1.25GHZ is just overclocked 1GHZ

Originally posted by itanium
How do you overclock it. I wan't to run 867 MHz G4 at 1.25 GHz. Show me how.

Oh wait, you can't? I guess that 1.25 GHz G4 is worth a little more to me than that 867 MHz G4. Hmmm, go figure!

Your argument is weak, all CPU manufacturers do this. Why the sudden interest in Motorola?

Also, not all 1 GHz G4s will run stable at 1.25 GHz. What don't you understand about this. This is what makes the 1.25 GHz G4 special and worth the additional cost. Lower yield equals higher cost. Tada!

You're repeating yourself and ignoring my answers, and your arguments are self-contradicting. For the last time; we're talking about Apple, not Motorola.
And go back and read my posts once more.

Sure I can show you how to overclock your 867 MHz. But I can't guarantee it will run at 1.25 GHz without additional cooling. PPC7455 bus multiplier is set by R2, R5, R8, R4 and R10 where R2 is PLL CFG0, R4 PLL CFG1, R5 PLL CFG2. R8 PLL CFG3 and R10 PLL EXT. The resistors are located left to the processors. If you add a peltier with water cooling, or cool it with liquid nitrogen to lower resistance, sure you can clock it to 2 GHz+.
What Apple has probably done here is to add aditional cooling to make 1 GHz chip run stable at 1.25 GHz. That's $650 for four ugly holes in the front and an even more noisy fan which makes the computer sound like a jumbojet heading at full speed against the ground.
I'd rather invest 70 bucks in a peltier, thermal paste and some quality fans to make a 1 GHz Quicksilver run at even higher speeds than 1.25 GHz. Then I wouldn't have to pay three times as much for useless DDR RAM, either.
 
Originally posted by plastic
Well, in Japanese Mac Mod shops, they are always OC-ing this and that... and you can find loads of Japanese Mac Sites doing this (second largest market in the world for Apple) and what they do not show you is....

How long does the OC-ed G4 run stable? Tada! They can get a G4-1Ghz to run at 1.25 for an hour? And after a month? What happens? Fried? Nah, they will not tell you this... so I kinda have to support what itanium mentioned. These are more stable chips that was used in 1.25 towers.

I dun care if they have found a new way to OC a 400Mhz G4 chip to a 5Ghz G5... as long as it works, I will part with my money for it. Simple as that. I think Apple does a decent QC job.

My opinion. Dun bite me.

So it's guaranteed to be stable and all good when Apple does it? :rolleyes:
 
This is not hard, m'kay?

Apple buys many G4s from Motorola. Some are rated by Motorola at 1GHz, some at 867 MHz. Apple knows that the chips rated at 1GHz have not been tested at higher speeds, since 1GHz is the highest Moto will test at. So they test the 1GHz-rated chips at 1.25GHz. Some fail at 1.25 GHz. These they sell as 1GHz processors, since they are already rated (by Moto) at that speed. Some pass at 1.25 GHz. These they sell as 1.25 GHz processors (with extra cooling, since more Hz => more W => more heat).

Now, for the grand prize of not getting kicked in the ass by Leroy the arthritic mule, answer these questions:

a)What is the difference between chips sold by Motorola as 1GHz chips and those (re)sold by Apple as 1GHz chips?

b)What would be the likely result if we were to overclock to 1.25GHz, and leave running for more than an hour or two, the chips sold by Apple as 1GHz chips?
 
Originally posted by scruffy
This is not hard, m'kay?

Apple buys many G4s from Motorola. Some are rated by Motorola at 1GHz, some at 867 MHz. Apple knows that the chips rated at 1GHz have not been tested at higher speeds, since 1GHz is the highest Moto will test at. So they test the 1GHz-rated chips at 1.25GHz. Some fail at 1.25 GHz. These they sell as 1GHz processors, since they are already rated (by Moto) at that speed. Some pass at 1.25 GHz. These they sell as 1.25 GHz processors (with extra cooling, since more Hz => more W => more heat).

Now, for the grand prize of not getting kicked in the ass by Leroy the arthritic mule, answer these questions:

a)What is the difference between chips sold by Motorola as 1GHz chips and those (re)sold by Apple as 1GHz chips?

b)What would be the likely result if we were to overclock to 1.25GHz, and leave running for more than an hour or two, the chips sold by Apple as 1GHz chips?

No.
More MHz does not necessary mean more watts. A real 1.25 GHz can have exactly the same watt rating as a 1 GHz.
Motorola doesn't test each invidual processor.
It's likely that Apple doesn't, either. That could be the reason for the extreme cooling system. Sure, any processor can be run at high frequencies with proper cooling.

