The real issue as you point out later is price. However, the high price of a processor like this is actually a function of amortizing the very high fixed costs of design and manufacturing capability rather than the actual variable costs of a little bit of copper and silicon.
In fact, the change in your pocket contains more commodity value (copper, tin, nickel etc.) than would a power4 processor (I'm assuming you have change in your pocket). IBM has to charge a ton for each processor because the only system they can put it in is their server system, of which they only sell thousands or tens of thousands of units (not hundreds of thousands or millions).
Manufacturing this chip in high volume would bring the per chip manufacturing costs (including R&D and equipment amortization and depreciation) to levels comparable with the G4.
</strong><hr></blockquote>
Now I thought I read that somewhere that the 32 Megs of L3 DDR cache per core were on the die (though that doesn't make much sense). That would raise the price a smidgen. Plus being such a big chip it probably takes the wafer space of 2-3 G4s at least, right? It would seem to make sense for IBM to get its money's worth for its fab efforts even if the design costs could be spread out.
Regarding Programmer's post, we already have news reports (not rumors) of IBM making custom variants of chips for valued customers (which was not their practice before). So it certainly is plausible it could happen (though I'd wager it'll be a distant cousin to the Power4).
All this talk about the Power4 is really exciting. Its single chip performance will be equivalent at its current clock speed to future Pentium 4s; however, Power4's currently being fabed on an old process. It would be wonderful to have parity once again, and it seems to me that the Power4 has a lot of room to grow.
Did anyone notice that the Power4 jumped from 700 to 1300MHz at one time? That's a really wonderful performance spike; who's to say Apple isn't working with IBM right now to boost 1300MHz to 2500MHz or so? It would be a wonderful change from the 500MHz quandry we were in for such a long period of time. It's important note that the Power4 sports a 17 stage pipeline, which is what gives it such great MHz mobility. (I never thought I'd be touting deep pipelines. ) Deep pipelines are also a marked departure from IBM's traditional stance of high IPC throughput.
Maybe the G5 is actually going to be Power4 at an advanced process. Perhaps instead of G5s we'll have Power5s at MWSF. Please Apple, please?
I don't think the G5 will show up this summer. A friend of mine works at Moto in one of the chip design divisions(he's like a mid level manager). Anyway, over a year ago he told me that Apple cancelled their contract with Moto for the G5. The reason had to do with the instruction set that the chips use. Apple wanted to use somthing called Apple Pi (or something like that) but the original design used somethign else (don't ask me because I don't know and frankly don't understand). Moto did a cost analysis and determined that they simply couldn't affort to make the changes that Apple wanted. The result - the contract was cancelled and Apple took their business to IBM. I asked about Altavec and he said he didn't know, but that he had heard some money had been exchanged inorder for IBM to develop a chip that used it. Now keep in mind that this was a year ago and I haven't spoken to the guy about it since because it's sort of rude to ask him about company secrets.
<strong>I don't think the G5 will show up this summer. A friend of mine works at Moto in one of the chip design divisions(he's like a mid level manager). Anyway, over a year ago he told me that Apple cancelled their contract with Moto for the G5. The reason had to do with the instruction set that the chips use. Apple wanted to use somthing called Apple Pi (or something like that) but the original design used somethign else (don't ask me because I don't know and frankly don't understand). Moto did a cost analysis and determined that they simply couldn't affort to make the changes that Apple wanted. The result - the contract was cancelled and Apple took their business to IBM. I asked about Altavec and he said he didn't know, but that he had heard some money had been exchanged inorder for IBM to develop a chip that used it. Now keep in mind that this was a year ago and I haven't spoken to the guy about it since because it's sort of rude to ask him about company secrets.</strong><hr></blockquote>
whoa! jinx on the pi thing. yes, your story exactly corresponds to what i've heard. exactly! ibm's project should be in volume production right now.
