IBM and Motorola G5's?

[MacManiac1224]

I just took a look at MOSR, and I saw a very interesting piece of info:
---------
Presently we are waiting for IBM's discussions of the new desktop
CPU at the forum later in the month to draw any new conclusions,
but it looks as if the G5 family might not be one lineage of chips from
one company -- Apple may be attempting to increase the competition
between IBM and Motorola by using both POWER4 and PPC
8500-based CPUs....
-----------

This sounds kinda crazy, but sounds really good. It is kinda like the AMD and Intel battle, but just over Mac cpu's, which could mean faster mac's in faster time if thier is a little compeition. Personally I would rather have the Power4 chip powering my Mac, but if Motorola comes out with a 64-bit G5 first, I don't think I can wait any longer, and I am going to take the plunge.

What do you guys think?

[MacManiac1224]

MOSR also says:
--------------
Don't count Motorola out of game just yet as far as future Mac
processors are concerned. Although Apple and IBM continue to work
on a solution that some have dubbed GPUL (Giga-Processor Ultra
Lite - although this is most certainly not the name anyone involved is
using; certain rumormongers just made it up, as they have numerous
other "codenames"), Motorola apparently has plans to supply Apple
with hundreds of thousands of PowerPC 8500-family processors
beginning in early 2003.
----------------
Early 2003? MWSF? I hope so! Maybe Motorola G5 in MWSF, and IBM G5 (or something else) in MWNY, but most likely MWSF 2004. I don't think Apple is going to release two very powerful chips in 1 year?

What do you guys think?

[Shrek]

What is MOSR? Can you provide a link?

[void]

http://www.macosrumors.com/

[Shrek]

Well, well, well. Well, did I not predict something like this just a few weeks ago? Looks like I could be right on target, again! :D Well, almost. :rolleyes:

[solvs]

Originally posted by Shrek
What is MOSR? Can you provide a link?

They're good for a laugh, and wishful thinking, but don't take anything they say too seriously. Someone once asked me how I would feel if MOSR and MacNN didn't exist anymore, and I told then I couldn't care less. And I really couldn't.

I wonder how much they like their ATI Radeon 8500 with ADC (there is no such creature, read their archive).

Even for rumors, they're pretty out there.

[j763]

Originally posted by MacManiac1224
I just took a look at MOSR

doesn't that answer the question? ;)

[Shrek]

Originally posted by j763


doesn't that answer the question? ;)

no

[alex_ant]

Mac OS Rumors is run by a guy whose only job is to cull rumors from other rumor sites (like this one) and present them in a single place. It is not really a rumor source - just a rumor summary.

There will be a line of Motorola MPC8500 (G5) processors. The only questions are: 1) when, and 2) will they appear in Macs. The answer to 1 is "when the fab in which they will be produced is finished," which will probably be sometime no sooner than Q2 of next year. The answer to 2 is "probably." If a Motorola G5 fulfilling all of Apple's requirements existed, there is no reason Apple wouldn't want to use it, even if they already had the IBM GPUL.

By this time next year, I think it is likely that Apple will be in the process of transitioning its entire product lines to these new CPUs. I think the G3 will be replaced by the G4 in the iBook next year. Whichever CPU the PowerBook gets will be dependent upon which of these two next-gen CPUs is the most energy-efficient.

[beatle888]

motorola G5's in january...
and the power4's towards the end of the year.

i think this would be a good guess.

[DaveGee]

MOSR has and is a joke...

They even claim GPUL was just 'made up' by a bunch of rumormongers. Give me a break they are fishing for people to prove them wrong.

Also ... while this is just speculation... Why would Jordan Hubbard (Apple's manager of BSD technologies) say to a room full of hacker types the following:

"One of the benefits of working with Motorola is that the hardware won't get faster."

Talk about a right hook to the jaw... No punches being pulled with that comment.

Do you really think he would have said that if he didn't already know that the next PPCs were NOT gonna have a big letter M on the top.

Maybe I'm wrong but the way I see it.

