots and lots of money.

[jefhatfield]

Originally posted by ffakr


That wasn't strange at all... Microsoft makes a lot of money off Apple users.
They didn't do it because they love Apple... They did it years ago because of the agreement in the stock deal... and they continue to produce Office because they make money on it... lots and lots of money.

i remember the really cold war period before the truce:p

[ffakr]

Originally posted by jefhatfield


i remember the really cold war period before the truce:p

I also remember hearing that MS office 4.2 (word 6 and such) sold around $400 Million a year... this is long before the truce...
that's a lot of money for such a crappy software package.

MS will continue to support Apple as long as the Mac Business Unit generates profit, and certainly as long as Apple can continue to make MS look like it isn't a monopoly (that is until they actually compete against MS in a real way.. then MS Smash Crush)

[peter2002]

You guys are missing the point. Don't go in another tantrum of which chip is better or faster. That is mute now.

The point is to get a faster G4 in a PowerMac without reinventing the wheel. The solution is to get Intel or maybe even AMD, which can at least hit 2.3GHZ which is way better than 1.25GHZ, to make it for them.

Peter :)

[Nipsy]

Originally posted by peter2002
You guys are missing the point. Don't go in another tantrum of which chip is better or faster. That is mute now.

The point is to get a faster G4 in a PowerMac without reinventing the wheel. The solution is to get Intel or maybe even AMD, which can at least hit 2.3GHZ which is way better than 1.25GHZ, to make it for them.

Peter :)

But that is essentially re-inventing the wheel...

The solution is IBM...

Neither Intel or AMD make PPC chips, therefore they would need to research, design, ramp-up, and produce the chip.

I hope you're not näive enough to believe that this can happen faster than we can get the IBM 970 (which has already been researched, design, and possibly sampled) from a company which has been making PPC chips for years.

Your next new PowerMac generation will be IBM 970 based.

[jefhatfield]

Originally posted by Nipsy


But that is essentially re-inventing the wheel...

The solution is IBM...

Neither Intel or AMD make PPC chips, therefore they would need to research, design, ramp-up, and produce the chip.

I hope you're not näive enough to believe that this can happen faster than we can get the IBM 970 (which has already been researched, design, and possibly sampled) from a company which has been making PPC chips for years.

Your next new PowerMac generation will be IBM 970 based.

so instead of moto being behind intel/amd, it will be ibm that will be our whipping post

G4 can scale up...ibm can not

ibm can go ddr, G4 cannot

what do we do?

make the 970 scale beyond 2+ ghz?

or make the G4 re-engineered to take ddr at some point?

the PC boys/gals are kicking our ass right now...but hopefully not for long;)