Will Apple's G5 come from IBM? - Page 3

taboo
That's very interesting, maybe, if I understand what your implying, IBM shouldn't have any trouble providing Altivec/Velocity Engine for Apple. No rewriting code, no loss in speed, no hassles w/ Motorola.

[quote]Originally posted by Rhumgod:
<strong>Well so far as the Apple crowd is concerned. They have done massive amounts for their own desktops/workstations/servers.</strong><hr></blockquote>

The 601/603/603e/604/604e/750/620 were all heavily contributed to by IBM. The 601 was an IBM design with a Motorola bus added. The only chip that IBM hasn't taken part in, publicly at least, is the G4. The first G4 was essentially just a G3 + 604's FPU + AltiVec unit. Since IBM was heavily involved in the VMX development, I'd say that they had their hand in there pretty heavily too. Only the G4+ has no signs of obvious IBM involvement, and that has only been out for the last 2 years.

Keep in the mind that the Somerset Design Center was responsible for almost all of the PowerPC desktop design work, and it was fully a joint operation.

Since the fracture in AIM Motorola has done the G4+ and (apparently -- no real proof here) some work toward a G5 for Apple which has since been abandoned. The G5 work that Motorola has done that is public is clearly aimed at the embedded market and not appropriate for Apple. IBM, on the other hand, has done the POWER4 (not to mention 2 & 3) as well as its fair share of embedded work. The recent news makes it clear that the POWER4 work IBM has done over the last two years is going to be far more relevent to Apple than the G5 work that Motorola has done in the same time period.

So which company hasn't been contributing to the Mac?

[quote]Originally posted by Programmer:
<strong>So which company hasn't been contributing to the Mac?</strong><hr></blockquote>

Intel?

"So which company hasn't been contributing to the Mac?"

Shrewd assessment.

Maybe IBM thought the same...hence the breakdown of Somerset.

I read somewhere that IBM felt Motorola didn't contribute enough to the Somerset venture. ie were being somewhat parasitical.

Apple went to Motorola to 'carry the ball'. Three years later...where are we?

Looking at your post...the cold hard facts tend to speak for themselves.

Lemon Bon Bon

[quote]Originally posted by Powerdoc:
<strong>Considering that the new chip will have only one core, and less cache (i consider that they will remove the L3 cache and limit the L2 cache between 512 k and 1 MB) : i say 30 MB for the core, add 10 M transistor for the altivec, and 10 millions more for the cache : it make 50 millions of transistors, like the Programmer said the same amount of transistors that the last release of pentium 4. Considering that the derived chip, let's call him the power VMX is built on SOI 0,13 , it will make less heat thant the pentium 4 on 0,13 (but without SOI).
There is no reason why such a chip can't work in a Apple. Considering that the latter version of the tower have a more big (efficient ventilation) the power VMX will be perfect in it. </strong><hr></blockquote>

You should also consider that if this processor is likely to ship in mid-to-late '03 then it will likely be on the next process... i.e. 0.1 or 0.09 microns. This will allow far more than the Pentium4's level of transistors -- and you can add IBM's back of tricks into that... SOI, low-K dialectrics, copper, and a more stuff that I don't pretend to understand. I'd be surprised if this chip only has a transistor count to match the current Pentium4. If it is really aimed at workstations and towers I suspect they'll beef up the heat sinks and go for it. With 8-way superscalar they'll probably have a huge number of execution units to back that up, big caches... and possibly even a multi-core version or two.

If you haven't seen it, go read <a href="http://www.tubbs.net/main.html" target="_blank">this</a>. If you have read it, read it again. It says he's gone to work on VMX for IBM, but below that it says he is going to work on the Sony / Toshiba / IBM "Cell" project which is the next-gen PowerPC. Yeah, its just a log of some guy and no its not iron-clad or for sure... but it makes a lot of sense and fits in with a bunch of other tidbits floating around (and I don't mean people spewing their unfounded opinions). IBM went 2-way multi-core with the POWER4... I don't think that's the farthest they're going to take the idea. You've got to do something with a billion transistors.

Hopefully this gives people a little confidence that maybe, just maybe, the future of the PowerPC looks pretty good. Mr. Jobs thinks so.

