Why are so many people suggesting that since we're only hearing about this new chip from IBM now, it means that we have 6 months to 2 years of waiting before it's available in quantity?
Is there any law that says that IBM MUST announce a new chip to the world long before it's being fabbed? IBM is ALREADY producing chips in their new plant. Who can say for certain that the chips being built there ARE NOT this 64-bit CPU?"
The specs of this upcoming chip look to almost everyone as if they were custom designed for Apple...to fit every need Jobs et al would demand from IBM to give them a persuasive reason to drop Motorola. Though some speculate IBM may have designed this on their own for their own use in their own servers, I think the specs are far too Apple-ish to believe that.
So if we can agree that this is the Motorola-killer Apple has been wishing for, and that Apple will most likely be the #1 customer for this chip, why can't we believe Apple was aware of this chip long ago, and involved in its design? And if we buy that, then why would it be impossible to believe that this chip is DONE, and currently in production?
Again...what law requires that IBM announce this BEFORE it's being produced? There is just as much possibility that it will be shipping BEFORE the October 13th microprocessor conference as there is that it won't be.
Some will say my logic goes against "what's been done before" regarding processor announcements. But what company is more likely to be involved in doing things "different" than the past than Apple...
</strong><hr></blockquote>
How do we predict anything? Because it fits with past patterns, known facts, and the likely probabilities. Talks at MDF are almost always about chips that are months away from availability, and Apple hasn't managed to convince IBM or Moto to keep them a secret in the past. This isn't Apple we're talking about, this is IBM. This chip doesn't fit into the rumours about the near-term future we've been seeing for a while now -- they imply that the PowerMac will be using the last Moto G4 version(s). If this chip is POWER4-based (very likely) then you should note that the POWER4 has only been available for about a year, which makes it a pretty tight timeline to come out with a derivative chip anytime soon.
And since when is this chip "clearly designed for Apple"?! Sure its design goals line up with Apple's needs, but that is because Apple needs a desktop processor! The only feature that could might be interpreted as Apple-specific is the SIMD implementation (assuming its VMX)... but even for that IBM was on the original design team and co-holds the patents for and it can be useful for IBM as well. In fact they've been talking up this kind of processor enhancement for the POWER5. This is a desktop processor that will be very useful to IBM's own lineup of machines, covering the low-end of the product range and useable in big high-end SMP machines.
While I agree with Programmmer and others that most of the evidence points to a minor speed bump Tuesday, with a switch to IBM coming at MWSF at the very earliest, there is one thing that points to an earlier release of the IBM chip: that leaked case design. That case seems to be designed for a hotter processor than anything we have seen from Motorola. My impression is that Motorola is not likely to come out with any chips that they cannot sell into the embedded market, and the embedded market doesn't like chips that require a "7 lb heat-sink." Unlike chip designs, I don't remember ever seeing leaked pictures of a case more than a couple weeks before the case was actually released.
<strong>While I agree with Programmmer and others that most of the evidence points to a minor speed bump Tuesday, with a switch to IBM coming at MWSF at the very earliest, there is one thing that points to an earlier release of the IBM chip: that leaked case design. That case seems to be designed for a hotter processor than anything we have seen from Motorola. My impression is that Motorola is not likely to come out with any chips that they cannot sell into the embedded market, and the embedded market doesn't like chips that require a "7 lb heat-sink." Unlike chip designs, I don't remember ever seeing leaked pictures of a case more than a couple weeks before the case was actually released.</strong><hr></blockquote>
If these new machines do anything to address the bandwidth issues across MPX then the rumoured high-end quad processor machine becomes plausible... and I'd expect to see that before I'd expect to see a completely new chip next week. Alternatively they are pushing the envelope on the processor clock rates (heat increases quickly with faster clocks). There are other possible explanations... like trying to use a bigger heat sink and quieter fans to reduce noise.
Well, there's quite a lot of possibilities on this topic. Here are my guesses.
Apple won't dump Motorola any time soon, even if they really, really want to on a personal level. It's not necessarily good business to limit your
self to a single source for some components. That's why IBM is still making G3s for Apple. Having a second source in your back pocket is a nice safety net.
Why not an expanded PowerMac matrix? Steve threw out the 4-product matrix a while ago with the Cube.
For Prosumers, we have the current G4 family from Motorola.
The rest get iMacs.
