<strong>Whilst it looks like the G4 replacement is going to be the Power4.
The most likely scenario is that the next revision to the PowerMac range will be a G4 with FSB support for DDR.
Do we know wether the MPC7470 actually exists or is it a product of unfounded rumours?</strong><hr></blockquote>
I have to say that it is the product of unfounded rumours -- somebody thought it would "make sense" that Motorola would go the same route as Intel & AMD. There is a big difference, however... Intel & AMD build their processors for the desktop market and they don't have to share their FSB with anything except the chipset. Motorola, on the other hand, sells most of the G4s to embedded customers who like MPX the way it is. A key Moto rep has publicly stated that MPX will not go DDR for this reason, but it would go to 166 MHz. This is what happened. While I suppose it is remotely possible that plans changed or he was lying it is far more likely that we will never see a DDR version of MPX. The rumours of a G4 with a RapidIO bus are more believable although again they are unfounded.
Remember a few days ago we saw the OFFICIAL info on GCC on Linux on PPC64 with AltiVec? This is a real deal .
Also.....I think the guy on the email is only talking about Power 4. But IBM is going to talk about a 64bit PPC chip in October. AFAIK Power 4 and PPC aren't totally identical......So when that guys said no Power4 on Mac is true. But new 64bit PPC on new Mac also is true......
A chief IBM engineering manager (allegedly) said: Well, it sounds like this CPU is not in Apple's
future .
Bad news, but it there are some questions:
Why is the vector unit i) incompatible with the PPC SIMD standard VMX/Altivec/Viscosity Engine*, which is ii) used by Apple (a significant potential customer, which would subsidise chip production with economies of scale)? It's not as if Altivec is seriously flawed and I doubt it's tied to MPX.
Apple is more than a customer, they're part of the AIM PowerPC group. Apple management would be privy to this since its inception, as it is very relevant to Apple's business.
What are the other alternatives that Apple could realistically use, considering their investment in Altivec?
G4++(RIO)? (mythical_beastie) G5?
As pointed out by xlr8yourmac folks, saying "Yes: in Macs in Q2 2003 and it's gonna totally ro><or" would be a quick way to unemployment, NDA violation penalties and angering Mr. Jobs (see ATi annoucement from a while back).
<optimism>Apple don't want to be upstaged </optimism>
Guess we'll find out from the Microprocessor Forum in October. What's the date?
Devils Advocate- Apple has asked IBM to throw out walls and window shades to keep all the secrets in house. How dumb would Apple (Steve) look if IBM came out announcing a chip that was so obviously intentded for Apple to use? Who would by anything if in October a 1.8ghz 64 bit chip with DDR and uses Velocity Engine was announced?
Steve likes surprises, and with the latest word being this chip from IBM being Apple's next gen when it's at least 6-9 months off, I'm sure he's putting a cap on the chatter. Oh, and a few denials always helps.
First, these are public companies we are talking about. The SEC does not look kindly on people outright lying about business deals. The IBM guy could try to hide the truth but not outright lie.
That said, I do find the gcc patch on IBM's website more convincing than this second hand rumor. We don't really know how this conversation went. It is possible that Hunter thought they were talking about the Power4 derivative but that the IBM guy thought they were talking about the Power4 itself. Misunderstandings like that happen all the time.
mm, so you are saying that in deed Apple will use this chip from IBM that IBM couldn't say that Apple isn't interested? Doesn't sound illegal to me especially Apple can change it's mind later and decide to get the chip.
<strong>mm, so you are saying that in deed Apple will use this chip from IBM that IBM couldn't say that Apple isn't interested? Doesn't sound illegal to me especially Apple can change it's mind later and decide to get the chip.</strong><hr></blockquote>
"and there are technical issues that would
prevent AltiVec from ever marrying with Power4 or its successors."
And then, about a month later, IBM presents a Power4 derivative with AltiVec instructions? And now of course, Apple's suddenly interested? Still doesn't sound like a lie to you?
