PDA

View Full Version : New Motorola chips faster than you think!


The Reaper
Aug 17, 2003, 02:05 AM
everyone has been noting that motorola's new chips will top out at 1.33GHz. remember that the 1.42 GHz G4 chips in the old powermacs were basically 1 GHz G4s from motorola, overclocked by apple. so the fastest 'old' G4 was a 1GHz chip, overclocked to 142% of the original speeds.

by extrapolation, if apple overclocks these new 1.33GHz G4s by the same amount, we will see G4 chips at 1.9GHz. although these overclocked chips will be hotter than the original 1.33ghz chips, remember that the new G4s are significantly cooler than the previous generation. this may mean that such high speed overclocked chips will run at the same temperature as the current 1GHz G4s in the powerbooks.

so we may end up with 1.9GHz G4 powerbooks. or, at the very least, something better than 1.33GHz.

this sounds like a good reason for apple to continue it's partnership with motorola (at least, until the G5 and it's supporting technologies have been engineered into a laptop). so the imacs and powerbooks can have one more upgrade cycle on the G4, and be fast, too.

what do you think?

Dreamail
Aug 17, 2003, 03:03 AM
Good idea, but your theory has two flaws unfortunately:

1.) Those 1.42GHz chips - contrary to urban legend - were not overclocked.
Motorola sold only 1GHz chips officially, because all they could produce at 1.42GHz went straight to Apple, leaving none for other clients.
This has been proven by people who actually removed the heat sink on their 1.42 CPUs (e.g. during installation of a water cooling system). The chips were clearly labelled by Motorola as 1.42GHz.

2.) Because of the reduction in the manufacturing process from 180nm to 130mn the new chips can go up in speed from 1GHz to 1.33GHz - while consuming exactlty the same amount of energy and hence producing still the same amount of heat.
True, the new 1GHz chip done at 130nm will be a lot cooler than the old 1GHz chip at 180nm, but the new 1.33GHz chip will have the same heat dissipation as the old 1GHz chip. That's actually how they come up with the 1.33GHz as far as I understand: they leave everything else constant (as to the old chip), things like energy consumption and see how fast the new one will go with that.

To sum it up: sorry, 1.33GHz it is.

Nevertheless, there could still be a small patch of chips produced at a higher clockrate, again one which Motorola does not advertise and only Apple will receive (real ones again, not overclocked ones!), but those will run hotter than the current 1GHz chips.

Sun Baked
Aug 17, 2003, 03:17 AM
In 99.5% of the material it'll be shown as a 1.3MHz part.

A couple places it'll show a speed range of 500-1333 MHz -- which means 1.33GHz still falls within most tolerances.

But the 1.3GHz limit is Motorola keeping their old 105C Junction temperature with the new chip. (Look at the PPC page temp is same all the way down the page.)

Sure you can crank up volts and MHz, but it'll need to be tested for a different set of tolerances. Like the old 1.42GHz 7455 r3.3

You should see the chips hitting the upgrade manufacturers with the high tolerances quickly, as Apple's need for the high performance chips is not as great this product cycle.

Apple will need the lower temp chips for portables, and the iMac/eMac.

tazo
Aug 17, 2003, 04:12 AM
I hope for the sake of any portable g4 users, Apple does not implement a 1.33 ghz g4 into its portable line; that is to say without including a complimentary burn kit :rolleyes: ;)

mim
Aug 17, 2003, 04:16 AM
The best rule of thumb is (unfortunaltley) - it's Motorola so don't be optimistic.

What we know of the chip, and of the G4's development history doesn't lead us to expect the kind of speed jumps you are suggesting.

Still - a jump to 1.33Ghz will be nice.

The Reaper
Aug 17, 2003, 08:40 AM
ah, ok, that makes sense. i was hoping that perhaps the old belief (that the 1.42s were overclocked) was correct.

so, just out of curiosity, is it possible that the 1.42s ARE overclocked 1GHz chips (by motorola)? and that the 1.33 is the new 'base' chip, ready to be overclocked to higher speeds (by them)? because if it IS motorola who is overclocking them, then all chips from motorola, including the 1.42s, would have the motorola label, as expected.

perhaps this is what the original overclocking statements were referring to, and people misinterpreted them to mean that apple was doing the overclocking. perhaps it was just another example of motorola incompetence (ie they couldn't make faster 'base' chips, so they just overclocked them before they shipped them to apple. now they have a new, 33% faster base chip to work with).

because if this was the case, then all current G4 lines could see a 33% performance boost. the imacs could see even more (they currently top at 1GHz, no higher, so they wouldn't steal the G4 Powermacs 'thunder'. i don't think that heating issues were the deciding factor in limiting the iMacs. so now, with the G4 Powermacs out of the way, the iMacs and eMacs have room to grow.)

just some random thoughts. correct me if i am wrong.

