Remember, the G4 as released was not the G4 that was on paper up until that point. It was missing certain features and was ultimately released as a crippled chip. If Motorola were crippling the G5 in a similar fashion, they could very well be near releasing a working chip.
When was the G4 released? January 2000 I think? Well, if Motorola just continued work on the "real" G4 for the past two years, by now they would have a G4 chip that could greatly surpass the G4 they released, and just call it a G5. I'm going to guess that the G5 as released WHEN it's released, will not be the chip that was outlined 3-4 years ago.
So a G5 in January is really just what a G4 should have been two years ago. No big deal if it's really faster in all respects.
Remember guys, we are now pretty darn sure Apple has become directly involved in development of the PowerPC. I am convinced that this is the reason we saw no increase for 18 months, then a mhz jump, followed by a modest jump later.
My theory is that jobs really DID tell MOT that if they didn't get their respective sh$t together than he was going to A) Dump them and convince IBM to make the things (legal or not) and Sue the hell out of them.
I think that with Apple and Job's Reality Distortion Field Energy behind development we really will see the G5 at Macworld.
My theory is that jobs really DID tell MOT that if they didn't get their respective sh$t together than he was going to A) Dump them and convince IBM to make the things (legal or not) and Sue the hell out of them.
</strong><hr></blockquote>
Moto's never had a real problem with designing nice chips. Their problem is they can't produce those chips in their fabs. They've been farming more and more of their high end products out to TSMC and other foundries because frankly their fabs are dirty and their engineers have some issues. Add to that the fact that they are bleeding money and cutting costs everywhere doesn't lend itself to the most productive MPU production.
And the AlmightyBabarahm is correct, those SPEC scores are BS. Most of that MOSR article on the G5 is just laughable. My favorite quote is:
[quote]The 32 bit version of the G5 will be solely targeted towards embedded applications, as 32-bit addressing is no longer adequate for desktop applications.<hr></blockquote>
That quote about 32bit addressing is laughable considering the 7450 has 36bit addressing (up to 64GB of real memory addressing) which is adequate. Why would the G5 downgrade in features? Doesn't make sense.
There are no indications that the G5 will be multicore in it's first generation but it can be very mulit-processor friendly. If you know something about the POWER4 it's that it was designed to operate with other processors at peak effeciency. You make the processors face each other in a certain way to create the shortest traces to maximum performance just the way IBM designed it. The G5 might take that into effect and also improve chip to chip performance to the point where the performance of 2 chips would approach the twice of what a single chip could do. They would need to implement a enormously fast bus though to make it happen. 500MHz SDR(for lowest possible latency) would be a good start and you would have to position the chips as close as physically possible.
If Apple had at least a 50% input on the principal design of the G5 then they would have included features on the chip primarily useful for desptop applications (workstations). They would not want to lag in MHz so a 10-14 stage pipeline is a must. L3 cache is getting more expensive as the core processor speed goes up so that would be on the chopping block, but to compensate you would need to bump up the L1/L2 cache accordingly. You might want to keep the L1 at 32/32KB or increase it to maybe 64KB data and 32KB instruction but the L2 would have to be bumped to 512KB or more if possible. A more precise branch predictor and maybe some advanced out of order excecution features. The ability to use either 64 or 32bit data and instructions on the fly (you would be able to have the OS, some 32bit apps, a few 64bit apps all running at the same time). A fast, P4 crushing, system bus. I suggest a 200MHz DDR (400MHz)bus to keep it simple and low latency running at 128 bit to processor from the main controller.
And lastly, memory controller on die! This would keep the processor feed and not starving like it is now! Especially for Altivec. Advantages of an on die mem controller would be the ability to impose a hugely fat bus to the processor (256bit?) at core speeds (either half (600-800MHz) or full core speed). Disadvantages would be bigger die (but it would eliminate the need for an external mem controller) and possibly more pin out (about 200 devoted to memory for DDR-SDRAM.
[quote] Moto's never had a real problem with designing nice chips. Their problem is they can't produce those chips in their fabs. They've been farming more and more of their high end products out to TSMC and other foundries because frankly their fabs are dirty and their engineers have some issues. Add to that the fact that they are bleeding money and cutting costs everywhere doesn't lend itself to the most productive MPU production. <hr></blockquote>
Look, I tend to think your reasoning is right, but the point is there was no clock speed increase in almost two years. Now, you and I both know that the PowerPC architecture is superior to most, but a 1.2 ghz deficit in clock speed is a bit tough to make up. From a marketing standpoint, it is a nightmare. Apple has got to know this.