Why is the vector unit i) incompatible with the PPC SIMD standard VMX/Altivec/Viscosity Engine*, which is ii) used by Apple (a significant potential customer, which would subsidise chip production with economies of scale)?<hr></blockquote>
It was sounding like the 64-bit ppc IBM is working on ties two 64 bit units together to do the 128 bit operations of AltiVec/VMX... which is fine for most of the operations.
There's only one (or two) ops that are particularly tough to do without separate registers etc - 'permute' comes to mind immediately. Permute is exceedingly cool -> I can see why a 160 instruction 'VMX' implementation wouldn't do it for Apple. Would it still be a very useful SIMD implementation? Sure! For _new_ code. But emulating permute just looks exceedingly painful - and wedging it into a chip that doesn't have a completely separate altivec unit could also be exceedingly painful.
Having Apple bump volume would probably be very nice - but IBM has plenty of reasons to develop & manufacture a successor chip to the 604 anyway.
The fact of the matter is, this is a second hand rumor that denies completely the presence of Altivec on the new IBM chip. We have first hand confirmation, on the other hand, that there is Altivec on IBM's new 64 bit chip. I'm inclined to believe what has already been presented publicly, rather than a rumored, private conversation between unknown parties. We know IBM has Altivec on its chip, and there's no really plausible reason to have Altivec unless the chip is going to Apple. Until I hear otherwise, this new rumor hasn't changed my thoughts on the subject.
[quote]Remember a few days ago we saw the OFFICIAL info on GCC on Linux on PPC64 with AltiVec?<hr></blockquote>
Correct me if I'm wrong, but didn't that patch simply add the proper AltiVec flags to correctly indicate the the 64-bit PPC in RS/6000 servers does NOT support AltiVec?
IOW, AltiVec support was added to gcc, along with a new flag for indicating which CPUs support AltiVec, but they forgot to add a value for that flag for teh RS/6000. So there was a patch to set a value for that flag, but the value indicated that it does NOT support AltiVec.
That's what I remember seeing in that diff. Is there something I missed?
prevent AltiVec from ever marrying with Power4 or its successors."
And then, about a month later, IBM presents a Power4 derivative with AltiVec instructions? And now of course, Apple's suddenly interested? Still doesn't sound like a lie to you?
123</strong><hr></blockquote>
No, not a lie if in deed something needs to be added or changed in order to meet Apple's specs.
The reason i find this rumor hard to believe is because i was told by someone that Apple would be using a power4 core by this time next year. A lot of info has surfaced recently to substantiate this until this article. That's why I'm more inclined to believe this is Steve's way of shrouding his next gen secret.
Well, I was extremely downcast after I read that 'rumour'.
It just cannot be true. I hope it aint.
Apple limps on with a G4 at 1.4 - 1.8 gig on the currently crippled DDR and BUS? Let's say we get that next April if we're lucky.
What?! IN THE FACE OF 3 GIG PLUS PENTIUM 4 AND A 64 BIT HAMMER WITH AN XP RATING OF 3.4 GIG?!!
No way. Absolutenly no way. No way. NO WAY will Apple be able to prop up 'power'Mac sales in the face of that.
Deluded or not, I refuse to believe that Apple will have no 64 bit answer to within 9 months of the Hammer. It's inconceivable that answer will try to compete with that.
A G4 on .9 from Moto with DDR memory controller.
Maybe. But excuse while I feel a little skeptical. Track record says Moto' will stay behind...and Apple will keep on charging full whack for out of date, insubstantial improvements.
If Apple thinks they are immune from the power and price competition from x86. They're making a foolish mistake. Dell are giving their education share a damn hammering. This IS one of Apple's key markets.
The only way to counter that is to go closer on price and performance.
In education they aint cheap enough.
In workstation land, they aren't powerful enough.
And they are damn expensive. They're gonna keep seeing those 'power'Mac sales shrink until they can deliver something that justifies the price.
The G4 is a flop. A disaster. It has failed to keep abreast of competing developments. Plain and simple.
