put this in your pipe and smoke it!
TidBits: Forums update, IBM & Moto agreement, more...
Architosh Forums Update
Previously we mentioned the Thanksgiving timeframe (which just ended yesterday) for the update to our forums. They will be integrated into the MacNN forums. However the tech team at MacNN may have gotten carried away with too much turkey and wine! We'll miss that target but not by much. We'll keep you updated.
IBM and Motorola PowerPC Agreement Ends?
A MacNN forum poster is stating that IBM and Motorola's PowerPC agreement ends on January 14, 2003. Whether this is true or not is not so interesting to us here at Architosh, as we have previously been told that IBM and Motorola were taking the PowerPC platform in different directions: the former very interested in the powerful but power-lite G3 architecture for embedded markets, and still interested in powerful and specialized PowerPC processors for the desktop and game machine market (GameCube and Playstation 3) , while the latter more interested in the embedded market period.
What is very interesting about this post is the statement that terms of the current agreement (the one that ends Jan 14) is the source of holdup on the now mythical Motorola G5 chip! The source of that disagreement appears to be HyperTransport vs RapidIO. Apple is a member of the HyperTransport Consortium. The assumption is that both companies are refusing to license each other's design technologies as originally agreed upon in the AIM alliance agreements. Moto is behind RapidIO while IBM may be behind HyperTransport.
Our take is that this rumor -- while very believable and in compliance with other information we have obtained -- is not totally correct. We believe that the real reason behind the Moto's G5 failure to come to market is the demand Apple placed on the use of the proprietary Apple Processor Interconnect (ApplePI-BIU) technology, something Moto had no use for with other G5 customers. Furthermore, we believe it may be that Apple's proprietary bus interface unit technology may have found better performance fit with HyperTransport rather than with RapidIO technology.
And contrary to other posts on the Net, Apple actually architected and created the ApplePI-BIU (Apple Processor Bus-Interface Unit) technology, not IBM. The "ApplePI" (trademarked perhaps?) moniker actually appears on Motorola G5 (code named Eleven) chip floor plan documents we have obtained going as far back as 2000. What we actually know is that back in 2000 some critical issues for Motorola's progress on the G5 chip included resolving the ApplePI technology from Apple both at the legal and specification level. We suspect that the legal issue back in 2000 involved the major PowerPC Alliance agreements between IBM and Motorola. The specification issue likely revolved around convincing Apple that RapidIO was the preferred technology for the ApplePI-BIU.
If the MacNN forum poster is correct about the PowerPC agreement terminations this January 14th, then one may imagine a future wherein IBM and Moto actually implement substantially different technologies with their PowerPC offerings, while still collaborating and sharing core technologies -- something that has been of much value to PowerPC customers. Apple would ultimately be better served if both Moto and IBM had products suitable for Apple's gear.
Architosh Forums Update
Previously we mentioned the Thanksgiving timeframe (which just ended yesterday) for the update to our forums. They will be integrated into the MacNN forums. However the tech team at MacNN may have gotten carried away with too much turkey and wine! We'll miss that target but not by much. We'll keep you updated.
IBM and Motorola PowerPC Agreement Ends?
A MacNN forum poster is stating that IBM and Motorola's PowerPC agreement ends on January 14, 2003. Whether this is true or not is not so interesting to us here at Architosh, as we have previously been told that IBM and Motorola were taking the PowerPC platform in different directions: the former very interested in the powerful but power-lite G3 architecture for embedded markets, and still interested in powerful and specialized PowerPC processors for the desktop and game machine market (GameCube and Playstation 3) , while the latter more interested in the embedded market period.
What is very interesting about this post is the statement that terms of the current agreement (the one that ends Jan 14) is the source of holdup on the now mythical Motorola G5 chip! The source of that disagreement appears to be HyperTransport vs RapidIO. Apple is a member of the HyperTransport Consortium. The assumption is that both companies are refusing to license each other's design technologies as originally agreed upon in the AIM alliance agreements. Moto is behind RapidIO while IBM may be behind HyperTransport.