Also, MOT is the supplier for Apple's CPU's (G4's). They are ultimately responsible for production, even if they outsource it. If I were Jobs I would have also "thrown the phone across the room". I truly believe they lied to Apple and screwed them HUGE. IMO, The only reason they didn't drop MOT is that there was no truly viable better option available.
This is why Apple has become intimately involved with the development of the PPC. Is there really anyone out there that DOESN'T believe Jobs told them something along the lines of "get your sh*t together or lose ALL of our business and face a lawsuit for making false performance predictions"?
It's just a bunch of rumors. Anyone with any licence to say anything is most likely constrained by NDA's. Anyone else is just speculating or passing on third hand information.
All in all, If I had to make my own educated guess on who's predictions are right, I'd say Powerdoc's more on the right track.
Will the G5 be faster than the G4, yes.
1.2-1.6 Gighz, most probably, but you won't see it in a Powermac in January.
Will it have that kind of IPC that makes it competitive with a Power4?
Highly Unlikely. it takes more than just strapping on extra execution units for that kind of performance. You need better BP, more reorder resources, Probably a recompile on existing code to maximize new capabilities, basically large improvements accross the board in everything to get that kind of IPC boost while also virtually doubling clock-speed.
And I don't recall Motorola having a history of being able to bring something like this to table in this manner.</strong><hr></blockquote>
Thanks JfW, you have answer in a much better way than i can do.
For the benchmarks it's funny to imagine that a G5 can beat in FP a monster such as a power 4
This is why Apple has become intimately involved with the development of the PPC. Is there really anyone out there that DOESN'T believe Jobs told them something along the lines of "get your sh*t together or lose ALL of our business and face a lawsuit for making false performance predictions"?
[QB]<hr></blockquote>
I agree with what you are saying. I just don't feel the problem neccessarily lies with Motorola's inability to design a processor for Apple. It lies in their ability to effectively manufacture that processor up to the performance level it was designed for. Apple has no expertise in this area and thus has nothing to offer Motorola. If Motorola is forced to outsource to IBM for production (who happens to be the most expensive foundry in the world) they are losing nearly all profits from these chips. Therefore they lose any incentive to contribute to PowerPC for desktop processors since there isn't any money in it for them.
I'm sure some of Moto's 500+ VPs are screaming their heads off at their fab managers to figure out how to make high end chips who are yelling at their engineers to run over to AMD and beg, steal, bribe solutions from them.
Yep, I think people are finally taking NDA's seriously now, even if it's just with Apple. People are thinking twice before opening up their mouths becuase they value their job. Especially nowadays where it's so easy for your employer to monitor your phone and network activity. I know some places that scan email for key words and tags them for further inspection.
If you are lucky enough to be on the hardware seeding list, would you risk it just to spew forth information to a bunch like us? Maybe on a minor hardware update, but certainly not on something that might be as revolutionary as the move from 9 to X is.
[ 11-16-2001: Message edited by: Hi Ho Quicksilver ]</p>
[quote]Although, I don't see Apple raising prices in this economy, do you?<hr></blockquote>
That 's just rediculous. If I understand how the new HiP fab works, there should be much higher yields thus making the chips cheaper to manufacture and sell.
The only reason Apple would have for raising prices is to pad the preofit margin even more. <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
Well, the Apple store just switched to euro's here, and prices have gone up slightly.
Furthermore, on a larger scale, I'm pretty certain that Appple powermac prices as a whole have actually gone up, for instance comparing the current line-up versus the yosemite line-up, in price. The top of the line dual 800 is over 5000 dollars here. The top of the line yosemite 400 never was that expensive. Sure now you get more for the money, but it's almost three years later so....
yep, too expensive in all but the ibook line-up. You can perhaps make an argument for the mid-range Ti, and only if you really absolutely need it (the low end tower).
Apple needs to keep the current monitor promotion permanently if they want to begin approaching more respectable pricing.
Can we stick to US dollars here? I'm not devaluing (no pun intended) your currency or anything, but I have to admit my conversions are a ittle rusty.
Anyway, Mr. StarfleetX, are you agreeing with me or saying I am ridiculous? I can't really tell, though I think you are agreeing that raising prices would be insane.