Sun Baked
Aug 17, 2003, 09:20 AM
Don't know what games Motorola is playing, since the public document seem to have been slightly revised before being publically posted. ;)

It's doubtful Apple got overclocked chips if they were in Motorola's 7455 r3.3 Part Number matrix. Seems like quite a few people actually have seen these documentation errors.

But the 7455 has been bumped to r3.4 also.

negrito
Aug 17, 2003, 03:57 PM
Originally posted by The Reaper
...perhaps this is what the original overclocking statements were referring to, and people misinterpreted them to mean that apple was doing the overclocking. perhaps it was just another example of motorola incompetence (ie they couldn't make faster 'base' chips, so they just overclocked them before they shipped them to apple. now they have a new, 33% faster base chip to work with).

you can't overclock processors without additional hardware. you never know at which speed a processor will run when you produce it. of course there are percentages of processors which can be expected to run at a certain speed. on one wafer there will be for example the main part (let's say 60%) that will run at 1.25ghz some will run at 1ghz and some at 1.33ghz. you have to test them to know their speed. and following to their speed they will be branded.

those speeds are recommended speeds! which are mainly on the top limit of the possible speed. with a motherboard you can may overclock those processors but they won't run stable.

in old days the pentium had a big margin of manufaturer recommended speed and overclocked speed. the amd had a smaller margin (maybe 50 to 100mhz without cooling exchange etc)

as the g4 speeds are raising in small steps you can expect thet the recommended speed is very near the overclocked speed. there is a very small gap and if you are looking for a low heat emmiting and energy saving proc (for powerbooks) you won't (and maybe can't) go 33% over the recommended speed.

rice_web
Aug 17, 2003, 05:00 PM
I can all but guarantee that these new chips will run 1.8GHz+

Well, at least some of them.

Maybe 5% of Motorola's past chips clocked past 1GHz, but now, with the new manufacturing process, yields of higher-clocked chips have increased. At least 50% of Motorola's chips probably now clock at 1.33GHz, with the top 5% clocking in the 1.8GHz range or so.

Given the limited number of purchases that Apple makes with Motorola, Apple will likely be able to offer G4 systems up to 1.8GHz, though the heat would be incredible (ruling out PowerBooks at this level), and no machines appear to be ready for a performance boost of this magnitude (if the iMacs, for example, were bumped to 1.8GHz, it'd be a jump from 1GHz, a more than significant increase).

Oh, and concerning the heat argument, a 1.33GHz G4 on .13 manufacturing will produce the SAME amount of heat as the 1GHz G4s that we have in the PowerBooks.

Overall, I wouldn't expect a speed bump for the PowerBooks/iMacs/eMacs beyond 1.3GHz, no matter what the yields from Motorola.

Daveman Deluxe
Aug 17, 2003, 05:24 PM
Core clock designation is really a simple affair.

After several wafers of chips are made, the cores are tested at various clock speeds to determine how fast they can run. Because there are "good" and "bad" portions of any silicon wafer, and because one can reliably predict what the quality of each section of the wafer is, the sampling of the first batch is used to statistically determine the maximum stable core clock frequency of a chip in a particular position on a silicon wafer.

This dramatically reduces overhead from having to test each and every processor core.

Furthermore, the market also determines what processors are rated at. A large part of the reason that it's common to be able to overclock Pentium IIs 100 MHz or higher without additional cooling is because of the way the market worked at the time. If Intel had sold every chip that could run 450 MHz as such, there would have been a glut of 450 MHz processors. Therefore, a lot of processors that could have reliably run at 450 MHz were sold as being 350 MHz cores. Supply and demand strikes again. Once a lot of geeks figured this out, they started buying 350 MHz cores and clocking them up to 450 MHz, which is why Intel introduced clock-locking.

The market is the reason Motorola only sold 1.42 GHz processors to Apple. Nokia has no need for a 1.42 GHz processor, but Apple did. The best processors and ONLY the best processors were sold to Apple, rated at 1.42 GHz. Had Moto released 1.42 GHz procs to the general public, prices of other cores that cost just as much to produce would have dropped.