Photoshop, Lightwave, Premiere...web pages...all supposed Apple strengths...G4 gets its ass kicked.
I recently update to Windows Xp. It aint 'silk'. But 'cotton' gets the job done. And it IS less painful to use.
I've yet to play with Jag'. But I'd like to be able to get a tower that doesn't limp along in Painter 7, that blows the doors off 300 dpi A4 Photoshop images. The dual Gig is still £2,000 etc. Paying 2K for time and a half performance. What? 1.5 gig? If you're lucky. It's a marketing con that is as transparent as the Cube's skirt.
It looks good on the surface but masks the fact that Apple are embarrassingly behind...
Yeesh. We're no further on nine months later than Jan 2002. What the hell is happening.
Apple have got to get serious about its 'power'Mac pro users. The latest round smacked of desperation.
They neglect their core base at their peril. 'power'Mac sales are still smarting and the apologists around here put it down to cycles.
Heck, I'll throw my hat in and put it down to the recession not a consumer chip outstaying its welcome in the workstation arena or hack served mobo.
I don't know why Apple just don't come out and say where the hell they are going with their towers.
I could at least plan my purchase accordingly.
Say what you like about x86 owners. At least they are told up front what's coming.
Yeesh.
Silver lining? Well, if Apple can turn a chip like 8 way superscalar away...they must have something dazzling up their sleeve.
Comments
The most likely scenario is that the next revision to the PowerMac range will be a G4 with FSB support for DDR.
Do we know wether the MPC7470 actually exists or is it a product of unfounded rumours?
<strong>Whilst it looks like the G4 replacement is going to be the Power4.
The most likely scenario is that the next revision to the PowerMac range will be a G4 with FSB support for DDR.
Do we know wether the MPC7470 actually exists or is it a product of unfounded rumours?</strong><hr></blockquote>
I have to say that it is the product of unfounded rumours -- somebody thought it would "make sense" that Motorola would go the same route as Intel & AMD. There is a big difference, however... Intel & AMD build their processors for the desktop market and they don't have to share their FSB with anything except the chipset. Motorola, on the other hand, sells most of the G4s to embedded customers who like MPX the way it is. A key Moto rep has publicly stated that MPX will not go DDR for this reason, but it would go to 166 MHz. This is what happened. While I suppose it is remotely possible that plans changed or he was lying it is far more likely that we will never see a DDR version of MPX. The rumours of a G4 with a RapidIO bus are more believable although again they are unfounded.
<strong>Whilst it looks like the G4 replacement is going to be the Power4.
The most likely scenario is that the next revision to the PowerMac range will be a G4 with FSB support for DDR.
Do we know wether the MPC7470 actually exists or is it a product of unfounded rumours?</strong><hr></blockquote>
From what I heard a few months ago, we will have the G4+ or whatever in January and the power4-core by summer. Course, we'll have to see how it goes.
[quote]artcat \t
posted September 06, 2002 07:46 Â*
<a href="http://bbs.xlr8yourmac.com/ubb/Forum26/HTML/000108.html"; target="_blank">http://bbs.xlr8yourmac.com/ubb/Forum26/HTML/000108.html</a>;
(depressing info if true)"<hr></blockquote>
"depressing info if true" is a grand understatement
[ 09-06-2002: Message edited by: rickag ]</p>
<strong>Just found this link posted on ArsTechnica
"depressing info if true" is a grand understatement
[ 09-06-2002: Message edited by: rickag ]</strong><hr></blockquote>
Yeah, I just saw that over at macnn
Definitely discouraging
<img src="graemlins/oyvey.gif" border="0" alt="[No]" />
First this is a rumor
Remember a few days ago we saw the OFFICIAL info on GCC on Linux on PPC64 with AltiVec? This is a real deal .
Also.....I think the guy on the email is only talking about Power 4. But IBM is going to talk about a 64bit PPC chip in October. AFAIK Power 4 and PPC aren't totally identical......So when that guys said no Power4 on Mac is true. But new 64bit PPC on new Mac also is true......
