Will the G6 be the G5 everybody's been waiting for?
A little tidbit from <a href="http://www.macbidouille.com/"; target="_blank">Macbidouille</a>for 07-25-02 points to THIS: [quote] There has been disappointment in the DSP world that the current proposed architecture for G5 (the MPC8500 series) is not AltiVec-enabled, leading many to continue banking on further improvements in the G4 line until release of the G6 -- perhaps by the end of 2003. The G5 architecture does, however, call for a built-in RapidIO interface, which also is something many in the industry are anticipating; the G6 is expected to have AltiVec and RapidIO, although performance parameters are defined by the telecommunications and computer industries, not the military.
"The intention is for RapidIO to propagate across all product lines, but we can't get into timeframes' says Motorola's Handa. "We are making a very strong investment in the G4 architecture and will take it all the way through the next process revisions to provide higher and higher performance levels. I can't comment about what comes beyond that, except to say we will draw out from the G4 architecture, while maintaining software and hardware compatibility; over the next 12 to 36 months. So those who are designing us in at up to 1GHz, can expect much more of the same while maintaining pin and software compatibility. As we map the G4 family into the next generation, for example, there will be dramatic decreases in heat at equivalent megahertz." <hr></blockquote>
(taken from <a href="http://www.infinibandta.org/parchives/clippings/msg00072.html"; target="_blank">the archives of the Infiniband Trade Association</a>)
Looks like there won't be a PM G5 at all... <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
"The intention is for RapidIO to propagate across all product lines, but we can't get into timeframes' says Motorola's Handa. "We are making a very strong investment in the G4 architecture and will take it all the way through the next process revisions to provide higher and higher performance levels. I can't comment about what comes beyond that, except to say we will draw out from the G4 architecture, while maintaining software and hardware compatibility; over the next 12 to 36 months. So those who are designing us in at up to 1GHz, can expect much more of the same while maintaining pin and software compatibility. As we map the G4 family into the next generation, for example, there will be dramatic decreases in heat at equivalent megahertz." <hr></blockquote>
(taken from <a href="http://www.infinibandta.org/parchives/clippings/msg00072.html"; target="_blank">the archives of the Infiniband Trade Association</a>)
Looks like there won't be a PM G5 at all... <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
Comments
So basically, the Motorola G5 will always be a low-end CPU (in the scheme of things).
The Motorola G4 will be developed until the Motorola G6 (assuming Moto survives that long).
One thing this does is once and for all confirm that the pre MWSF02 Motorola G5 rumors at MOSR and The Reg were all false.
And hey Dorsal, what was that about Motorola G5 development going well? <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
Barto
[EDIT: Put Motorola in front of every Gx]
[ 07-29-2002: Message edited by: Barto ]</p>
<strong>A little tidbit from <a href="http://www.macbidouille.com/"; target="_blank">Macbidouille</a>for 07-25-02 points to THIS:
(taken from <a href="http://www.infinibandta.org/parchives/clippings/msg00072.html"; target="_blank">the archives of the Infiniband Trade Association</a>)
Looks like there won't be a PM G5 at all... <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" /> </strong><hr></blockquote>
Information like this is starting to make me think that Apple really has been working on another CPU strategy, when these new intiatives bear fruit is another story.
Based on the water cooler conversation I've been hearing of late, 2004 looks like a more realistic target than 2003 for releasing a new CPU architecture for the Mac. For the rumormongers out there, I think the REAL dirt is going to come from people who are actively working on the existing G4 platform.
I wonder? What information is there about current G4 development? Anyone have any links to articles that speculate where the G4 is heading in the near-term?
<strong>
So basically, the G5 will always be a low-end CPU (in the scheme of things).
The G4 will be developed until the G6 (assuming Moto survives that long).
One thing this does is once and for all confirm that the pre MWSF02 G5 rumors at MOSR and The Reg were all false.