That last paragraph is speculation, but I think it's a fairly good one.

ColoJohnBoy
Aug 17, 2003, 05:47 PM
Can someone point me in the direction of the heat/power specs of the 7457 that were posted on the sight a couple months back? I'd be curious to see how they compare to my 1GHz G4.

I can't use this as a true laptop when doing gaming, DVD burning, or editing in Final Cut Express: it just gets too hot, hotter even than any 12" PowerBook I've come across.


Visit Blue Pudding!
http://bluepudding.1hwy.com

Rezet
Aug 17, 2003, 05:48 PM
Originally posted by Dreamail
Good idea, but your theory has two flaws unfortunately:

1.) Those 1.42GHz chips - contrary to urban legend - were not overclocked.
Motorola sold only 1GHz chips officially, because all they could produce at 1.42GHz went straight to Apple, leaving none for other clients.
This has been proven by people who actually removed the heat sink on their 1.42 CPUs (e.g. during installation of a water cooling system). The chips were clearly labelled by Motorola as 1.42GHz.

2.) Because of the reduction in the manufacturing process from 180nm to 130mn the new chips can go up in speed from 1GHz to 1.33GHz - while consuming exactlty the same amount of energy and hence producing still the same amount of heat.
True, the new 1GHz chip done at 130nm will be a lot cooler than the old 1GHz chip at 180nm, but the new 1.33GHz chip will have the same heat dissipation as the old 1GHz chip. That's actually how they come up with the 1.33GHz as far as I understand: they leave everything else constant (as to the old chip), things like energy consumption and see how fast the new one will go with that.

To sum it up: sorry, 1.33GHz it is.

Nevertheless, there could still be a small patch of chips produced at a higher clockrate, again one which Motorola does not advertise and only Apple will receive (real ones again, not overclocked ones!), but those will run hotter than the current 1GHz chips.

Well if they were not overclocked, why did apple give up on them so quickly? They don't even want to sell out old supplies right now. Why arent they selling 1.42 anymore? Seems something isnt right with that processor...

Bear
Aug 17, 2003, 06:09 PM
Originally posted by Rezet
Well if they were not overclocked, why did apple give up on them so quickly? They don't even want to sell out old supplies right now. Why arent they selling 1.42 anymore? Seems something isnt right with that processor... I see a few possibilities:[list=1] Apple wanted to only sell one G4 configuration and they wanted to keep one that would support Mac OS 9.
Motorola had a hard time producing enough 1.42GHz chips
Apple wants to leave a bigger differentiation between G4 and G5 systems.
[/list=1]I suspect it's the first reason. And that may as well be why the 15" Powerbook hasn't been updated/replaced with a 15" Aluminum Powerbook.

solvs
Aug 17, 2003, 07:04 PM
Originally posted by Rezet
Well if they were not overclocked, why did apple give up on them so quickly? They don't even want to sell out old supplies right now. Why arent they selling 1.42 anymore? Seems something isnt right with that processor...

I see someone beat me to it by the time I hit reply, but I'll still put my $0.02 in.

The 1.42 GHz G4s are hot. They were not exactly overclocked, as Motorola marked them as 1.42s. By definition, oc'ing them would be running them past manufacturers rated speeds. Since Moto and Apple produced them, they can't be called oc'ed. You would be seeing a lot more problems anyway. Many of the machines would have died or had stabiltiy issues. Moto rated them at 1.42, Apple sold them at 1.42, therefore they are 1.42. End of story.

Now that that's out of the way, we'll probably be seeing PBs, and later eMacs, in the 1 to 1.33 GHz mark. I would be happy to see higher, but I'm not counting on it. Now the iMacs may be higher, if they can deal with the heat issue. The iBooks will probably stay with the G3, but at 1 to 1.1 (or 1.2) at first. Higher when (if) the G3+ Altivec ships. Maybe they can put that in a nice, cheap headless Mac. Cool, fast, cheap.

But I don't see PB being too much faster thatn 1.33. Unless Apple really does oc them this time. 9 fans and 1 hour of battery life!

Just like a Dell (http://www.theregister.co.uk/content/54/28245.html).

Sun Baked
Aug 17, 2003, 07:57 PM
Missed the other 7455 r3.3 part number thingie. The 1400 is on this one. ;)

Looks like they did some more updates last couple weeks, so the r3.4 7455 stuff is starting to show up.

The last letter in part number "F" is rev 3.3 -- the r3.4 is probably a "G".