Now back to real life
[ 09-06-2002: Message edited by: Leonis ]</p>
future .
Bad news, but it there are some questions:
- Why is the vector unit i) incompatible with the PPC SIMD standard VMX/Altivec/Viscosity Engine*, which is ii) used by Apple (a significant potential customer, which would subsidise chip production with economies of scale)? It's not as if Altivec is seriously flawed and I doubt it's tied to MPX.
- Apple is more than a customer, they're part of the AIM PowerPC group. Apple management would be privy to this since its inception, as it is very relevant to Apple's business.
- What are the other alternatives that Apple could realistically use, considering their investment in Altivec?
- As pointed out by xlr8yourmac folks, saying "Yes: in Macs in Q2 2003 and it's gonna totally ro><or" would be a quick way to unemployment, NDA violation penalties and angering Mr. Jobs (see ATi annoucement from a while back).
- <optimism>Apple don't want to be upstaged
</optimism>
Guess we'll find out from the Microprocessor Forum in October. What's the date?G4++(RIO)? (mythical_beastie) G5?
*not really
<strong>Time to jump in
Also.....I think the guy on the email is only talking about Power 4.
Now back to real life
</strong><hr></blockquote>
[quote]
From: Craig Hunter <c.a.hunter who is at larc dot nasa dot gov>
Date: Thu Sep 5, 2002 5:30:18 PM US/Eastern
To: <[email protected]>, <[email protected]>
Subject: IBM Power4
Folks,
....After the briefing, I asked the
presenter (a chief engineering manager from IBM) about the Power4 derivative
for desktops and low end servers to be announced in October....... Well, it sounds like this CPU is not in Apple's
future --.......would
prevent AltiVec from ever marrying with Power4 or its successors.
Furthermore, the guy came right out and said that they have pitched the
desktop Power4 to Apple, but Apple was not interested........
Craig
--
Dr. Craig Hunter
NASA Langley Research Center
AAAC/Configuration Aerodynamics Branch<hr></blockquote>
He was referring to the processor the be discussed @ the MPF, not the Power 4.
note: I added the bold to the email.
Steve likes surprises, and with the latest word being this chip from IBM being Apple's next gen when it's at least 6-9 months off, I'm sure he's putting a cap on the chatter. Oh, and a few denials always helps.
That said, I do find the gcc patch on IBM's website more convincing than this second hand rumor. We don't really know how this conversation went. It is possible that Hunter thought they were talking about the Power4 derivative but that the IBM guy thought they were talking about the Power4 itself. Misunderstandings like that happen all the time.
<strong>mm, so you are saying that in deed Apple will use this chip from IBM that IBM couldn't say that Apple isn't interested? Doesn't sound illegal to me especially Apple can change it's mind later and decide to get the chip.</strong><hr></blockquote>
"and there are technical issues that would
prevent AltiVec from ever marrying with Power4 or its successors."
And then, about a month later, IBM presents a Power4 derivative with AltiVec instructions? And now of course, Apple's suddenly interested? Still doesn't sound like a lie to you?
123
Why is the vector unit i) incompatible with the PPC SIMD standard VMX/Altivec/Viscosity Engine*, which is ii) used by Apple (a significant potential customer, which would subsidise chip production with economies of scale)?<hr></blockquote>
It was sounding like the 64-bit ppc IBM is working on ties two 64 bit units together to do the 128 bit operations of AltiVec/VMX... which is fine for most of the operations.
There's only one (or two) ops that are particularly tough to do without separate registers etc - 'permute' comes to mind immediately. Permute is exceedingly cool -> I can see why a 160 instruction 'VMX' implementation wouldn't do it for Apple. Would it still be a very useful SIMD implementation? Sure! For _new_ code. But emulating permute just looks exceedingly painful - and wedging it into a chip that doesn't have a completely separate altivec unit could also be exceedingly painful.
Having Apple bump volume would probably be very nice - but IBM has plenty of reasons to develop & manufacture a successor chip to the 604 anyway.