Our take is that this rumor -- while very believable and in compliance with other information we have obtained -- is not totally correct. We believe that the real reason behind the Moto's G5 failure to come to market is the demand Apple placed on the use of the proprietary Apple Processor Interconnect (ApplePI-BIU) technology, something Moto had no use for with other G5 customers. Furthermore, we believe it may be that Apple's proprietary bus interface unit technology may have found better performance fit with HyperTransport rather than with RapidIO technology.
And contrary to other posts on the Net, Apple actually architected and created the ApplePI-BIU (Apple Processor Bus-Interface Unit) technology, not IBM. The "ApplePI" (trademarked perhaps?) moniker actually appears on Motorola G5 (code named Eleven) chip floor plan documents we have obtained going as far back as 2000. What we actually know is that back in 2000 some critical issues for Motorola's progress on the G5 chip included resolving the ApplePI technology from Apple both at the legal and specification level. We suspect that the legal issue back in 2000 involved the major PowerPC Alliance agreements between IBM and Motorola. The specification issue likely revolved around convincing Apple that RapidIO was the preferred technology for the ApplePI-BIU.
If the MacNN forum poster is correct about the PowerPC agreement terminations this January 14th, then one may imagine a future wherein IBM and Moto actually implement substantially different technologies with their PowerPC offerings, while still collaborating and sharing core technologies -- something that has been of much value to PowerPC customers. Apple would ultimately be better served if both Moto and IBM had products suitable for Apple's gear.
Comments
<strong>...there's going to be a PowerPC in the PS3?
I vaguely remember having read somewhere that the "Cell", Sony's chip for PS3, is indeed a PPC and is codeveloped with IBM.
<strong>What's microshaft going to use?</strong><hr></blockquote>
Pentium IV
<strong>Sony was impressed by the chip in the Gamecube and approached IBM.</strong><hr></blockquote>
An army of G3 cores in a Cell chip...I've flirted with the idea often, but never imagined they would actually do it. The G3 may not be walking into the sunset just yet, if what you say is true. Imagine 16 cores of 1 Ghz G3's running on highspeed buses and fast embedded memory...OUTSTANDING!
first of all this whole ibm970 thing has been blown WAY outta porportion.
do you people realize that apple has intel and amd motherboards running os/x as we speak!
i doesnt mean they are on the verge on introducing a product based on these designs!
apple has NEVER let it be known what chip it would be using in upcoming powermacs,why would it now?
the ibm announcement in light of this is strange only IF apple will be using this new processor.
also...ibm has "other" chips on its powerpc roadmap and the gallery has been silent on the possibility of apple using these new chips.
it has been surmised that ibm will put out a g3 variant at 1ghz plus that has altivec and ddr support!!!
it also mentions multipe "cores" in upcoming chips.
these designs,if they are released would be more attractive from a price/performance/ standpoint and also the power drain is VERY important.
dont think power drain is not a HUGE issue.
everything in this damn indusrty is getting smaller,cooler and cheaper.
now for motorola.....................
lets talk about the 7557.
the fact the we know about it before its even here troubles me.
i think it could be a distraction from a more powerful chip.
remember motorola WILL be releasing its 8640,8560 G5 embedded chips VERY soon.
these chips are "book e" compatible,have built in support for 333mhz ddr ram,are fully compatible with existing designs and have rapid i/o.
granted they dont have altivec or FPU's but that COULD be added.
and less not forget that the g5 was ALMOST DONE!
its possible that we could get a G5 here folks!
motorola was having problems with the manufacturing of the G5.
which brings me to my next assertion.....AMD.
we have all heard rumour of NDIVIA chipsets and HYPERTRANSPORT.
AMD is THE founding member of the hypertransport consordium.
apple love HYPERTRANSPORT and has designed a chip "apple pi" to allow motorola and ibm powerpc designs which come with rapid i/o to communicate with a hypertransport based motherboard.
now....
who better to either fab the chips or even fab the whole motherboard than AMD!!!
remember a lot of engineers over at AMD came from motorola,im sure they are quite familiar with motorola powerpc architectures.
remember the "shattering anouncement" that was suppose to happen at comdex in regard to apple and AMD?
well......there is something going on.