They'll probably need to keep the G4 in the Powerbooks, but what do I know.
If they go this route, they can drop Sahara from the iBook, and move it to the G4, because they'll be buying a high-end PPC from IBM. IBM would probably like this mix better, as they would increase their margins. (Well, this would be better for IBM if the volumes or margins are high enough.)
And then, right after all that work, Apple can switch to Intel, just like that <a href="http://news.com.com/2100-1001-948239.html"; target="_blank">Neff guy claims.</a>
- New IBM chip for a 'new' high-end desktop & XServe
- Motorola's G4 for PowerMac
- Motorola's G5 for PowerBook, iMac, eMac
- IBM's Sahara for iBook
Leonis,
This makes the most sence and I too think this is what we might see going forward... This is also the ONLY way I could see Motorola doing a new '74??' CPU that has HyperTransport (and DDR) support. Motorola would have never invested the time and money on adding such features to the G4 if Apple was going IBM (new chip) for 100% of it's 'pro' line. Lets face it MOT would have HAD to know such a thing was happening since most deals with suppliers are done via a signed contract and Apple would have had to explain WHY next years contract wasn't being signed or the number of chips ordered was quite a bit less than the year before.
The next (13th or 24th) PowerMacs will tell us ALOT about what will happen over the next year.
My guess:
If the new boxes DO support DDR and are HyperTransport then G4's will be available for at least a year.
If the new boxes DON'T then... This will be the last of the MOT G4's in Apple desktop boxes.
If the stories are true that IBM is just now wooing Apple, and the arguments here are true about a 64 bit PPC processor being a year or so away, I will be surprised, and my high respect for Steve Jobs as Apple's CEO will vanish. It would mean that he was not preparing to move away from Motorola, and not actively seeking an alternate vendor. For the past two years he has been putting up with a PPC vendor who isn't doing what it takes to be a leader, but just getting by. If this is all true, I had Steve Jobs pegged wrong, and I'll be very disappointed in that. But I don't expect to be disappointed. Likely, Jobs realized Motorola would never finish and deliver a G5 for Apple. He made a deal with IBM, and Apple worked closely with them in development. Apple then had to tread water with Motorola G4s until the G5 is ready for prime time in Macs.
Regarding when a G5 will ship, I never speculated on that, just that it will be announced before October 15 and likely much sooner. IBM has a lot to talk about at the Microprocessor Forum. They get into all the technical challenges of how to build something like this, and what the design trade-offs were. What we will find out from Apple is just a little more than we already know from the forum agenda. We will hear a lot about G5 performance for running applications, and what this means to the Macintosh's future.
The real speculation is when and how the G5 will be introduced. New PowerMac? Maybe not. It may first appear in a new high-end workstation-like Mac, which could be a rackmount with a server version too. It may be priced much higher, and scheduled to ship in three to six month. But I would not rule out shipping sooner. This is speculation. The G5 announcement is 85 percent sure in my mind, if Steve Jobs is who I think he is.
<strong>It would mean that he was not preparing to move away from Motorola, and not actively seeking an alternate vendor. For the past two years he has been putting up with a PPC vendor who isn't doing what it takes to be a leader, but just getting by.</strong><hr></blockquote>
And now to post a response (forgive the new guy):
The G4 has been out for three years and the speed-dump issue was an immediate problem. The rate at which the clock speed has been improving is appalling. So Moto has been letting Apple down for 3 years not 2.
"If the stories are true that IBM is just now wooing Apple, and the arguments here are true about a 64 bit PPC processor being a year or so away, "
Well there not "true",just idle speculation and perhaps a tinge of pessimism.No one here knows the status of that processor,the current state of the G4 or what is comming next for that matter.Apple needs a revolution but all we have had the last few years has been a dissapointment.
Is there a law it has to be 6-24 months? No. Does it make sense? Yes. Can it be less than 6? Yes, but it'd tick a lot of people off. Can it be more than 24? Yes, but it'd tick a lot of people off, so they'd have to deliver better than promised.</strong><hr></blockquote>
Typically microprocessor convention announcements are not product announcements, they are technology announcements. Yes, something like a microprocessor is a complex thing that has been in the works for some time even before IBM would announce the technology/market direction.
Yes, Apple has prototypes of said processors that they're playing with (but they also have all sorts of other technologies they are playing with that haven't been announced, and may never see the light of day as a product).