Correct me if I'm wrong, but didn't that patch simply add the proper AltiVec flags to correctly indicate the the 64-bit PPC in RS/6000 servers does NOT support AltiVec?
IOW, AltiVec support was added to gcc, along with a new flag for indicating which CPUs support AltiVec, but they forgot to add a value for that flag for teh RS/6000. So there was a patch to set a value for that flag, but the value indicated that it does NOT support AltiVec.
That's what I remember seeing in that diff. Is there something I missed?
<strong>
"and there are technical issues that would
prevent AltiVec from ever marrying with Power4 or its successors."
And then, about a month later, IBM presents a Power4 derivative with AltiVec instructions? And now of course, Apple's suddenly interested? Still doesn't sound like a lie to you?
123</strong><hr></blockquote>
No, not a lie if in deed something needs to be added or changed in order to meet Apple's specs.
The reason i find this rumor hard to believe is because i was told by someone that Apple would be using a power4 core by this time next year. A lot of info has surfaced recently to substantiate this until this article. That's why I'm more inclined to believe this is Steve's way of shrouding his next gen secret.
It just cannot be true. I hope it aint.
Apple limps on with a G4 at 1.4 - 1.8 gig on the currently crippled DDR and BUS? Let's say we get that next April if we're lucky.
What?! IN THE FACE OF 3 GIG PLUS PENTIUM 4 AND A 64 BIT HAMMER WITH AN XP RATING OF 3.4 GIG?!!
No way. Absolutenly no way. No way. NO WAY will Apple be able to prop up 'power'Mac sales in the face of that.
Deluded or not, I refuse to believe that Apple will have no 64 bit answer to within 9 months of the Hammer. It's inconceivable that answer will try to compete with that.
A G4 on .9 from Moto with DDR memory controller.
Maybe. But excuse while I feel a little skeptical. Track record says Moto' will stay behind...and Apple will keep on charging full whack for out of date, insubstantial improvements.
If Apple thinks they are immune from the power and price competition from x86. They're making a foolish mistake. Dell are giving their education share a damn hammering. This IS one of Apple's key markets.
The only way to counter that is to go closer on price and performance.
In education they aint cheap enough.
In workstation land, they aren't powerful enough.
And they are damn expensive. They're gonna keep seeing those 'power'Mac sales shrink until they can deliver something that justifies the price.
The G4 is a flop. A disaster. It has failed to keep abreast of competing developments. Plain and simple.
Photoshop, Lightwave, Premiere...web pages...all supposed Apple strengths...G4 gets its ass kicked.
I recently update to Windows Xp. It aint 'silk'. But 'cotton' gets the job done. And it IS less painful to use.
I've yet to play with Jag'. But I'd like to be able to get a tower that doesn't limp along in Painter 7, that blows the doors off 300 dpi A4 Photoshop images. The dual Gig is still £2,000 etc. Paying 2K for time and a half performance. What? 1.5 gig? If you're lucky. It's a marketing con that is as transparent as the Cube's skirt.
It looks good on the surface but masks the fact that Apple are embarrassingly behind...
Yeesh. We're no further on nine months later than Jan 2002. What the hell is happening.
Apple have got to get serious about its 'power'Mac pro users. The latest round smacked of desperation.
They neglect their core base at their peril. 'power'Mac sales are still smarting and the apologists around here put it down to cycles.
Heck, I'll throw my hat in and put it down to the recession not a consumer chip outstaying its welcome in the workstation arena or hack served mobo.
I don't know why Apple just don't come out and say where the hell they are going with their towers.
I could at least plan my purchase accordingly.
Say what you like about x86 owners. At least they are told up front what's coming.
Yeesh.
Silver lining? Well, if Apple can turn a chip like 8 way superscalar away...they must have something dazzling up their sleeve.
But ask yourself. Who is delivering it.
Motorola?
AMD?
Marklar looks the more realistic of the two.
Lemon Bon Bon
[ 09-06-2002: Message edited by: Lemon Bon Bon ]</p>