Adding to the speculation, what about the G4? IBM might be building one of those too, but the 64 bit processor gets the attention. The last statement is, "... implements a system interface capable of up to 6.4GB/s." The system interface a weak link of current G4s, yes? Low end PowerMacs would likely use a G4, but current G4s may not be compatible with the G5 for motherboard functions. If IBM builds a G5 for Apple, it would be simple to do a G4 with the same system interface.

[quote]Originally posted by snoopy:
<strong>Adding to the speculation, what about the G4? IBM might be building one of those too, but the 64 bit processor gets the attention. The last statement is, "... implements a system interface capable of up to 6.4GB/s." The system interface a weak link of current G4s, yes? Low end PowerMacs would likely use a G4, but current G4s may not be compatible with the G5 for motherboard functions. If IBM builds a G5 for Apple, it would be simple to do a G4 with the same system interface.</strong><hr></blockquote>

That's part of why I say, now and then, we may see an IBM G3+VMX. IBM's answer to the notebook requirement Apple has. They've already modified the bus to a 200 MHz spec, its not unthinkable that they could replace it with something else -- especially something like RIO or HT which is, in some ways, simpler than a 60x bus.

This chip seems to be the real deal. I hope Apple would talk about the future direction of their architecture a little more. Still, under Jobs...secrecy is what keeps them ahead of the game.

This baby feels like the real deal. Only a year to wait? My wife's iBook will have to do in the mean time!



"power VMX "

I think I like the sound of that...

'Powermac VMX'

Well 'ard' sounding...

Pull those x86 whoop-ass kickin' boots on...

Lemon Bon Bon

Well, when Moto' dropped the G4 fall...they must have really P*ssed Steve Jobs off.

You only have to look at Ati's position in 'power'Macs when they tried to trump his keynote!

Not a guy you want to jerk off!

'It'll be great in two years time when we aren't using you anymore' (Jobs to Moto' exec'...)

That may be folklore...but...it looks Moto' are well and truly on their way out.

Add to that, comments Canada Moto' guy said about 'G5' desktop not even on their radar...

...seems to confirm what some of us refused to believe.

I'll raise my glass to IBM...

Lemon Bon Bon

PS. Personally, I find it quite exciting that the 'Cell' is going to be PowerPC based!

Apple and Sony, two 'cool' hardware brands going to the same place for their cpus...

Add to that Nintendo on IBM tech'..and the PowerPC tech' has alot of marketing cred'...

Beat that X-box!

[ 08-08-2002: Message edited by: Lemon Bon Bon ]</p>

As someone who usually posts long diatribes on theoretical Apple policy and behavior, I'll make this short as it's a question for the tech-heads.

The IBM built G3, currently used in the 700 MHz iBook, allows for a 200 MHz bus speed. If IBM took that G3, and tacked on an AltiVec compatible SIMD extention set, it seems you'd have a pretty kick-arse processor for portable use...possibly even iMac/eMac desktop use.

Can someone in detail do a side-by-side comparison of the IBM 750FX series against a low-power G4? Explain the advantages (if any) of Apple to continue using the G4 vs. a SIMD enhanced IBM G3.

Is there any chance the G4 could go bye-bye, and Apple would use the IBM G3+SIMD for consumer machines, and the Power4 derivative for Pro machines, and leave Motorola behind entirely?

-- Ensoniq

[ 08-08-2002: Message edited by: Ensoniq ]</p>

[quote]Originally posted by Matsu:
<strong>Well, IBM has NOT been any great contributor to the PPC. They ***NEVER*** had any G3 running faster than equivalent G4's over the past 2 and a half years.</strong><hr></blockquote>

The Apple-Moto-IBM alliance had a deal that specified IBM cannot produce chips for Apple faster than what Moto can produce. If it weren't for that, the state of the Mac would be a whole lot better right now.

[quote]Originally posted by Jerome:
<strong>

The Apple-Moto-IBM alliance had a deal that specified IBM cannot produce chips for Apple faster than what Moto can produce. If it weren't for that, the state of the Mac would be a whole lot better right now.</strong><hr></blockquote>

Any evidence of this alleged agreement? Not that it isn't true, but I am very skeptical any such agreement ever existed.

