Apple on Non-PPC: A Rumors Perspective

[alex_ant]

Re: Re: Re: MacOS X on alternate processor

Okay, let me just pick this apart:

Originally posted by Booga
I mean using any PowerPC chips for desktop use. Of course PowerPC chips are major players in the embedded market, but that basically means very low power chips that are underpowered for desktop or server use. POWER, a close relative, is a very high end chip that's nice if you can afford it.

But doesn't the fact that there's this great architecture that can scale from tiny handheld devices to massive servers say something? I think everyone will agree that the G4 is underpowered and that Apple needs a much faster chip, but I don't understand how that rules out a next-generation PPC chip and necessitates a leap to the "enemy" platform.

In the middle, there is only Apple using PowerPCs. There's no way they can stay cost competitive producing an order of magnitude or two fewer parts.

I'm not sure what you mean by cost competitive. In price/performance, they haven't been cost-competitive for a long time, yet they've still been able to gain market share and profit for the most part. I think if Apple were truly out to achieve cost competitiveness, they would do what e.g. Dell does: change all their hardware to bottom-dollar components, slash their margins, and ship massive numbers of very cheap computers. I don't see that as a goal for Apple; Apple is run by a man who is perfectly content selling very nice, very expensive machines to the few who can afford them, and that's actually working for them. So I don't understand the urge to commoditize their hardware; I can understand how consumers would want it, but I can't understand how it could become a sustainable business model for Apple.

1800+ Athlons are RETAILing for $100 these days (I dropped one into my PC last weekend. I'd have rather put it in my Mac.) The PowerPC originally promised double the performance or half the price. Now x86 is double the performance AT half the price. It's all economy of scale. Elegance is all nice and good, but I'd rather run on a fast x86 than an elegant but slow PowerPC. And don't kid yourself-- MacOS X would be a whole lot faster on a 2100+ Athlon than a dual 1GHz G4. Wouldn't that be great?

For a variety of reasons, I don't think OS X on x86 would be a good thing - one of the main reasons being that there is no scenario I've yet heard that would not result in the complete financial decimation of Apple, x86 Macs included. I think OS X on a much faster PPC would be a great thing, but not on commodity hardware.

Those are just my thoughts...

Alex

 

[dw1]

Apple will move to 64bit IBM Power4

http://www.mdronline.com/mpf/conf.html

From that link:
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.

Who do you think IBM added all of those vector instruction sets for???

 

[lem0nayde]

So, let me ask a question for those of you who are "in the know" with these various possible technologies. Which of the chips being discussed, from AMD, Intel, IBM, and nVidia would require the large, seemingly insane need for cooling in the photos we saw of the new supposed PowerMac enclosure?

Wouldn't that at least help to narrow down the possibilities? Or are we looking more likely at a Quad-1ghz G4 configuration to fill the interim time between now and when Apple decides where it is going (like maybe after Ocotober when IBM officially announces it's new chip).

I just wonder if any of the clues we DO have add up to more than speculation...not saying they do, just wondering.

 

[nuckinfutz]

Ahhhh Booga you were doing sooo good

Your comments about Apple not switching to a platform that is in transition are spot on but you blew it on PPC.

I mean using any PowerPC chips for desktop use. Of course PowerPC chips are major players in the embedded market, but that basically means very low power chips that are underpowered for desktop or server use. POWER, a close relative, is a very high end chip that's nice if you can afford it. In the middle, there is only Apple using PowerPCs. There's no way they can stay cost competitive producing an order of magnitude or two fewer parts.

www.ibm.com/content/home/shop_ShopIBM/en_US/eServer/pSeries/pSeries.html]Note Entry level IBM Servers using 604e chips [/URL]

Embedded markets is a Misnomer and marketing hype. Cisco, Nintendo and other all benefit from the PPC Instruction set. What they're selling is a core that is extensible. You may make the distinction between low power embedded versus whatever but the end result is the same. The Power4 is too expensive for Desktop use...soooo

Is there any suprise IBM will announce a Desktop varient with aSIMD Vector unit IBM announces 64bit 8 way superscalar proc

That sound you hear is the idea of OSX on 64bit X86 crashing to the ground. Why go through the effort if a good chip is in your backyard?