If Jordan DIDN'T know what was going on in the CPU department he wouldn't have said what he did.

If Jordan DID know what was going on in the CPU department and the next big CPU *was* gonna come from MOT he wouldn't have said what he did.

If Jordan DID know what was going on in the CPU department and the next big CPU *was not* gonna come from MOT then yea what the heck... Take a jab it ain't gonna hurt none.

Dave

[ddtlm]

alex_ant:

There will be a line of Motorola MPC8500 (G5) processors. The only questions are: 1) when, and 2) will they appear in Macs.
A1: They are shipping and have been for a while. A2: No, current 85xx chips are totally unsuitable for Macs. They lack things such as floating point units, which really does lower the ol Photoshop performance, doesn't it. :)

Here is an example: 8540.
http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?code=MPC8540&nodeId=01M98655

And the similar 8560:
http://e-www.motorola.com/webapp/sps/site/prod_summary.jsp?code=MPC8560&nodeId=01M98657

The chips aren't even high-clocking. Moto lists 1ghz as the Max, not good at all, certainly not much of a G5.

I do not see any reason why a G4 exactly as we know it except fabbed on 90nm or 130nm processes would not be entirely suitable for iBooks, iMacs and PBooks for some time. There is no need to reinvent the wheel if IBM is going to come in and do the high-end chips. Let Moto be slow, thats fine for some models.

[cubist]

Originally posted by DaveGee

If Jordan DIDN'T know what was going on in the CPU department he wouldn't have said what he did.

If Jordan DID know what was going on in the CPU department and the next big CPU *was* gonna come from MOT he wouldn't have said what he did.

If Jordan DID know what was going on in the CPU department and the next big CPU *was not* gonna come from MOT then yea what the heck... Take a jab it ain't gonna hurt none.


You have hit the nail right on the head! Mot's high-speed PPCs are for routers and other dedicated applications. They don't have any other cakes in the oven.

More speculation: IBM's new chip is likely to show up in a low-end RS-6000, which they would position as a small-business database server. It'll give them something to compete with Sun in the low-end server area. (Nobody would seriously use a winblows machine as a database server.) As to whether the jeepul will be used in Macs, that's more speculative. While desktop macs wouldn't be viewed as competitive with a low-end RS-6000, a jeepul xserve might. So, it may be interesting after all that IBM is conducting a survey about whether DB/2 should be ported to mac os x server. My guess is that IBM wouldn't mind losing a few RS6000 sales to xserves, as long as their chip and their software are getting out there.

I don't think we'll get any answers on Oct. 7. But Mac Rumors is the place to keep an eye on!
:D

[Beigean]

ddtlm,

alex_ant was talking about a G5 8500 specifically. had he said 85xx then that would include the -40 and -60 models. however the general assumption is that a desktop Macable model would be labeled 8500, and no such processor is shipping or has been officially announced.

[beatle888]

i just saw a comercial for IBM personal
computers. they had an Intel Pentium4.

i wonder when they would start using their
power4 light version for their own personal PC's

[ibjoshua]

Originally posted by beatle888
i just saw a comercial for IBM personal
computers. they had an Intel Pentium4.

i wonder when they would start using their
power4 light version for their own personal PC's

correct me if i'm wrong but PCs run windows and windows doesn't run on powerpc chips.
so probably never.
linux desktops run on powerpc chips of course. do IBM make linux desktops?

i_b_joshua

[scem0]

I think that they will release the g5 in Jan. It will be called the g5 ( Big suprise :eek: ) and it will have speeds at about 1.2 GHz 1.4 GHz & 1.6 GHz. Then later in the year, maybe right b4 christmas, apple will release the GPUL mac with speeds at 1.5, 1.7, and 2.0 GHz. Total guesses, but this is what I am guessing.

[snoopy]

In just ten more days we should know a lot more. My guess is that the IBM G5 (also called by the IBM code name GPUL) is very far along in development, and may be sampling already. I believe Motorola no longer has plans for an Apple G5.