[ 08-08-2002: Message edited by: rickag ]</p>

[quote]Originally posted by blabla:
<strong>


IBM has a tradition of talking about chips years ahead of actual release... While this is not a completly new chip, I dont expect it to show up as early as next week.

My guess: Early next year, when IBMs new chip facility is moving to 0.13.</strong><hr></blockquote>

edit: Whoops! didn't read far enough into the tread. Sorry! :edit

This may be a stupid question but...
Why would IBM build a fab that was
behind the latest technology. Engineered obsolescence? It makes more sense that it's already to go at 0.13 now. Doesn't it? <img src="graemlins/hmmm.gif" border="0" alt="[Hmmm]" />

[ 08-08-2002: Message edited by: tsukurite ]</p>

[quote]Originally posted by Programmer:
<strong>

You should also consider that if this processor is likely to ship in mid-to-late '03 then it will likely be on the next process... i.e. 0.1 or 0.09 microns. This will allow far more than the Pentium4's level of transistors -- and you can add IBM's back of tricks into that... SOI, low-K dialectrics, copper, and a more stuff that I don't pretend to understand. I'd be surprised if this chip only has a transistor count to match the current Pentium4. If it is really aimed at workstations and towers I suspect they'll beef up the heat sinks and go for it. With 8-way superscalar they'll probably have a huge number of execution units to back that up, big caches... and possibly even a multi-core version or two.

If you haven't seen it, go read <a href="http://www.tubbs.net/main.html" target="_blank">this</a>. If you have read it, read it again. It says he's gone to work on VMX for IBM, but below that it says he is going to work on the Sony / Toshiba / IBM "Cell" project which is the next-gen PowerPC. Yeah, its just a log of some guy and no its not iron-clad or for sure... but it makes a lot of sense and fits in with a bunch of other tidbits floating around (and I don't mean people spewing their unfounded opinions). IBM went 2-way multi-core with the POWER4... I don't think that's the farthest they're going to take the idea. You've got to do something with a billion transistors.

Hopefully this gives people a little confidence that maybe, just maybe, the future of the PowerPC looks pretty good. Mr. Jobs thinks so.</strong><hr></blockquote>
you are probabily right, my bet was pessimistic by purpose : i wanted to demonstrate that there was not a heat or watt issue preventing to put this chip in the mac.

And with SOI 0,09 even a power 4 chip with 170 millions chip will work in a mac.

I think that there is a chance for Apple to come back in the raw power competition

[ 08-08-2002: Message edited by: Powerdoc ]</p>

[quote]Originally posted by rickag:
<strong>

Any evidence of this alleged agreement? Not that it isn't true, but I am very skeptical any such agreement ever existed.

[ 08-08-2002: Message edited by: rickag ]</strong><hr></blockquote>

I don't remember where I read that, but I didn't invent it. Sorry guys...

I still have a feeling that this new chip will be used in a *brand new* hardware line up. Above PowerMac line

This year I only expect to see Mot's 7470 (in PowerMac) & 7460 (in PowerBook)

But in next year I would expect to see something like this

The "G" thing is totally gone in order to avoid confusion of the lineups

*New* Mac / XServe: IBM's new chip

PowerMac: Mot's PPC 7500

iMac (17" TFT): Mot's PPC 7470

iMac / eMac / PowerBook: Mot's PPC 7460

iBook: IBM 750CXE


Apple needs more than one supplier for CPU. Same as they need both ATI and nVidia for video chips

[ 08-08-2002: Message edited by: Leonis ]</p>

[quote]Originally posted by Jerome:
<strong>

I don't remember where I read that, but I didn't invent it. Sorry guys...</strong><hr></blockquote>

Perhaps you read that in a rumor site , personally i already read that in AI, but i read also answers saying that was pure crap.

[quote]Originally posted by DaveLee:
<strong>

I think Apple will have to realise soon that they HAVE to make an announcement of some form or other with regard to where their architecture is going. There has been too much rampant speculation about what Apple must/will do, among others this includes much of the 'legitimate' press (i.e. Macworld magazine) and these 'analysts' that are beginning to make waves. Apple really should clarify their commitment to PowerPC and yield some information about what we can expect from them in the future. Jeez, they're not dealing with national security here.</strong><hr></blockquote>


Would it make sense to publish an upgrade plan? Buy a new PowerMac 1.4ghz, and in 10 months, you can slap a new IBM cpu in. That may assuage some of the reluctance to buy now. People are afraid of getting shafted right after buying. (and let's be honest, Apple has shafted users several times). It may help sales now and in the short term.