IBM will get economies of scale by using their chips in their workstation/servers. Apple will benefit from that. Don't you wonder why IBM just built a new chip foundry?

This week has yielded GREAT information. X86 for Macs is just FUD

 

[topicolo]

Originally posted by gopher
As much as some people don't like it, in order for Apple to survive, they are going to need to maintain Classic support in one form or another for several more years to come. Sometimes it takes a decade for educators to upgrade their systems. Think for a moment where we were a decade ago? The age of LCs, and some of the II machines. No PCI support, and finding an ethernet card if you didn't purchase one back then is going to be very hard today. If they want those educational institutions to upgrade to a newer Mac, they are going to need to make migration as easy as possible, and as inexpensive as possible. If they manage to pull this off without hurting some people it will be nothing short of a miracle.

In order for Apple to survive, they need to DROP support for classic as soon as possible. Without a forced move to OS X, Application developers and hardware developers will be more reluctant to produce things for the mac because they see a fragmentation of an already small market, making it even less attractive. By supporting classic, Apple also won't have the same options for the future of the mac, because it still has so much legacy equipment/software dragging it down. The sooner everybody settles down to an all OS X environment, the better. As soon as that happens (2 years?), Apple can start considering the switch to x86 processors if they still wish to, and since many chips will be 64bit by then, they can choose between the AMD Clawhammer/Opteron or the IBM produced G5 (notice how I didn't mention Mot)

 

[gopher]

Originally posted by topicolo


In order for Apple to survive, they need to DROP support for classic as soon as possible. Without a forced move to OS X, Application developers and hardware developers will be more reluctant to produce things for the mac because they see a fragmentation of an already small market, making it even less attractive. By supporting classic, Apple also won't have the same options for the future of the mac, because it still has so much legacy equipment/software dragging it down. The sooner everybody settles down to an all OS X environment, the better. As soon as that happens (2 years?), Apple can start considering the switch to x86 processors if they still wish to, and since many chips will be 64bit by then, they can choose between the AMD Clawhammer/Opteron or the IBM produced G5 (notice how I didn't mention Mot)

I disagree. Classic support is fundamentally important. Unless of course Apple can offer a Classic system that provides full support of Mac OS 9 compatible hardware without rebooting, Apple can ill afford to leave those who can't afford new machine out in the dark. At what point should Apple turn off Classic support? When the vast majority of its userbase has switched to Mac OS X. That won't happen until it becomes affordable to upgrade to newer systems for the vast majority of consumers. And when will that happen? When those machines that can run Mac OS X efficiently are easily available on the market for less than the cost of the operating system itself. When the cost of buying hardware is no different than upgrading the software. That has happened to the early PowerMacs now. So that means 7 years down the road from the release of the first AGP G4s, unless you can convince all the G3 owners that it is fast enough to run on an earlier G3 and will run all existing software. With educators strapped for cash this is going to be very very difficult. As I previously posted some school systems only upgrade their computers once every 10 years. Hardware sales are still relatively sluggish and will be until the economy picks up. When that will happen is still unclear.

 

[TechLarry]

Originally posted by zulgand04
now i know im probly wrong with what im going to say but, i'll say it anyways.

I know moto is really killing the mac with there slow development, but switching to AMD or a nightmarish thought of Intel . isn't a switch like that makeing a 180, i mean apple has stayed away from any connection to M$ compatable products. I would't mind paying an extra 200$ for a IBM processor. If apple ever whent to a AMD, or Intel Prcessor, it just would't seem like a mac any more but instead a fancy pc, with a fancy os. not a totaly diffrent computer.

Do you REALLY car what is running your software, as long as it's fast and is the SAME software with the same features and the exact same look and feel ?

Brand loyalty is an ancient thought processes. That stopped some time ago when the corporations started to cease practicing customer loyalty.

TL

 

[Booga]

Re: MacOS X on alternate processor

Okay, let me just pick this apart:

But doesn't the fact that there's this great architecture that can scale from tiny handheld devices to massive servers say something?