[ddtlm]

Beigean:

Oooh, I get it, "8500" kinda like 7460, 7470, 7500, and whatever other names rumormongers have decided on. :)

I seriously doubt it would be called 8500 though, cause it could have little or nothing in common with the other 85xx's.

[Macrazy]

Apple is going to use both Motorola and IBM for two reasons.

First reason Apple will use Motorola PPC for low end user such as ibook, iMac, eMac, PowerMac, and xServe. Second reason Apple will use IBM power4 lite for high end user. Such as PowerMac, iMac, Powerbook, xServe and or PowerxServe. Also maybe (Who knows) that Apple will use Power4 (not Lite) for HIGH END Server for xServe.

Anyway, for example IBM offer two things. Intel Pentium4 on IBM computer and run Mirco*****. IBM server run Power4 run Linux Server.
What If.. Apple decide to offer Intel too? (Jeez)

Let's wait till tomorrow.

[macmunch]

We see a G5 Moto at MWSF 2003

And over the year it will be ....

iBook --> G4
iMac ---> G4
emac ---> G4
Powerbook ---> G5 (moto)
PowerMac ---> G5 (moto)
Servers (Xserve) ---> G5 (IBM)

Dont forgett the IBM chip will be not so cheap like a G4 !!

So it look for me like a good server chip
:)

[Macrazy]

Well.. I learned that IBM Power4 Lite same price as Motorola's future G5. That what my friend told me. Who work at IBM. :D

[DaveGee]

Originally posted by Macrazy
Well.. I learned that IBM Power4 Lite same price as Motorola's future G5. That what my friend told me. Who work at IBM. :D

Given both the mot 'g5' and the 'IBM Power4-lite' are both priced the same then I don't see why they would both be used what would the advantage be?

Dave

[vniow]

If the Power4mini-me is half as hot as the Power4, then you don't wat to see it in laptops. or a tightly-packed, system like an iMac.
If they are the same price, then the Moto G5 could go in iMacs, eMacs and Powerbooks, while the IBM G5 could go in Powermacs and xServes.:)

[macmunch]

good idea edvniow ...
:)

[alex_ant]

Originally posted by DaveGee
Given both the mot 'g5' and the 'IBM Power4-lite' are both priced the same then I don't see why they would both be used what would the advantage be?
The advantage would be the competition supposed to be happening between Motorola and IBM that would lead each to try to make their chips faster than the other, bringing better/faster chips to Apple.

[Macaddict16]

Introduce Moto's G5 at MWSF in PowerMacs and xServe and at MWNY introduce the POWER4 chip (different name than G5) and place those in PowerMacs and xServes and place Moto's G5 in iMacs/PowerBook (G4 in iBook and eMac)

[Macrazy]

Originally posted by DaveGee


Given both the mot 'g5' and the 'IBM Power4-lite' are both priced the same then I don't see why they would both be used what would the advantage be?

Dave

My mistake about " Will be the same price"

I just spoke with my friends who work at IBM. They told me that both processors are not the same price. The price is uncomfirmed yet. They think the cost will be (IBM Power4 lite) $385.00 (64 bits). They are still discuss about costs. They will not sell one piece. They required to sell for 2,000 items for this price.

They sure the cost will be (Motorola G5 PowerPC) $275.00 (32 bits). They had prototype of Motorola G5. It was year ago. Umm... I wonder why the G5 have 32bits? Why not 64 bits? Good Question.

One is 64 bits and other is 32bits. We do see why we have two different advantages.

I hope its true!

[Macrazy]

Power4 May Give Insight into Next-Generation Apple CPU
by Remy Davison, Insanely Great Mac
October 10th 2002

IBM next week will provide a preview of its 64-bit Power4 processor. Designed for both desktops and entry-level servers, the new PowerPC chip could find its way into Apple desktops sometime in the future, according to a report in the EETimes.

The Power4 will be the first IBM PowerPC processor which incorporates Motorola's AltiVec instruction set, which has long been a feature of the Motorola G4s used in Apple's Power Macs, PowerBooks, flat-panel iMacs and eMacs.