Just my thoughts.

[quote]Originally posted by tsukurite:
<strong>


Would it make sense to publish an upgrade plan? Buy a new PowerMac 1.4ghz, and in 10 months, you can slap a new IBM cpu in. That may assuage some of the reluctance to buy now. People are afraid of getting shafted right after buying. (and let's be honest, Apple has shafted users several times). It may help sales now and in the short term.

Just my thoughts. </strong><hr></blockquote>

It's a cool idea and I like it... but SJ not, for him it's better to sell 2 times a full system


Moreover, if the new macs will be only dual 1.4 GHz even with FSB DDR... I won't buy one 'cause they will be too slow (3.0 GHz P4 and 2.8 GHz Athlon for end of year)... except if I can upgrade to new CPU during the next year at a very very very low cost.

Aw

An interesting note taken from <a href="http://www.digit-life.com/articles/ibmpower4/" target="_blank">XBT</a>:

The 32-bit GX Bus running at 1/3 of the processor's frequency is used to connect to the I/O subsystem (e.g., by the PCI bridge) and a commutator in case of multiple nodes which contain POWER4 chips for creating clusters.


This is the bus used to connect to the memory/peripheral controller asics. First off, it's only 32bit but runs at a third of the core speed (on a 1GHz processor that's 333MHz SDR and on the 1.3GHz version of the POWER4 it's 433MHz SDR). It would be equivalent to a 64bit 166MHz DDR bus, but with less latency. And less pins out of the processor package. BUT if I understand that article the actual memory bus piggybacks off the L3 cache bus.

A 333MHz GX bus at 32bits would only give 1.3GBps and a 433MHz would give 1.7GBps. Not too much better than MPX.

More fuel for the fire: <a href="http://www.mdronline.com/mpf/conf.html" target="_blank">http://www.mdronline.com/mpf/conf.html</a>

check out the second line:

Session One: PC Processors
Kevin Krewell, Senior Editor, Microprocessor Report; General Manager, MDR

Breaking Through Compute Intensive Barriers - IBM's New 64-bit PowerPC Microprocessor
Peter Sandon, Senior Processor Architect, Power PC Organization,
IBM Microelectronics IBM is disclosing the technical details of a new 64-bit PowerPC microprocessor designed for desktops and entry-level servers. Based on the award winning Power4 design, this processor is an 8-way superscalar design that fully supports Symmetric MultiProcessing. The processor is further enhanced by a vector processing unit implementing over 160 specialized vector instructions and implements a system interface capable of up to 6.4GB/s.

Congrats Macrat you managed to find the exact same info that started this thread to begin with... Maybe you should start reading from page one... <img src="graemlins/lol.gif" border="0" alt="[Laughing]" />

[ 08-08-2002: Message edited by: DaveGee ]</p>

[quote]Originally posted by Powerdoc:
<strong>

Perhaps you read that in a rumor site , personally i already read that in AI, but i read also answers saying that was pure crap. </strong><hr></blockquote>

I don't know about the stipulation between Moto and IBM, but I do remember announcements of IBM getting the G3 to MUCH higher clockspeeds when the G4 was stuck at 500.

And it wasn't on rumor sites.

[quote]Originally posted by Ensoniq:
<strong>

The IBM built G3, currently used in the 700 MHz iBook, allows for a 200 MHz bus speed. If IBM took that G3, and tacked on an AltiVec compatible SIMD extention set, it seems you'd have a pretty kick-arse processor for portable use...possibly even iMac/eMac desktop use.

Can someone in detail do a side-by-side comparison of the IBM 750FX series against a low-power G4? Explain the advantages (if any) of Apple to continue using the G4 vs. a SIMD enhanced IBM G3</strong><hr></blockquote>

The 700 Mhz 750FX that's in the iBooks currently, despite its bus being clocked at only 100 Mhz (half of the possible 200 Mhz) is actually faster at non-altivec floating point intensive code than the 700 Mhz G4 in the iMac and eMac. It is even faster in some applications than the 800 Mhz G4 in the TiBook with 1 MB DDR L3 cache . If it had Altivec, it'd be faster across the board.