Yes, it is possible IBM will pull a shocker and say they are available for sampling immediately -- but that's different from full scale production, and that's different from someone (like Apple) already having a product built around them.
Certainly there are plenty of possibilities, but remember, forums like this (the microprocessor forum) are about presenting technologies and future direction. Things Intel, Motorola, and others have historically announce at said conferences to not see the light of day for some time after they are announced.
There is a big difference between what is possible, and what is likely. It is likely you won't see these chips in anything Apple produces for a year or so. I'd be glad to be wrong about that -- I'd be in line to buy one.
<strong>While I agree with Programmmer and others that most of the evidence points to a minor speed bump Tuesday, with a switch to IBM coming at MWSF at the very earliest, there is one thing that points to an earlier release of the IBM chip: that leaked case design. That case seems to be designed for a hotter processor than anything we have seen from Motorola. </strong><hr></blockquote>
From what I'm told, it is simply because the airflow within the case changed due to the internals being re-organized, so changes had to be made to increase the cooling capacity.
Of course, it is also possible that they are making the case for hotter processors as part of a forward-thinking design. But you'd logically think that when Apple introduces an entirely new CPU, they'd want the case to be distinctive and announce this to the world, too...
Comments
quote:
------------------------------------------------------------------------
Originally posted by Ensoniq:
Why are so many people suggesting that since we're only hearing about this new chip from IBM now, it means that we have 6 months to 2 years of waiting before it's available in quantity?
Is there any law that says that IBM MUST announce a new chip to the world long before it's being fabbed? IBM is ALREADY producing chips in their new plant. Who can say for certain that the chips being built there ARE NOT this 64-bit CPU?"
------------------------------------------------------------------------
<strong>
The specs of this upcoming chip look to almost everyone as if they were custom designed for Apple...to fit every need Jobs et al would demand from IBM to give them a persuasive reason to drop Motorola. Though some speculate IBM may have designed this on their own for their own use in their own servers, I think the specs are far too Apple-ish to believe that.
So if we can agree that this is the Motorola-killer Apple has been wishing for, and that Apple will most likely be the #1 customer for this chip, why can't we believe Apple was aware of this chip long ago, and involved in its design? And if we buy that, then why would it be impossible to believe that this chip is DONE, and currently in production?
Again...what law requires that IBM announce this BEFORE it's being produced? There is just as much possibility that it will be shipping BEFORE the October 13th microprocessor conference as there is that it won't be.
Some will say my logic goes against "what's been done before" regarding processor announcements. But what company is more likely to be involved in doing things "different" than the past than Apple...
</strong><hr></blockquote>
How do we predict anything? Because it fits with past patterns, known facts, and the likely probabilities. Talks at MDF are almost always about chips that are months away from availability, and Apple hasn't managed to convince IBM or Moto to keep them a secret in the past. This isn't Apple we're talking about, this is IBM. This chip doesn't fit into the rumours about the near-term future we've been seeing for a while now -- they imply that the PowerMac will be using the last Moto G4 version(s). If this chip is POWER4-based (very likely) then you should note that the POWER4 has only been available for about a year, which makes it a pretty tight timeline to come out with a derivative chip anytime soon.
And since when is this chip "clearly designed for Apple"?! Sure its design goals line up with Apple's needs, but that is because Apple needs a desktop processor! The only feature that could might be interpreted as Apple-specific is the SIMD implementation (assuming its VMX)... but even for that IBM was on the original design team and co-holds the patents for and it can be useful for IBM as well. In fact they've been talking up this kind of processor enhancement for the POWER5. This is a desktop processor that will be very useful to IBM's own lineup of machines, covering the low-end of the product range and useable in big high-end SMP machines.
[ 08-11-2002: Message edited by: Programmer ]</p>
<strong>
How do we predict anything? Because it fits with past patterns, known facts, and the likely probabilities.</strong><hr></blockquote>
Not on AI. Here we predict things from what we really, really want.