I find your lack of faith disturbing...</strong><hr></blockquote>
awesome man! I like you.
but...."your sad devotion to that ancient religon has not helped you conjure up the stolen data tapes, or given you clarvoyance enough to find the location of the hidden rebel fort-uh"
As pjn23 and admactanium stated above, I also heard that Apple had turn down Moto's proposed G5. The reason I heard was that the memory scheme was not what Apple wanted. I had assumed that was RapidIO instead of HyperTransport. But if IBM is building the new chip, are they not committed to RapidIO for powerPC chips?
I read in February I think about 'looking to someone rather then MOTO for Apple's next chip". Basically saying IBM was going to do it. So there seems to be a lot of sharing of the same sentiment that IBM will be doing Apple's next chip.
Dare I say it's CONFIRMED that IBM is doing Apple's next chip
Introducing the new PowerMac G5, starting at $11,999.
Anyone who seriously thinks the Power4 is a viable alternative is on something.
Oh, and moki knows nothing.</strong><hr></blockquote>
Nobody is suggesting that the POWER4 will suddenly appear in the PowerMac G5. The existance of the POWER4, however, demonstrates that IBM knows how to build faster processors with the POWER/PowerPC instruction set. The long pipelines, wide superscalar design, and 0.13 micron process are directly applicable. The multi-core design may also be applicable. The enormous caches and fast+wide buses are not applicable -- and this (along with massive I/O subsystems) is where most of the cost of the POWER4 comes from.
Moki may very well know nothing, but he's clearly having fun at it.
Now if it takes 3-5 years of full bore effort to design a chip and they just had this falling out last year isn't MWNY 02 or MWSF 03 a little optimistic?
Of course if there is a new uber PowerPC coming from IBM then my guess that the XServe's chipset was the trailblazer not for the next PowerMacs but for the next iMac may be borne out.
I think I'm good either way as long as something good is coming . Coming soon
Moki may very well know nothing, but he's clearly having fun at it.</strong><hr></blockquote>
Everything I know about moki makes me believe he wouldn't trifle with us. He obviously wouldn't divulge anything he couldn't legally state - what would be the incentive there? He's a responsible head of a prominent Mac shareware company. I'm not going to swear by any testimony I hear around here, but what moki says does have credence, IMO.
I just don't think there is going to be a G5 this summer and the reason is because all the Apple execs are filing to <a href="http://biz.yahoo.com/t/a/aapl.html"; target="_blank">sell stock</a>. That tells me that the iMac sales have flattened out and that they don't expect to see a significant boost from the power mac line (which they would if the G5 was released). And don't you think they would have held the release of the Xserver if the G5 was just around the corner?
Comments
<strong>
The real issue as you point out later is price. However, the high price of a processor like this is actually a function of amortizing the very high fixed costs of design and manufacturing capability rather than the actual variable costs of a little bit of copper and silicon.
In fact, the change in your pocket contains more commodity value (copper, tin, nickel etc.) than would a power4 processor (I'm assuming you have change in your pocket). IBM has to charge a ton for each processor because the only system they can put it in is their server system, of which they only sell thousands or tens of thousands of units (not hundreds of thousands or millions).
Manufacturing this chip in high volume would bring the per chip manufacturing costs (including R&D and equipment amortization and depreciation) to levels comparable with the G4.
</strong><hr></blockquote>
Now I thought I read that somewhere that the 32 Megs of L3 DDR cache per core were on the die (though that doesn't make much sense). That would raise the price a smidgen. Plus being such a big chip it probably takes the wafer space of 2-3 G4s at least, right? It would seem to make sense for IBM to get its money's worth for its fab efforts even if the design costs could be spread out.
Regarding Programmer's post, we already have news reports (not rumors) of IBM making custom variants of chips for valued customers (which was not their practice before). So it certainly is plausible it could happen (though I'd wager it'll be a distant cousin to the Power4).
Did anyone notice that the Power4 jumped from 700 to 1300MHz at one time? That's a really wonderful performance spike; who's to say Apple isn't working with IBM right now to boost 1300MHz to 2500MHz or so? It would be a wonderful change from the 500MHz quandry we were in for such a long period of time. It's important note that the Power4 sports a 17 stage pipeline, which is what gives it such great MHz mobility. (I never thought I'd be touting deep pipelines.