IBM will detail the chip at the Microprocessor Forum, running October 14-17. According to the report EETimes, the chip will support extensions which will provide compatibility with PowerPC architecture, with backward-compatibility with 32-bit environments. Commentators in the article argue that this will ensure compatibility with the Mac OS.

"Apple would have to be crazy not to use this part," Peter Glaskowsky. editor-in-chief of The Microprocessor Report, the forum's host, said to EETimes.

However, there are obstacles to the Power4's use: Apple would need to invest significant time in OS development to support a 64-bit environment. Moreover, software developers would need to rework their applications to take full advantage of the chip.

The upsides, should Apple decide to adopt the Power4, include a superscalar pipeline and symmetric multiprocessing.

Analysis: Jobs has had a sneak preview of the PowerPC roadmap that we aren't privileged to, but it's safe to say he's happy with the 'options'. The work needed to rework OS X to get the best performance out of the Power4 may take considerable time and investment; however, it's not beyond the bounds of possibility that Apple could implement the Power4 and use the existing 32-bit architecture in its boxes. It's been done before and they're doing it now: the current PM G4s are a case in point, employing processors which cannot take full advantage of DDR.

Similarly, Apple has pulled such stunts in the past, with some deplorable, half-baked Performas. Power4 - like the early PM G4/350 - might give Apple the marketing edge it sorely needs while it competes with the next-generation Intel/AMD chips. To be able to use Power4 in a 32-bit environment while it waits for the OS to catch up may prove a temptation for Apple that is too great to ignore.

[MacCoaster]

Originally posted by Macrazy
They sure the cost will be (Motorola G5 PowerPC) $275.00 (32 bits). They had prototype of Motorola G5. It was year ago. Umm... I wonder why the G5 have 32bits? Why not 64 bits? Good Question.
Motorola G5 isn't prototyping anymore. It's in production; been in production for a good while. Those prices are probably the current G5 85xx. Those processors are not suited for desktop Macintoshes.

[snoopy]

Originally posted by Macrazy
Power4 May Give Insight into Next-Generation Apple CPU
by Remy Davison, Insanely Great Mac
October 10th 2002

. . . However, there are obstacles to the Power4's use: Apple would need to invest significant time in OS development to support a 64-bit environment. Moreover, software developers would need to rework their applications to take full advantage of the chip. . .



I think the article over states the difficulty. All things should work fine in 32 bits. As applications beging to take advantage of 64 bits, they will just work better. I may be mistaken, but much of the BSD Unix may already be 64 bit ready.

[nixd2001]

Originally posted by snoopy


I think the article over states the difficulty. All things should work fine in 32 bits. As applications beging to take advantage of 64 bits, they will just work better. I may be mistaken, but much of the BSD Unix may already be 64 bit ready.

Agreed. There may be effort to get extra benefit, but Apple appears sufficiently sensible (based on examining xnu source code etc) that there just shouldn't be major issues to recompile a program so that it can benefit from a 64 bit pointer. These 32/64 bit issues are not new to the industry 0 I was bashing poorly written code to make it work with 64 bit pointers on Alpha machines about 8 years ago - the world has improved a lot since then (and the BSD family has generally been pretty cleanly written code for ages anyway).

[Macrazy]

Originally posted by MacCoaster

Motorola G5 isn't prototyping anymore. It's in production; been in production for a good while. Those prices are probably the current G5 85xx. Those processors are not suited for desktop Macintoshes.

I said it was prototype a year ago. I know they are producton now.

[snoopy]

Check this out.

http://www.forbes.com/technology/newswire/2002/10/13/rtr749520.html

It is the PPC 970, a single core and available the second half of 2003. Oh yes, 1.8 GHz. We even learned that IBM will make the Playstation 3 chip, though it didn't state that directly.

[ddtlm]

snoopy, nixd2001:

I'm sure you both realize that most applications that run in 32 bits now simply do not need any more bits, and so they will continue to be 32-bit for quite some time and possibly "forever". 64-bitness will only speed up certain applications, in particular those that need more than 4GB of RAM. Native 64-bit integers are also useful for a handful of tasks but I don't think thats a big deal.