OK, to answer your questions;
a) If Apple does test each invidual processor, a 1 GHz chip from Motorola can potentially be run on higher frequencies that the ones you get from Apple, and the average temperature/wattage is lower because the ones from Apple are found to be unstable or not to run at all on 1.25 GHz with Apple's cooling.

b) Again, if Apple does test each invidual processor, that wouldn't work on a new DDR G4. With a Quicksilver CPU module, though, it could work just fine with good cooling (unless Apple did the same when they sold Quicksilvers, and have collected chips that can be run on 1.25 GHz for a long time).
I've overclocked a G3 from 300 to 440 MHz, and my current G4 from 400 to 500 MHz, and both of them still run perfectly stable.
 
Chip makers do test every CPU they make. And something like 25% of them don't run acceptably at any speed, and need to be thrown out. (A couple of years back, someone used FPGA's to make a low-cost supercomputer out of partially functional CPUs that would otherwise have been thrown out, but that's a digression)

a) If Apple does test each invidual processor, a 1 GHz chip from Motorola can potentially be run on higher frequencies that the ones you get from Apple, and the average temperature/wattage is lower because the ones from Apple are found to be unstable or not to run at all on 1.25 GHz with Apple's cooling.

Exactly. Can consumers buy them direct from Motorola? That would be especially interesting if one could get a commodity motherboards that would run OS X and hold G4s

b) Again, if Apple does test each invidual processor, that wouldn't work on a new DDR G4. With a Quicksilver CPU module, though, it could work just fine with good cooling (unless Apple did the same when they sold Quicksilvers, and have collected chips that can be run on 1.25 GHz for a long time).
I've overclocked a G3 from 300 to 440 MHz, and my current G4 from 400 to 500 MHz, and both of them still run perfectly stable.

Did Quicksilvers come up to 1GHz? That is an interesting thought, certainly.

Second part - so you've overclocked chips before, and gotten lucky. Congratulations, and may your good luck continue (should you OC more chips). Still, I wouldn't run any scientific apps on those machines... It should be noted that the testing consists of more than just "does the OS run stable", which isn't really much of a workout for the CPU. Generally they will do things like calculate pi to the 50,000th digit, and if even one number is out it's considered an unacceptable chip at that speed
 
Originally posted by ksv
I tend to be the last poster in discussions...

no way, i'm the designated thread killer here! (and pretty much everywhere)

i'm going to go ahead and accredit it to irrefutable arguments ^_^.
 
Originally posted by scruffy
Did Quicksilvers come up to 1GHz? That is an interesting thought, certainly.

What do you mean? I have a QS Dual-1G. Or do you mean they were overclocked Dual 867s now... :(
 
I mean I wasn't sure. I check exact specs of new Apple computers when I'm actually thinking of buying one, or helping someone else pick one out.

Just to give you an idea of how often the former happens, I have a 300 MHz B/W G3.
 
Originally posted by scruffy
Exactly. Can consumers buy them direct from Motorola? That would be especially interesting if one could get a commodity motherboards that would run OS X and hold G4s
Exactly. Can consumers buy them direct from Motorola? That would be especially interesting if one could get a commodity motherboards that would run OS X and hold G4s
You'd still need the rest of the processor card (clock, L2/L3 cache etc)

Originally posted by scruffy
Did Quicksilvers come up to 1GHz? That is an interesting thought, certainly.

The last ones were 800, 933 and 2x1000 MHz, I think.

Originally posted by scruffy
Second part - so you've overclocked chips before, and gotten lucky. Congratulations, and may your good luck continue (should you OC more chips). Still, I wouldn't run any scientific apps on those machines... It should be noted that the testing consists of more than just "does the OS run stable", which isn't really much of a workout for the CPU. Generally they will do things like calculate pi to the 50,000th digit, and if even one number is out it's considered an unacceptable chip at that speed

When I leave it calculating mandelbrot fractals for 24 hours, and doesn't crash, I consider it as stable ;)
I've overclocked many other G3s, too, and none of them have had instability problems.
 
i've read this thread in full, and it has some interesting merits. but can someone tell me why it matters if you chip is overclocked? i agree with the statement that this happens in the pc industry all the time. so why is this news? if my processor was "approved" in some sense to run faster, you can bet that i would find a way to overclock it. but if you cant promise me that i wont fry it, well then...

i just dont see what all the hub-ub is about. at most the only real news here is that now people have more reason to bash moto for not putting out a new chip.
 
The point is not the fact that the 1GHz and 1.25GHz chips use the same fab process. That's no big deal. It's made painfully obvious that the circuit diameter didn't shrink and the core voltage remained the same by the size of the heat sink and the consumed wattage. What pains me is the manner in which the material is presented. Both by posters here and by the link provided just above.