And hey Dorsal, what was that about G5 development going well? <img src="graemlins/smokin.gif" border="0" alt="[Chilling]" />
</strong><hr></blockquote>
I disagree... as I pointed out in another thread (causing you to sit in the corner for 10 minutes), the "G5" terminology is very loose. Apple could call its next machine the "G5" regardless of what Motorola (or whomever makes the thing) calls the processor that goes into it. TheRegister's rumours about the 7500 represent most of what I expect to see in the "G5", and that could easily be the chip that they were refering to last August through December. There were discussions at the time about Apple calling a new machine the "G5" even though the processor was still technically a G4... but if that G4 was the 7500 then it is a reasonable thing to do.
As Motorola likes to keep pointing out, the G4 is actually scaling pretty well now that they've got the G4+ architecture... there was just a "flat spot" where the 7400/7410 was stuck at <500 MHz. If they continue to make changes like that to the 7xxx family it could carry on quite respectably. Moto itself has said that 1.8 GHz, 166 MHz MPX are both coming and the 7500 rumours point to RapidIO and an on-chip memory controller. The only weak point then is the single FPU, but they could correct that too. The only thing really missing from the "G5" then is that its not 64-bit... but the POWER series is, so that's where Apple might change tracks (or, more likely, use both). I also have to wonder whether the G4's core could up upgraded to 64-bits. The difference in the programming model is actually pretty minor (widening the registers isn't hard, the internal pathways are already at least that wide for FPU/VPU, and if they are already lengthening the pipelines again...?).
Rumour sources who get their info from prototype hardware are always too optimistic about delivery times -- they don't have visibility to the full list of problems, supply issues, etc. I've experienced this phenomena myself (although I don't post any NDA stuff <img src="graemlins/oyvey.gif" border="0" alt="[No]" /> ).
[ 07-28-2002: Message edited by: Programmer ]</p>
<strong>and you know this because...</strong><hr></blockquote>
I'm only guessing, but ... he probably said it because of that rumor a while back that said Apple cancelled it's request for Mot's G5 because Mot wouldn't include certain instruction sets (??) ("apple pie") into the new chip. I also remember reading that there were multiple chip manufacturers answering Apple's request for the new chip.
(edit: spelling)
[ 07-28-2002: Message edited by: MacJedai ]
[ 07-28-2002: Message edited by: MacJedai ]</p>
If it's a major structural change. I'd agree. eg a modified G4 core into a Rio mobo. I think I'd accept that... However...
"TheRegister's rumours about the 7500 represent most of what I expect to see in the "G5", and that could easily be the chip that they were refering to last August through December. There were discussions at the time about Apple calling a new machine the "G5" even though the processor was still technically a G4... but if that G4 was the 7500 then it is a reasonable thing to do."
The fabled 7500. I always thought this WAS supposed to be the 'G5'. Stretched pipleine. More mhz. Better Mobo... Years ago...I read about this. And I waited...and waited...and I'm still waiting...
"As Motorola likes to keep pointing out, the G4 is actually scaling pretty well now that they've got the G4+ architecture... there was just a "flat spot" where the 7400/7410 was stuck at <500 MHz."
(In Homer Simpson angry voice..., 'Why I oughta...' choke, throttle Moto'...etc...)
"If they continue to make changes like that to the 7xxx family it could carry on quite respectably."
Fair enough. But the next gen' PCs will be getting mobo changes too? What else will be in the '7500'? Just Rio? Just a pipeline stretch? Which (as we saw with the original G4 vs the G4+)
means less work done per cycle. The benefit of the extra mhz won't be seen until the 'new' pipeline gets ramped up...ala G4+ ala Pentium 4? No?
"Moto itself has said that 1.8 GHz, 166 MHz MPX are both coming and the 7500 rumours point to RapidIO and an on-chip memory controller."
That sounds okay. BUT...(!)
"The only weak point then is the single FPU, but they could correct that too."