<strong>While I agree with Programmmer and others that most of the evidence points to a minor speed bump Tuesday, with a switch to IBM coming at MWSF at the very earliest, there is one thing that points to an earlier release of the IBM chip: that leaked case design. That case seems to be designed for a hotter processor than anything we have seen from Motorola. My impression is that Motorola is not likely to come out with any chips that they cannot sell into the embedded market, and the embedded market doesn't like chips that require a "7 lb heat-sink." Unlike chip designs, I don't remember ever seeing leaked pictures of a case more than a couple weeks before the case was actually released.</strong><hr></blockquote>
If these new machines do anything to address the bandwidth issues across MPX then the rumoured high-end quad processor machine becomes plausible... and I'd expect to see that before I'd expect to see a completely new chip next week. Alternatively they are pushing the envelope on the processor clock rates (heat increases quickly with faster clocks). There are other possible explanations... like trying to use a bigger heat sink and quieter fans to reduce noise.
yes. that came to me all well.
if tues = duallies then we get to start a new AI countdown. For the power4 mac.
Apple won't dump Motorola any time soon, even if they really, really want to on a personal level. It's not necessarily good business to limit your
self to a single source for some components. That's why IBM is still making G3s for Apple. Having a second source in your back pocket is a nice safety net.
Why not an expanded PowerMac matrix? Steve threw out the 4-product matrix a while ago with the Cube.
That could mean a high-end, IBM-based PPC used to <a href="http://www.apple.com/shake/"; target="_blank">Shake,</a> Rattle and <a href="http://www.emagic.de/english/news/index.html"; target="_blank">Roll.</a> This is for Pros.
For Prosumers, we have the current G4 family from Motorola.
The rest get iMacs.
They'll probably need to keep the G4 in the Powerbooks, but what do I know.
If they go this route, they can drop Sahara from the iBook, and move it to the G4, because they'll be buying a high-end PPC from IBM. IBM would probably like this mix better, as they would increase their margins. (Well, this would be better for IBM if the volumes or margins are high enough.)
And then, right after all that work, Apple can switch to Intel, just like that <a href="http://news.com.com/2100-1001-948239.html"; target="_blank">Neff guy claims.</a>
New IBM chip for a new,higher-end (need to mention more expensive?) desktop model , XServe
Motorola's 7470 & 7500 for PowerMac
Motorola's 7460 for PowerBook, iMac, eMac
IBM's Sahara for iBook
By doing this everyone will be happy
[ 08-11-2002: Message edited by: Leonis ]</p>
- Motorola's G4 for PowerMac
- Motorola's G5 for PowerBook, iMac, eMac
- IBM's Sahara for iBook
Leonis,
This makes the most sence and I too think this is what we might see going forward... This is also the ONLY way I could see Motorola doing a new '74??' CPU that has HyperTransport (and DDR) support. Motorola would have never invested the time and money on adding such features to the G4 if Apple was going IBM (new chip) for 100% of it's 'pro' line. Lets face it MOT would have HAD to know such a thing was happening since most deals with suppliers are done via a signed contract and Apple would have had to explain WHY next years contract wasn't being signed or the number of chips ordered was quite a bit less than the year before.
The next (13th or 24th) PowerMacs will tell us ALOT about what will happen over the next year.
My guess:
If the new boxes DO support DDR and are HyperTransport then G4's will be available for at least a year.
If the new boxes DON'T then... This will be the last of the MOT G4's in Apple desktop boxes.
Dave
[ 08-11-2002: Message edited by: DaveGee ]</p>
<strong>what processor and bus speed yields 6.45 GB/s bandwidth?</strong><hr></blockquote>
Dual Channel DDR. FSB = 166MHz.
what processor and bus speed yields 6.4 GB/s bandwidth?</strong>
Lots of ways:
1. 800 MHz 32 bit Hypertransport (800 MHz * 4 bytes * 2 DDR)
2. Dual channel 800 MHz 16 bit RapidIO (800 MHz * 2 bytes * 2 channels * 2 DDR)
3. Dual channel 400 MHz 32 bit Hypertransport.
4. QDR 64 bit 200 MHz processor bus (200 MHz * 8 bytes * 4 QDR)
For memory:
1. Quad channel PC800 Rambus (400 MHz * 2 bytes * 2 DDR * 4 channel)
2. Dual channel PC3200 DDR SDRAM (200 MHz * 8 bytes * 2 DDR * 2 channel)
MPC74xx = G4
MPC75xx = G5
IBM's new 64bit PPC = G6
G4 in the eMac, iBook and the new Cube2:-))
G5 in the xserve, Powerbook, iMac.
G6 in the Powermac.