AMD's Hammer is said to be faster in 64-bit code, but that is only because it gets more general purpose registers (8) that way as opposed to the pathetic 4 x86-32 has.

[nixd2001]

Originally posted by ddtlm
snoopy, nixd2001:

I'm sure you both realize that most applications that run in 32 bits now simply do not need any more bits, and so they will continue to be 32-bit for quite some time and possibly "forever". 64-bitness will only speed up certain applications, in particular those that need more than 4GB of RAM. Native 64-bit integers are also useful for a handful of tasks but I don't think thats a big deal.

AMD's Hammer is said to be faster in 64-bit code, but that is only because it gets more general purpose registers (8) that way as opposed to the pathetic 4 x86-32 has.

Yep.

I'd expand the 4GB of RAM to 4GB of address space though. Mapping files in to memory is probably the other big gain from such a large address space. Along with some more esoteric "tagged address" applications, but we don't need to go there :D

[benixau]

Moving to 64bit asap does have its benefits. It does mean that in the future when current RAM limits are too small for the biggest of applicatinos the people at apple will be able to just up the RAM. Intel etc. will have to sell its clients a brand new processor and the vendors entirley new computers. If apple makes the move within the next year then once again, technology wise, apple will be ahead.

Who else ehas an OS that is future prrof until the year 99999???

[scem0]

I do not believe in a future proof OS.

[benixau]

Originally posted by scem0
I do not believe in a future proof OS.

Winblows is only compatible until 9999, Mac OS is compatible until 99999. Lets see that means that is your hardware lasted aslong you could theoretically get 90 000 years more usage out of a mac. hmm ....... which computer is more future proof.

Furuture proofing an OS is making sure it will run a few apps that are released for machines with better versions. Even if updates are needed, eg OS 9.2.2 with carbonlib. It is obvious which one lasts longer.

[ddtlm]

benixau:

Huh? What you have said does not reflect the real world.

[benixau]

doesnt have to, its a statement of technological superiority and foward thinking.

[MacBandit]

Originally posted by ddtlm
snoopy, nixd2001:

I'm sure you both realize that most applications that run in 32 bits now simply do not need any more bits, and so they will continue to be 32-bit for quite some time and possibly "forever". 64-bitness will only speed up certain applications, in particular those that need more than 4GB of RAM. Native 64-bit integers are also useful for a handful of tasks but I don't think thats a big deal.

AMD's Hammer is said to be faster in 64-bit code, but that is only because it gets more general purpose registers (8) that way as opposed to the pathetic 4 x86-32 has.

The programs don't necessarily ened to be written in 64bit integers. The whole process can work just like Altivecs 128bit pathway or the Sony PS2s 128bit pathway. They both can take in 4 x 32bits at a time or 8 x 16 bits. I think you get the point. What it simply does is widden the pipeline so more tasks can be taken care of at once.

[nixd2001]

Originally posted by MacBandit


The programs don't necessarily ened to be written in 64bit integers. The whole process can work just like Altivecs 128bit pathway or the Sony PS2s 128bit pathway. They both can take in 4 x 32bits at a time or 8 x 16 bits. I think you get the point. What it simply does is widden the pipeline so more tasks can be taken care of at once.

Probably not. The key thing large registers in Altivec can do is "break the carry bit". In other words, depending on the size of the "words" being operated, there is no carry from some bits to the next bit up for additions, etc. Creating this "split carry" architecture would take extra work, followed by the need to create additional instructions to use the capability. I think they'll stick with letting Altivec handle SIMD operations and leave the INT cores alone (IMHO).

[MacBandit]

Originally posted by nixd2001


Probably not. The key thing large registers in Altivec can do is "break the carry bit". In other words, depending on the size of the "words" being operated, there is no carry from some bits to the next bit up for additions, etc. Creating this "split carry" architecture would take extra work, followed by the need to create additional instructions to use the capability. I think they'll stick with letting Altivec handle SIMD operations and leave the INT cores alone (IMHO).