This is the one thing I fear most. Surely...in the name of sanity...SURELY (sounds angst ridden, eh?) 1 fpu only...would be ridiculous in light of all the other modifications? Surely...going to Rio would be the ideal time to beef up the fpu? I'm sure the 'Sledgehammer's fpu unit won't be standing still... Reputed, the next gen 'Hammer' can get an extra 30% per cycle. Based upon the available rumours... Rio and a pipeline stretch *only* (while good in context of stagnant
"The only thing really missing from the "G5" then is that its not 64-bit..."
Yep.
"but the POWER series is, so that's where Apple might change tracks (or, more likely, use both). I also have to wonder whether the G4's core could up upgraded to 64-bits. The difference in the programming model is actually pretty minor (widening the registers isn't hard, the internal pathways are already at least that wide for FPU/VPU, and if they are already lengthening the pipelines again...?)."
fpu. More... Doesn't the Athlon have 3 fpu units? Does about as much work as a G4 per clock?
Hmmm. I hope the 7500 does more per cycle regardless of Rio. For 64 bit. I hope Apple do an IBM escape route...
What about dual Core G4 on Rio? Now that would be juicy... What's the chance of Moto' sticking two of them on one chip? I heard about that for the G5 a while back too. Or is that their plan for the G6?
Y'know...I'd like Moto' and Apple to be more open about their cpu strat'...
Lemon Bon Bon
[ 07-28-2002: Message edited by: Lemon Bon Bon ]</p>
(multi-core and all, you know.)
My theory is that Apple wants to stick with the G4 for a little while, IF they boost the speed the FSB, add RapidIO, and add an on chip memory controller, they could justify changing the name, in terms of marketing, to the "G5"
This wouldn't be reflective of what Motorola called the chip, but who cares? The only really notable difference at that point would be the 32/64 bit gap, and none of the apps i run (FCP, DVDSP, PS7) are 64 bit compatable anyway, including Mac OS X.
I bet they release a screaming fast processor, crank up the front side bus, slap an on board memory controller onto the chip, speed up the AGP, and maybe speed up the PCI bus. To make the thing an extra sweet sell... They call it a G5
And who would really know the difference? Who would care, as long as we were the platform on top again. Any of you running 64 bit apps? Nope... I didn't think so..
I don't mind speculation but when people start treating it like fact : " now that we know this then that must be true " .
Something that does tend to lend credence to a report is like the new Powermac case pictures. When Apple has something removed then you may really have something.
Might as well get used to it. Apple has clamped down on the rumor machine tight. It would be nice to get something juicy but just because something appears on some minor web site doesn't make it true.
So really nobody knows except Apple and maybe only SJ.
I thought the G4 was 128-bit, the velocity engine processes data at that rate so why is it said to be only 32-bit. The FPU is 32-bit according to Apple though. Can somone explain this to me, I know technically what it is, i know its a measurement but I thought the G4 was 128 and OS X was 128 also, what am I missing. Sorry if this is really stupid
Fair point.
"And that 30% improvement in the Hammer apparently (I'm just going on what I've read about the Hammer in various articles) comes primarily from moving the memory controller on-chip."
So...if the 'G4' has a memory controller 'on-chip'...that should cancel out that advantage?
"The actual core of the beast is not hugely different from the Athlon's core."
Hmmm.
"If the same benefits show up when the G4 gets an on-chip memory controller and an 80% clock boost, then perhaps it doesn't actually need a second FPU."
Well, there are rumours of the 'Hammer' having a hard time scaling!
"Of course the competition is a moving target, so a second one would be awfully nice..."
Well, ah...yes. I thought the 'Hammer' was going to launch at 3 gig plus? Or...maybe...that will only be the 'XP' rating.
IF the G4 can get close to 2 gig by San Fran' on Rio and...squeak an extra fpu in there... Then...then it could be game on if the Hammer is stuck at 1.6 ghz with an xp rating of 3 gig 'Xp'.
Hmm. The current Athlon has been labouring up to 1.8 gig. If Apple/Moto can get 1.4 gig out on DDR...then that's fair 'catch up' for now...all things considered.
1.8 gig and on...for next Jan...I think I can see your point(S).