[ 08-08-2002: Message edited by: timortis ]</p>

[quote]Originally posted by pi radians:
<strong>

I don't know about the stipulation between Moto and IBM, but I do remember announcements of IBM getting the G3 to MUCH higher clockspeeds when the G4 was stuck at 500.

And it wasn't on rumor sites.</strong><hr></blockquote>
So this stipulation that IBM cannot make faster chips (in mhz at least) than Mot was false.

Well..I think what he meant about the IBM chips not being able to go past whatever Moto had was not : They cannot engineer one. But the agreement was that Apple could not sell them higher than Moto, and IBM could not necessarily publicize them faster.

Or atleast thats what I got from his posts.

Apple has made me so cynical that I fully expect a lame 867 G4 lowend to a duel 1067 G4 top end next week but still it is interesting that the new Ibm chip is for "desktops".Now no one else realy uses the ppc in a desktop but Apple.This clearly points to apple! I need to keep reminding myself of that for it to sink in.But what about a time scale? There is no rule that they must announce a new processor line far in advance at such a forum.Reference to past events is no good to make a decision on.This is not really a "new" chip,just a desktop derrivative based on power4.It is quite possible that they have been under some nondisclosure agreement with apple.Apple may have even helped design and fund it.An nda makes sense because if apples move to a new powerful chip was publicly know it would have killed G4 tower sales.So it is possible Apple will release the new chip in the powermacs soon,maybe even before the conference.
Well we can dream cant we?

Some interesting papers from IBM RD journal and news

<a href="http://www-1.ibm.com/servers/eserver/pseries/hardware/whitepapers/power3wp.html" target="_blank">POWER 3 -&gt; PPC 64</a> (from october 1998 !)

<a href="http://www.research.ibm.com/journal/rd46-1.html" target="_blank">POWER4 all you want to know on POWER4</a> (january 2002 -&gt; shipping on june 25 in pSeries 690)

Future of Mac is in IBM with 6-12 months transition with Moto


Aw

<strong>Originally posted by pi radians:
I don't know about the stipulation between Moto and IBM, but I do remember announcements of IBM getting the G3 to MUCH higher clockspeeds when the G4 was stuck at 500.

And it wasn't on rumor sites.</strong>

It was from rumor sites and was blatantly false. The 750 (G3) and 7400 (G4) have virtually the same clock speed frequencies on the same fabrication process. The only way IBM could have shipped much higher G3 clock speeds was a better fabrication process, and it wouldn't have been much of an improvement since the fab improvement was very minor from 0.2 micron to 0.18 micron.

Once Moto shipped the 7450, the 7450 clockspeed was and will always be higher than any G3.

[ 08-08-2002: Message edited by: THT ]</p>

<strong>Originally posted by Ensoniq:
Can someone in detail do a side-by-side comparison of the IBM 750FX series against a low-power G4? Explain the advantages (if any) of Apple to continue using the G4 vs. a SIMD enhanced IBM G3.</strong>

A 0.13 micron 7450 G4 will clock 50% higher that an IBM 750fx. That pretty much should be the end of the story right there in the G4's favor.

<strong>Is there any chance the G4 could go bye-bye, and Apple would use the IBM G3+SIMD for consumer machines, and the Power4 derivative for Pro machines, and leave Motorola behind entirely?</strong>

A 0.13 micron 7450 based G4 is a sweetspot chip. I think Apple should us it if Moto fabs it.

So whats the 7470? Some MOSR dream-product? AFAIK, Motorola havent announced any new G4 chip after the SOI G4.

To be honest, I dont think Motorola is going to add new features to the G4 line. I suspect the Motorola "high-end" PPC is dead.

hahahah the Myth that won't die. How these misconceptions about the G3 have come about it amazing...it's folklore.

[quote]Originally posted by eddively:
<strong>Well..I think what he meant about the IBM chips not being able to go past whatever Moto had was not : They cannot engineer one. But the agreement was that Apple could not sell them higher than Moto, and IBM could not necessarily publicize them faster.