This is industry norm. Not industry freak. And should be represented as such. Also, the big difference between me overclocking my processor and Apple overclocking my processor is ... that it IS guaranteed by Apple. For a year. Or 3 for about 10% over the cost of the machine. What a guaranteed seller of chips says is the speed, is the speed. The facts are true, their presentation is severely flawed. This is an indication that moto isn't introducing new designs very fast. ... we know that already.
 
I am working in the semiconductor industry.

There is a difference between me overclocking my G3 from 300 to 400 MHz and Motorola testing the same chip under controlled conditition to rate it 300 or 400 MHz.

What we don't know here is who decided these G4 were good for 1.25 GHz. Did Motorola test these at 1.25 GHz (and deliver only to Apple), did Apple receive test info from Moto and use these accordingly, or does Apple just overclock and cool ?

Or is the speed limitation related to temperature, in which case just cooling down the die would allow for higher speed ? This would also be under control.

The only concern is that a device may be unreliable if it is operated outside its specification. We don't have the real specification of the device (only the published ones are available on Moto web site, but the real one may differ), neither do we have the test specification.... therefore all our discussion is... speculation.
 
I wholehartedly agree with not buying a new G4, and picking up a new pentium with XP. I just bought the new G4 Dual 1GHz with 1 gig RAM and jaguar and I'm utterly disturbed and profoundly disappointed with it's performance in every aspect. I used to be a Mac person, it's the sad truth, Apple has finally dropped the ball and gone south. We Mac people need to come out of our shell of denial about Jaguar. People need to know the truth about this. You can read my full report in the Opinion forum....
 
Originally posted by chevy
I am working in the semiconductor industry.

There is a difference between me overclocking my G3 from 300 to 400 MHz and Motorola testing the same chip under controlled conditition to rate it 300 or 400 MHz.

What we don't know here is who decided these G4 were good for 1.25 GHz. Did Motorola test these at 1.25 GHz (and deliver only to Apple), did Apple receive test info from Moto and use these accordingly, or does Apple just overclock and cool ?

Or is the speed limitation related to temperature, in which case just cooling down the die would allow for higher speed ? This would also be under control.

The only concern is that a device may be unreliable if it is operated outside its specification. We don't have the real specification of the device (only the published ones are available on Moto web site, but the real one may differ), neither do we have the test specification.... therefore all our discussion is... speculation.

Why should Motorola sell chips capable of running at 1.25 GHz rated at 1 GHz? That would just make less profit...
Motorola's chip's clock frequency ratings aren't related to temperature. Sure, cooling down the die allows the processor to run at higher frequencies, but I'd rather do that myself than paying $650 for Apple doing it.
 
You didn't get me right.

Either Moto only delivers to Apple because they have a contract to do so, or because Apple takes all the production at the best price, or because Apple accepts to pay a premium for extra testing and binning (sorting).

Or by cooling the chip Apple is able to run it faster. Managing the temperature of a component is not that easy ! And development of a new cooling scheme also costs, someone has to pay for it...

BTW, the only solutions for a much faster Mac now are:
Wider processing (64 bits or above)
New technology (faster clocks or parallel computing)
Distributed processing (Apple is already going this route with Quartz Xtreme), this is the option with the highest acceleration potential, but also the least versatile and probably the most expensive as all processors become specialized. This is the way biology is operating and it as been demonstrated to be quite efficient....
 
Originally posted by chevy
You didn't get me right.

Either Moto only delivers to Apple because they have a contract to do so, or because Apple takes all the production at the best price, or because Apple accepts to pay a premium for extra testing and binning (sorting).

Or by cooling the chip Apple is able to run it faster. Managing the temperature of a component is not that easy ! And development of a new cooling scheme also costs, someone has to pay for it...

If Motorola sorts out 1.25 GHz chips, they'd sell them as 1.25 GHz, not as 1 GHz. If Apple sorts them, the overall quality of the 1 GHz G4s is lower, and can't be overclocked as easily as pre-DDR G4 Macs, like I've said three or four times now ;)
Cooling down a processor effectively isn't harder than using a peltier and a good fan. A peltier costs about $20. I'm sure I could clock a 1 GHz G4 way beyond 1.25 GHz for $650!
But, we can of course not say anything for sure until someone actually have checked the 1.25 GHz if they really are rated by Motorola at the speed they are running at.

I'm not saying Apple is doing anyhing wrong as long as their computers are stable, I'm just saying I'd rather buy a dual 1 GHz Quicksilver than a dual 1 GHz DDR.
 
Back
Top