Still, I'd still like an extra fpu to keep the other company...yeesh. Three years and still only 1 fpu...
(Y'know, I was really excited when the original G4 came out...so much promise...((whimsy...)) I await the 7500...to deliver on that promise.)
Lemon Bon Bon
<strong>
I'm only guessing, but ... he probably said it because of that rumor a while back that said Apple cancelled it's request for Mot's G5 because Mot wouldn't include certain instruction sets (??) ("apple pie") into the new chip. I also remember reading that there were multiple chip manufacturers answering Apple's request for the new chip.
(edit: spelling)
[ 07-28-2002: Message edited by: MacJedai ]
[ 07-28-2002: Message edited by: MacJedai ]</strong><hr></blockquote>
Let's just say a little bird told me.
<strong>Just answer, don't flame:
I thought the G4 was 128-bit, the velocity engine processes data at that rate so why is it said to be only 32-bit. The FPU is 32-bit according to Apple though. Can somone explain this to me, I know technically what it is, i know its a measurement but I thought the G4 was 128 and OS X was 128 also, what am I missing. Sorry if this is really stupid</strong><hr></blockquote>
The velocity engine can handle 128 bit chunks of data but the proc still runs software compiled for 32bit. Therefore I assume that the VE should simply be considered a co processor of sorts that can process larger bits of data in an otherwise 32bit app.
<strong>Just answer, don't flame:
I thought the G4 was 128-bit, the velocity engine processes data at that rate so why is it said to be only 32-bit. The FPU is 32-bit according to Apple though. Can somone explain this to me, I know technically what it is, i know its a measurement but I thought the G4 was 128 and OS X was 128 also, what am I missing. Sorry if this is really stupid</strong><hr></blockquote>
The G4 is a 32 bit chip, and no you're not stupid for thinking its a 128 bit chip. In fact, the Velocity engine can process certain commands at 128 bits. Its a pretty amazing chip, acctually, and the full potential is only recently being explored - see FCP 3 and DVDSP both of which rely on the velocity engine to boost their performance to nothing short of amazing.
Because most code is 32 bit, however, the G4 has been built to process it in 128 bit chunks. That means it takes 4 32 bit instructions and processes them at once.
This is a good thing, because if the G4 were only capable of 128 bit instructions, then the 32 bit code would have to be run in emulation, and it could only process one at a time. by being a 32 bit processor with the ability to crunch 128 bit instructions, the G4 is extremely fast and flexible. My illustration of the situation may be crude, if you want more info go to Arstechnica.com, or <a href="http://www.apple.com/powermac/processor.html"; target="_blank">Here</a>.
And I'll save you a little more time and cut and paste from apple's site too :
The Velocity Engine
Behind the PowerPC G4?s phenomenal performance is its aptly named Velocity Engine. The Velocity Engine processes data in huge 128-bit chunks, instead of the smaller 32-bit or 64-bit chunks used in traditional processors (it?s the 128-bit vector processing technology used in scientific supercomputers ? except that we?ve added 162 new instructions to speed up computations). In addition, the PowerPC G4 can perform four (in some cases eight) 32-bit floating-point calculations in a single cycle ? two to four times faster than processors found in PCs.
I have no problem whatsoever continuing to call the chips G4. Why not simply continue to call them G4 7500? If Apple suddenly decides they need the marketing hype of releasing G5 machines without the true (my idea of true) G5 processor it will look like another case of smoke and mirrors. Just a lie, with credibility slipping farther and farther. I don't know, maybe I just have issues, but I've waited so long for the G5 that a compromise now would put me into a year-long funk.
[ 07-28-2002: Message edited by: rambo47 ]</p>
I wouldn't mind a significantly updated G4 labeled "G5". A different possessor deserves a different name. However, the "G" stands for generation and that kind of upgrade it's a true generation change. So, I have some mixed feelings on the subject. Hopefully, the actual G5 would be best.
Suggestions for such a CPU:
-nVidia Style -- G4 Ultra
-AMD Style -- G4-II
-Intel Style -- G6