Or atleast thats what I got from his posts.</strong><hr></blockquote>

As I recall, the agreement that you are refering to was for when IBM was working as a back-up supplier of G4's for Apple when Moto could not meet the supply demands that they were contracted for, and did not apply to the G3.

Just becouse IBM is not making 1000mhz G3's does not neccessarely mean that they cant. If they do not have a customer for them they would be stupid to make them, and Apple wont/cant clock a consumer G3 faster than a Pro G4 becouse that would be really BAD marketing.

Most of the other customers for G3's are embeded or specializer processors such as Nintendo, and the Game Cube probably wont scale in speed, but will be replaced by a new console in 3 years. The ballance of IBM's G3 customers are probably buying in low encough quantities that it may not be economically feasable for IBM to put the faster chips into production.

Back to the topic at hand...

Apparently, IBM announced this processor last year, but labeled it as an "embedded" processor. With this in mind, can those that have experience in this area speak to the possibility of intro'ing this CPU now? Is a year long enough? Given that this is something like Son of POWER4, it seems that it should be fairly straight forward to pull the various bits together.

I really can't figure out if I should be excited yet or not.

Anyone? <img src="confused.gif" border="0">

[ 08-08-2002: Message edited by: tsukurite ]</p>

Not my field of expertise, but I am in the electronics mfg field. From what I know of chip design it is fully possible that it could ship in the timeframe you mention. It would need to be an aggresive schedule, based on existing technology would help, and then there is always lady luck on or against you.

[quote]Originally posted by tsukurite:
<strong>Back to the topic at hand...

Apparently, IBM announced this processor last year, but labeled it as an "embedded" processor. With this in mind, can those that have experience in this area speak to the possibility of intro'ing this CPU now? Is a year long enough? Given that this is something like Son of POWER4, it seems that it should be fairly straight forward to pull the various bits together.

I really can't figure out if I should be excited yet or not.

Anyone? :confused:

[ 08-08-2002: Message edited by: tsukurite ]</strong><hr></blockquote>

NOW will be the BEST OPTION for Apple but I don't think we will see a new IBM chip in Mac before at least MWSF 03... And it depends on the fact that Apple can say bye-bye to moto now or not... ??? Someone could tell us ?

Aw

If this chip is being made for Apple and seeing as how its not an embedded processor, wouldn't Apple get to announce models using it before IBM spews out technical details?

<strong>Originally posted by JCG:
Just becouse IBM is not making 1000mhz G3's does not neccessarely mean that they cant. If they do not have a customer for them they would be stupid to make them, and Apple wont/cant clock a consumer G3 faster than a Pro G4 becouse that would be really BAD marketing.</strong>

Once more onto the breach... due to the microarchitecture of the G3, the way the G3 is designed, IBM cannot ship a faster 750 based G3 than a 7450 based G4 on the same fab process (or thereabout). We've been talking about execution pipeline stages for a long time now. The 750 G3 (and the 7400 G4) has 4 stage pipeline. The 7450 G4 has a 7 stage pipeline. Due to the simplicity in circuitry for each stage in a deeper pipelined processor, those stages can be clocked higher allowing the procesor clock higher. This is why the 10 stage Athlon processor clocks higher than the 7450 G4. This is why the 20 stage Pentium 4 clocks higher than all other desktop uprocessors. And this is why the those G3 rumors were false.

In short, it means IBM can't.

[quote]Originally posted by apple.otaku:
<strong>If this chip is being made for Apple and seeing as how its not an embedded processor, wouldn't Apple get to announce models using it before IBM spews out technical details?</strong><hr></blockquote>

Maybe not but lot of hope in that...

Aw

[quote]Originally posted by apple.otaku:
<strong>If this chip is being made for Apple and seeing as how its not an embedded processor, wouldn't Apple get to announce models using it before IBM spews out technical details?</strong><hr></blockquote>

No..

1: I doubt this chips are made only for Apple
2: Just about every PPC used by Apple was first announced by IBM/Motorola. 7400, 7450,7455, 750cxe, 750fx. Probably the original 750 too, but I dont remember.