Maybe the G5 is actually going to be Power4 at an advanced process. Perhaps instead of G5s we'll have Power5s at MWSF. Please Apple, please?
<strong>These aren't the G5's you're looking for... move along.</strong><hr></blockquote>
yay mokis back!
Introducing the new PowerMac G5, starting at $11,999.
Anyone who seriously thinks the Power4 is a viable alternative is on something.
Oh, and moki knows nothing.
<strong>Oh, and moki knows nothing.</strong><hr></blockquote>
I find your lack of faith disturbing...
<strong>
I find your lack of faith disturbing...</strong><hr></blockquote>
moki, i'd like a nice slice of apple 3.1415
<strong>I don't think the G5 will show up this summer. A friend of mine works at Moto in one of the chip design divisions(he's like a mid level manager). Anyway, over a year ago he told me that Apple cancelled their contract with Moto for the G5. The reason had to do with the instruction set that the chips use. Apple wanted to use somthing called Apple Pi (or something like that) but the original design used somethign else (don't ask me because I don't know and frankly don't understand). Moto did a cost analysis and determined that they simply couldn't affort to make the changes that Apple wanted. The result - the contract was cancelled and Apple took their business to IBM. I asked about Altavec and he said he didn't know, but that he had heard some money had been exchanged inorder for IBM to develop a chip that used it. Now keep in mind that this was a year ago and I haven't spoken to the guy about it since because it's sort of rude to ask him about company secrets.</strong><hr></blockquote>
whoa! jinx on the pi thing. yes, your story exactly corresponds to what i've heard. exactly! ibm's project should be in volume production right now.
<strong>
I find your lack of faith disturbing...</strong><hr></blockquote>
awesome man! I like you.
but...."your sad devotion to that ancient religon has not helped you conjure up the stolen data tapes, or given you clarvoyance enough to find the location of the hidden rebel fort-uh"
<strong>The only thing Moki knows is how to rip off Star Wars dialogue.</strong><hr></blockquote>
hey being "good at[ ripping of star wars dialogue] is one thing but, good against the real...thats something else"
<strong>
hey being "good at[ ripping of star wars dialogue] is one thing but, good against the real...thats something else"</strong><hr></blockquote>
Actually, It's "good against the living" not "good against the real".
:cool:
[ 06-10-2002: Message edited by: AusTex ]</p>
Dare I say it's CONFIRMED that IBM is doing Apple's next chip
<strong>Yes, lets use the Power4.
Introducing the new PowerMac G5, starting at $11,999.
Anyone who seriously thinks the Power4 is a viable alternative is on something.
Oh, and moki knows nothing.</strong><hr></blockquote>
Nobody is suggesting that the POWER4 will suddenly appear in the PowerMac G5. The existance of the POWER4, however, demonstrates that IBM knows how to build faster processors with the POWER/PowerPC instruction set. The long pipelines, wide superscalar design, and 0.13 micron process are directly applicable. The multi-core design may also be applicable. The enormous caches and fast+wide buses are not applicable -- and this (along with massive I/O subsystems) is where most of the cost of the POWER4 comes from.
Moki may very well know nothing, but he's clearly having fun at it.
Of course if there is a new uber PowerPC coming from IBM then my guess that the XServe's chipset was the trailblazer not for the next PowerMacs but for the next iMac may be borne out.
I think I'm good either way as long as something good is coming . Coming soon
<strong>
[SNIP]
Moki may very well know nothing, but he's clearly having fun at it.</strong><hr></blockquote>
Everything I know about moki makes me believe he wouldn't trifle with us. He obviously wouldn't divulge anything he couldn't legally state - what would be the incentive there? He's a responsible head of a prominent Mac shareware company. I'm not going to swear by any testimony I hear around here, but what moki says does have credence, IMO.
[ 06-11-2002: Message edited by: Big Mac ]</p>
[ 06-11-2002: Message edited by: pjn23 ]</p>