[ 08-11-2002: Message edited by: Fat Freddy ]</p>
<strong>[qb]Originally posted by Bigc:
what processor and bus speed yields 6.4 GB/s bandwidth?</strong>
Lots of ways:
1. 800 MHz 32 bit Hypertransport (800 MHz * 4 bytes * 2 DDR)
2. Dual channel 800 MHz 16 bit RapidIO (800 MHz * 2 bytes * 2 channels * 2 DDR)
3. Dual channel 400 MHz 32 bit Hypertransport.
4. QDR 64 bit 200 MHz processor bus (200 MHz * 8 bytes * 4 QDR)
For memory:
1. Quad channel PC800 Rambus (400 MHz * 2 bytes * 2 DDR * 4 channel)
2. Dual channel PC3200 DDR SDRAM (200 MHz * 8 bytes * 2 DDR * 2 channel)[/QB]<hr></blockquote>
All sounds like this chip is down the road aways then, hey?
Not really.It supports a maximum bus of this bandwith but it doesnt have to use it if released now.
Regarding when a G5 will ship, I never speculated on that, just that it will be announced before October 15 and likely much sooner. IBM has a lot to talk about at the Microprocessor Forum. They get into all the technical challenges of how to build something like this, and what the design trade-offs were. What we will find out from Apple is just a little more than we already know from the forum agenda. We will hear a lot about G5 performance for running applications, and what this means to the Macintosh's future.
The real speculation is when and how the G5 will be introduced. New PowerMac? Maybe not. It may first appear in a new high-end workstation-like Mac, which could be a rackmount with a server version too. It may be priced much higher, and scheduled to ship in three to six month. But I would not rule out shipping sooner. This is speculation. The G5 announcement is 85 percent sure in my mind, if Steve Jobs is who I think he is.
<strong>It would mean that he was not preparing to move away from Motorola, and not actively seeking an alternate vendor. For the past two years he has been putting up with a PPC vendor who isn't doing what it takes to be a leader, but just getting by.</strong><hr></blockquote>
And now to post a response (forgive the new guy):
The G4 has been out for three years and the speed-dump issue was an immediate problem. The rate at which the clock speed has been improving is appalling. So Moto has been letting Apple down for 3 years not 2.
Well there not "true",just idle speculation and perhaps a tinge of pessimism.No one here knows the status of that processor,the current state of the G4 or what is comming next for that matter.Apple needs a revolution but all we have had the last few years has been a dissapointment.
<strong>
Really? Where's the Motorola G4.5 you promised?*
</strong><hr></blockquote>
It is already in current shipping Macs.
<strong>
Is there a law it has to be 6-24 months? No. Does it make sense? Yes. Can it be less than 6? Yes, but it'd tick a lot of people off. Can it be more than 24? Yes, but it'd tick a lot of people off, so they'd have to deliver better than promised.</strong><hr></blockquote>
Typically microprocessor convention announcements are not product announcements, they are technology announcements. Yes, something like a microprocessor is a complex thing that has been in the works for some time even before IBM would announce the technology/market direction.
Yes, Apple has prototypes of said processors that they're playing with (but they also have all sorts of other technologies they are playing with that haven't been announced, and may never see the light of day as a product).
Yes, it is possible IBM will pull a shocker and say they are available for sampling immediately -- but that's different from full scale production, and that's different from someone (like Apple) already having a product built around them.
Certainly there are plenty of possibilities, but remember, forums like this (the microprocessor forum) are about presenting technologies and future direction. Things Intel, Motorola, and others have historically announce at said conferences to not see the light of day for some time after they are announced.
There is a big difference between what is possible, and what is likely. It is likely you won't see these chips in anything Apple produces for a year or so. I'd be glad to be wrong about that -- I'd be in line to buy one.
<strong>While I agree with Programmmer and others that most of the evidence points to a minor speed bump Tuesday, with a switch to IBM coming at MWSF at the very earliest, there is one thing that points to an earlier release of the IBM chip: that leaked case design. That case seems to be designed for a hotter processor than anything we have seen from Motorola. </strong><hr></blockquote>
From what I'm told, it is simply because the airflow within the case changed due to the internals being re-organized, so changes had to be made to increase the cooling capacity.
Of course, it is also possible that they are making the case for hotter processors as part of a forward-thinking design. But you'd logically think that when Apple introduces an entirely new CPU, they'd want the case to be distinctive and announce this to the world, too...
hmmmm....