<strong>first of all this whole ibm970 thing has been blown WAY outta porportion.
do you people realize that apple has intel and amd motherboards running os/x as we speak!
i doesnt mean they are on the verge on introducing a product based on these designs!
apple has NEVER let it be known what chip it would be using in upcoming powermacs,why would it now?</strong><hr></blockquote>
It hasn't. Everything about 970s in Macs has been speculation, and clearly identified as such.
Question: Given that the 970 is a great chip, aimed right at the more lucrative end of their market, coming soon, with none of the downsides of switching architectures, why wouldn't Apple adopt it? Even if someone (IBM or Mot, or another company) is "sandbagging," to use moki's word, it's a great PowerMac and server processor.
[quote]<strong>also...ibm has "other" chips on its powerpc roadmap and the gallery has been silent on the possibility of apple using these new chips.
it has been surmised that ibm will put out a g3 variant at 1ghz plus that has altivec and ddr support!!!</strong><hr></blockquote>
The gallery has been buzzing about those chips since November last year, actually. But the 970 is a newer and more powerful revelation, so it's taken over. But if you look around, people are still mentally putting that G3 in iBooks and iMacs.
One big issue is whether the SIMD unit is compatible with VMX (AltiVec). If not, it loses a lot of its appeal for Apple.
[quote]<strong>it also mentions multipe "cores" in upcoming chips.</strong><hr></blockquote>
That's a generalized IBM strategy. The huge, hot POWER architecture has already been there for what, two years now? If you look around, there is a lot of speculation around multicore processors in the 970 family - not surprising, given that it's derived from the POWER 4.
[quote]<strong>these designs,if they are released would be more attractive from a price/performance/ standpoint and also the power drain is VERY important.</strong><hr></blockquote>
That's good for notebooks and iMacs. PowerMacs can use pricier and hotter CPUs, though, and so could any server offering Apple releases that's bigger than 1U.
The 970, though big, is not especially hot. It's about where the 1.25GHz G4 is, actually.
[quote]<strong>lets talk about the 7557.
the fact the we know about it before its even here troubles me.
i think it could be a distraction from a more powerful chip.
remember motorola WILL be releasing its 8640,8560 G5 embedded chips VERY soon.</strong><hr></blockquote>
First off, it's the 7457, unless you've gotten information that no-one else is privy to. It's a straightforward die shrink of the current 7455 which will, actually, make it attractive for notebooks and iMacs. If PowerLogix is correct, a 1.2GHz 7457 requires less power than the 800Mhz 7455 currently humming in the iMac, despite running 50% faster. Hello iMac. Hello iBook.
As for the 85xx series, you can't add AltiVec without breaking Book E compliance, so that's not going to happen, and Apple already has its choice of PPCs that already have FPUs on board. If the long-rumored Mot G5 actually does surface, it will probably appear as the MPC7500; 8xxx designates embedded processors.
[quote]<strong>motorola was having problems with the manufacturing of the G5.
which brings me to my next assertion.....AMD.
we have all heard rumour of NDIVIA chipsets and HYPERTRANSPORT.
AMD is THE founding member of the hypertransport consordium.
apple love HYPERTRANSPORT and has designed a chip "apple pi" to allow motorola and ibm powerpc designs which come with rapid i/o to communicate with a hypertransport based motherboard.</strong><hr></blockquote>
I haven't heard this before, although it makes sense. Source?
[quote]<strong>who better to either fab the chips or even fab the whole motherboard than AMD!!!</strong><hr></blockquote>
Heh.
That would explain the persistence of the Apple/AMD rumors without requiring anything as radical as OS X/x86. <a href="http://www.siliconstrategies.com/story/OEG20020412S0034"; target="_blank">This article</a>, dated April '02, says that Mot and AMD are at the end of a tech-sharing agreement that takes them to .1 micron, after which Mot moves their advanced fabrication to Grenoble, France in a partnership with Philips and STM. So the 7457 might be the last fruit of the AMD/Mot alliance.