As far as the PowerMac prices go, it doesn't seem they have gone up.....i don't really remember the G3's pricing but the only thing that has gone up is the entry level G4 as far as I know.....as far as the pro series is concerned.
Comments
[QB][/QB]<hr></blockquote>
So what's that mean PM G4? Are you going to say something?
When was the G4 released? January 2000 I think? Well, if Motorola just continued work on the "real" G4 for the past two years, by now they would have a G4 chip that could greatly surpass the G4 they released, and just call it a G5. I'm going to guess that the G5 as released WHEN it's released, will not be the chip that was outlined 3-4 years ago.
So a G5 in January is really just what a G4 should have been two years ago. No big deal if it's really faster in all respects.
but... no comments so far?
btw: scheisskopf, nice nick you have.... heh
My theory is that jobs really DID tell MOT that if they didn't get their respective sh$t together than he was going to A) Dump them and convince IBM to make the things (legal or not) and
I think that with Apple and Job's Reality Distortion Field Energy behind development we really will see the G5 at Macworld.
[ 11-15-2001: Message edited by: SDW2001 ]</p>
<strong>
My theory is that jobs really DID tell MOT that if they didn't get their respective sh$t together than he was going to A) Dump them and convince IBM to make the things (legal or not) and
</strong><hr></blockquote>
Moto's never had a real problem with designing nice chips. Their problem is they can't produce those chips in their fabs. They've been farming more and more of their high end products out to TSMC and other foundries because frankly their fabs are dirty and their engineers have some issues. Add to that the fact that they are bleeding money and cutting costs everywhere doesn't lend itself to the most productive MPU production.
And the AlmightyBabarahm is correct, those SPEC scores are BS. Most of that MOSR article on the G5 is just laughable. My favorite quote is:
[quote]The 32 bit version of the G5 will be solely targeted towards embedded applications, as 32-bit addressing is no longer adequate for desktop applications.<hr></blockquote>
<strong>hey... PM G4 is back!!!!! excellent, excellent.
but... no comments so far?
btw: scheisskopf, nice nick you have.... heh</strong><hr></blockquote>
What is that, scheiss?
"Neue Rechtschreibreform" or what?
<strong>
What is that, scheiss?
"Neue Rechtschreibreform" or what?</strong><hr></blockquote>
It's "Neuschreib"!
That quote about 32bit addressing is laughable considering the 7450 has 36bit addressing (up to 64GB of real memory addressing) which is adequate. Why would the G5 downgrade in features? Doesn't make sense.
There are no indications that the G5 will be multicore in it's first generation but it can be very mulit-processor friendly. If you know something about the POWER4 it's that it was designed to operate with other processors at peak effeciency. You make the processors face each other in a certain way to create the shortest traces to maximum performance just the way IBM designed it. The G5 might take that into effect and also improve chip to chip performance to the point where the performance of 2 chips would approach the twice of what a single chip could do. They would need to implement a enormously fast bus though to make it happen. 500MHz SDR(for lowest possible latency) would be a good start and you would have to position the chips as close as physically possible.
If Apple had at least a 50% input on the principal design of the G5 then they would have included features on the chip primarily useful for desptop applications (workstations). They would not want to lag in MHz so a 10-14 stage pipeline is a must. L3 cache is getting more expensive as the core processor speed goes up so that would be on the chopping block, but to compensate you would need to bump up the L1/L2 cache accordingly. You might want to keep the L1 at 32/32KB or increase it to maybe 64KB data and 32KB instruction but the L2 would have to be bumped to 512KB or more if possible. A more precise branch predictor and maybe some advanced out of order excecution features. The ability to use either 64 or 32bit data and instructions on the fly (you would be able to have the OS, some 32bit apps, a few 64bit apps all running at the same time). A fast, P4 crushing, system bus. I suggest a 200MHz DDR (400MHz)bus to keep it simple and low latency running at 128 bit to processor from the main controller.
And lastly, memory controller on die! This would keep the processor feed and not starving like it is now! Especially for Altivec. Advantages of an on die mem controller would be the ability to impose a hugely fat bus to the processor (256bit?) at core speeds (either half (600-800MHz) or full core speed). Disadvantages would be bigger die (but it would eliminate the need for an external mem controller) and possibly more pin out (about 200 devoted to memory for DDR-SDRAM.
Look, I tend to think your reasoning is right, but the point is there was no clock speed increase in almost two years. Now, you and I both know that the PowerPC architecture is superior to most, but a 1.2 ghz deficit in clock speed is a bit tough to make up. From a marketing standpoint, it is a nightmare. Apple has got to know this.