The Sony Playstation 2 currenty, "split carries," as you put it.

[nixd2001]

Originally posted by MacBandit



The Sony Playstation 2 currenty, "split carries," as you put it.

Within the MIPS core? Pretty funky if so.

The first I remember of split carry processors was the TMS340 graphics accelerator from Texas Instruments in the mid 90s. Become a bit more popular since then.

On the other hand, it's possible the mainframe guys did this back in the 60s - many of the "new" acceleration techniques got invented in the 60s or 70s by IBM for mainframes and then waited in the wings until microprocessors could catch up :D

[MacBandit]

Originally posted by nixd2001


Within the MIPS core? Pretty funky if so.

The first I remember of split carry processors was the TMS340 graphics accelerator from Texas Instruments in the mid 90s. Become a bit more popular since then.

On the other hand, it's possible the mainframe guys did this back in the 60s - many of the "new" acceleration techniques got invented in the 60s or 70s by IBM for mainframes and then waited in the wings until microprocessors could catch up :D

Here's an excerpt from a thread that we were discussing it in recently. I may be wrong but I think this is what you thinking of.

Ars has a nice article about the chip itself here that has a few good explanations. One quote is this:

"Finally, the Emotion Engine contains a 10-channel DMA controller (DMAC) to manage up to 10 simultaneous transfers on the Emotion Engine's internal 128-bit, 64-bit, and 16-bit buses."

and this:

"The two, fully-pipelined 64b integer ALU's are interesting, because they can either be used independently of each other (like in a normal CPU), or they can be locked together to do 128-bit integer SIMD in the following configurations: sixteen, 8-bit ops/cycle; eight, 16-bit ops/cycle; four, 32-bit ops/cycle. Pretty sweet.

To take advantage of the integer and FP SIMD capabilities that COP2 (COP2 = VU0) and the iALUs provide, Toshiba used extensions to the MIPS III ISA that include a comprehensive set of 128-bit SIMD instructions."

So it has full 128-bit busses and instructions (although they are channeled to outside chips as far as I can understand.)

Check out the thread here the inof is towards the bottom of the second page.
http://forums.macrumors.com/showthread.php?s=&threadid=11781&pagenumber=4

Also check out the info here at ArsTechnica.

http://www.arstechnica.com/reviews/1q00/playstation2/ee-1.html

[nixd2001]

Originally posted by MacBandit
Here's an excerpt from a thread that we were discussing it in recently. I may be wrong but I think this is what you thinking of.


It is the sort of thing I was thinking of. I hadn't realised the MIPS core had been tweaked that much, but I guess they've got the volume behind it to make it worth it.

I'd wager good beer that IBM won't go that route though (the catch is you've got to work out how to get the beer if I'm wrong:D ) It sounds like too much compilication for a more general purpose CPU than the hassle is probably worth.

Hopefully, we'll find out more details before the week is out. The memory bandwidth side of things is sounding on track though!

[MacBandit]

Originally posted by nixd2001


It is the sort of thing I was thinking of. I hadn't realised the MIPS core had been tweaked that much, but I guess they've got the volume behind it to make it worth it.

I'd wager good beer that IBM won't go that route though (the catch is you've got to work out how to get the beer if I'm wrong:D ) It sounds like too much compilication for a more general purpose CPU than the hassle is probably worth.

Hopefully, we'll find out more details before the week is out. The memory bandwidth side of things is sounding on track though!

Seeing how I live in Oregon the land of beer I don't think getting good beer will be a problem. I do accept paypal and billpay.

[nixd2001]

Originally posted by MacBandit


Seeing how I live in Oregon the land of beer I don't think getting good beer will be a problem. I do accept paypal and billpay.

:D

[MacBandit]

Originally posted by nixd2001


:D

Here's a new signature for you.

Dude you're getting 86'd.

Following Dells announcement to cancel the, "Dude," adds. That there description not mine. Someone at Dell actually called themt the, "Dude," adds.