[quote]<strong>remember the "shattering anouncement" that was suppose to happen at comdex in regard to apple and AMD?
well......there is something going on.</strong><hr></blockquote>
Perhaps. AMD undoubtedly would appreciate the business right about now...
[ 12-07-2002: Message edited by: Amorph ]</p>
<strong>which brings me to my next assertion.....AMD.
we have all heard rumour of NDIVIA chipsets and HYPERTRANSPORT.
AMD is THE founding member of the hypertransport consordium.
apple love HYPERTRANSPORT and has designed a chip "apple pi" to allow motorola and ibm powerpc designs which come with rapid i/o to communicate with a hypertransport based motherboard.
now....
who better to either fab the chips or even fab the whole motherboard than AMD!!!</strong><hr></blockquote>
Why would Apple develop ApplePI if they are about to switch to AMD?
<strong>Sony was impressed by the chip in the Gamecube and approached IBM. There's a consortium doing it but from what has been said it will be a very interesting chip in the PS3.</strong><hr></blockquote>
Furthermore, game developers find the Gamecube's architecture a lot easier to write for than the rather convoluted parallel processing units in the PS2's Emotion Engine.
There are interesting rumours that Nintendo will be getting out of the console hardware market (much as Sega did), so Sony would be nicely positioned to pick up developer expertise...
<strong>CEO of AMD even denied relations with Apple... </strong><hr></blockquote>
You make that sound so... Clintonesque.
"I did not have relations with that Apple!"
<strong>I haven't heard this before, although it makes sense. Source?</strong><hr></blockquote>
You were asking about AMD-Apple connections unrelated to Mac OS X on AMD (but that part of the quote got stolen by martians).
That AMD is the founder/prime motivator behind HyperTransport is freely available info, but I couldn't find a quote that said _exactly_ that.
<a href="http://www.amd.com/us-en/Corporate/VirtualPressRoom/0,,51_104_543_4493~7490,00.html"; target="_blank">HT Consortium Initial PR and founder list</a>
What is Apple doing on that HT initial PR list?
<a href="http://www.hypertransport.org/about.html"; target="_blank">HT Charter Member list</a>
<a href="http://www.hypertransport.org/faqs.html"; target="_blank">HT FAQ</a>
Note (from the FAQ) questions #2 (nVidia), 4, 5, 7, 8, 11, 16, 23, 35, 37. The questions I noted are interesting to me. They include: How does 1394 integrate, speeds @ 6.4 GB/s, nForce chipset, what about PCI-X, what about SMP configs... What does 1394 have to do with _anything_ in AMDland? I didn't see anything about firewire on the motherboard after wandering around ASUS for awhile.
Motorola and IBM are both on the <a href="http://www.rapidio.org/about/list/"; target="_blank">RapidIO Steering Committee</a>. HT & RapidIO are sufficiently similar, that both HT & RapidIO folk claim 'It is easy (read CHEAP in chip real estate) to bridge from here to the other protocol'. Whether we should believe this, I don't know.
The whole 'Apple on AMD' rumors _seemed_ to start from a couple of distinct reports like 'Apple engineer sighted near/at AMD' or vice versa (on different days).
And we hear rumors of 'ApplePI', the Apple Processor Interconnect, which is supposed to be the bus that the IBM PPC 970 has on it! A third bus with exactly the same buzz-words attached to it: point to point, packetized, 6.4 GB/s, maintains coherency for SMP....
These puzzle pieces just have to fit together somehow. They don't add to 'Mac OS X on Athlon64 RSN!' to me. One piece that isn't clear is timing. Is the ppc 970 the first to use one of these new bus plans? Or is it the last to use something _other_ than one of these bus plans.
Edit: Replaced a bracket and a slash found to be missing in action.
[ 12-08-2002: Message edited by: Nevyn ]</p>
Apple on Hypertransport panel? That's a fact and one cannot rule out someform of AMD/Nvidia/Apple and IBM alliance.
AMD and Nvidia are very snug at the moment.
Apple and IBM (from Moki's assertions...)
Add those two together and you can get a potent brew.
I can't wait to see 'THAT' Powermac...with Geforce FX on top.
Lemon Bon Bon