Also, MOT is the supplier for Apple's CPU's (G4's). They are ultimately responsible for production, even if they outsource it. If I were Jobs I would have also "thrown the phone across the room". I truly believe they lied to Apple and screwed them HUGE. IMO, The only reason they didn't drop MOT is that there was no truly viable better option available.
This is why Apple has become intimately involved with the development of the PPC. Is there really anyone out there that DOESN'T believe Jobs told them something along the lines of "get your sh*t together or lose ALL of our business and face a lawsuit for making false performance predictions"?
I thinks not.
[ 11-16-2001: Message edited by: SDW2001 ]</p>
<strong>
How can anyone really comment on it?
It's just a bunch of rumors. Anyone with any licence to say anything is most likely constrained by NDA's. Anyone else is just speculating or passing on third hand information.
All in all, If I had to make my own educated guess on who's predictions are right, I'd say Powerdoc's more on the right track.
Will the G5 be faster than the G4, yes.
1.2-1.6 Gighz, most probably, but you won't see it in a Powermac in January.
Will it have that kind of IPC that makes it competitive with a Power4?
Highly Unlikely. it takes more than just strapping on extra execution units for that kind of performance. You need better BP, more reorder resources, Probably a recompile on existing code to maximize new capabilities, basically large improvements accross the board in everything to get that kind of IPC boost while also virtually doubling clock-speed.
And I don't recall Motorola having a history of being able to bring something like this to table in this manner.</strong><hr></blockquote>
Thanks JfW, you have answer in a much better way than i can do.
For the benchmarks it's funny to imagine that a G5 can beat in FP a monster such as a power 4
[QB]
This is why Apple has become intimately involved with the development of the PPC. Is there really anyone out there that DOESN'T believe Jobs told them something along the lines of "get your sh*t together or lose ALL of our business and face a lawsuit for making false performance predictions"?
[QB]<hr></blockquote>
I agree with what you are saying. I just don't feel the problem neccessarily lies with Motorola's inability to design a processor for Apple. It lies in their ability to effectively manufacture that processor up to the performance level it was designed for. Apple has no expertise in this area and thus has nothing to offer Motorola. If Motorola is forced to outsource to IBM for production (who happens to be the most expensive foundry in the world) they are losing nearly all profits from these chips. Therefore they lose any incentive to contribute to PowerPC for desktop processors since there isn't any money in it for them.
I'm sure some of Moto's 500+ VPs are screaming their heads off at their fab managers to figure out how to make high end chips who are yelling at their engineers to run over to AMD and beg, steal, bribe solutions from them.
[ 11-16-2001: Message edited by: Hi Ho Quicksilver ]</p>
[quote] Talk is that the low end G5 model will sell for slightly more than the current 867Mhz G4. <hr></blockquote>
Wow...so I guess the high end would be like the dual 800....or more.
Although, I don't see Apple raising prices in this economy, do you?
[ 11-16-2001: Message edited by: SDW2001 ]</p>
That 's just rediculous. If I understand how the new HiP fab works, there should be much higher yields thus making the chips cheaper to manufacture and sell.
The only reason Apple would have for raising prices is to pad the preofit margin even more. <img src="graemlins/bugeye.gif" border="0" alt="[Skeptical]" />
<strong>From Rumors......
Wow...so I guess the high end would be like the dual 800....or more.
Although, I don't see Apple raising prices in this economy, do you?
[ 11-16-2001: Message edited by: SDW2001 ]</strong><hr></blockquote>
Well, the Apple store just switched to euro's here, and prices have gone up slightly.
Furthermore, on a larger scale, I'm pretty certain that Appple powermac prices as a whole have actually gone up, for instance comparing the current line-up versus the yosemite line-up, in price. The top of the line dual 800 is over 5000 dollars here. The top of the line yosemite 400 never was that expensive. Sure now you get more for the money, but it's almost three years later so....
Apple needs to keep the current monitor promotion permanently if they want to begin approaching more respectable pricing.
Anyway, Mr. StarfleetX, are you agreeing with me or saying I am ridiculous? I can't really tell, though I think you are agreeing that raising prices would be insane.
As far as the PowerMac prices go, it doesn't seem they have gone up.....i don't really remember the G3's pricing but the only thing that has gone up is the entry level G4 as far as I know.....as far as the pro series is concerned.
Any other thoughts?