I am not arguing that the PowerPC is not a great, efficient, relatively elegant architecture. I'm arguing that PowerPC chips are underpowered, overpriced chips that, barring some breakthrough, will cause Apple to cede all professional work on the platform in a matter of years.

I'm not sure what you mean by cost competitive.

Then let me explain. If Apple can save $100 per CPU, and ships a couple hundred thousand units a quarter, they are saving as much money as they currently report in profits each quarter. (ie. double their profits, which is what it's all about.) There are very cheap chips out there that run rings around any current PowerPC chip, and no technical reason that Apple can't migrate to them. AltiVec is a nice too, but AMD's vector processor isn't too far behind. (By the way, I pulled the $100 price out of the air, but the last price I saw on a G4/1000 was two months ago, when it cost $300 in quantity, while an Athlon/2000 costs about half that and gets significantly better performance. At 50W, it's also not that out of line with power consumption per MHz.)

For a variety of reasons, I don't think OS X on x86 would be a good thing - one of the main reasons being that there is no scenario I've yet heard that would not result in the complete financial decimation of Apple, x86 Macs included. I think OS X on a much faster PPC would be a great thing, but not on commodity hardware.

Who said anything about commodity systems? As has been pointed out repeatedly, CPU != system. Apple could have a motherboard that uses an nVidia chipset, an x86 processor, DDR memory, and whatever else you want, and have it still not run Windows. Likewise with MOSX-- they could still have it not run on standard Wintel hardware, but take advantage of the technological advancement and cost reductions associated with economies of scale.

My current proposal is for Apple to move the server stuff first, where performance matters most. Then professionals, once the major third party apps are recompiled. Consumers can stay on the slow PowerPCs for awhile before finally moving and letting Apple either shave $100 off the price tag or earn that much more profits.

 

[topicolo]

Originally posted by gopher


I disagree. Classic support is fundamentally important. Unless of course Apple can offer a Classic system that provides full support of Mac OS 9 compatible hardware without rebooting, Apple can ill afford to leave those who can't afford new machine out in the dark. At what point should Apple turn off Classic support? When the vast majority of its userbase has switched to Mac OS X. That won't happen until it becomes affordable to upgrade to newer systems for the vast majority of consumers. And when will that happen? When those machines that can run Mac OS X efficiently are easily available on the market for less than the cost of the operating system itself. When the cost of buying hardware is no different than upgrading the software. That has happened to the early PowerMacs now. So that means 7 years down the road from the release of the first AGP G4s, unless you can convince all the G3 owners that it is fast enough to run on an earlier G3 and will run all existing software. With educators strapped for cash this is going to be very very difficult. As I previously posted some school systems only upgrade their computers once every 10 years. Hardware sales are still relatively sluggish and will be until the economy picks up. When that will happen is still unclear.

That may be what you think should happen, but if history is any indication, the forced migration to OS X will happen much sooner. When Apple first introduced the PPC macs back in (April?) of 1994, it began the switch to PPC from 68k macs. By '97-'98, essentially all research and development on hardware and software was focussed solely on PPC macs. That was only a 3-4 year time span. The introduction of OS X is of a similar magnitude as the original 68k to PPC migration and it will probably take a similar amount of time for the mac to migrate to an all OS X platform.

Besides, Steve Jobs has already shown everyone that he intends to migrate the mac completely to OS X soon when he announced that OS 9 was dead. These rumors of Pinot just add fuel to that fire

 

[nuckinfutz]

I am not arguing that the PowerPC is not a great, efficient, relatively elegant architecture. I'm arguing that PowerPC chips are underpowered, overpriced chips that, barring some breakthrough, will cause Apple to cede all professional work on the platform in a matter of years.

And that arguement would be false. If we take the performance of the G4 chips and isolate them from the rest of the computer we will see that the G4 is a good performing. Check Distributed.net for benchmarks



Athlon and Pentium scores


PPC scores

Not the most scientific but it shows that the PPC processor has more power than many of us typically think.

There are very cheap chips out there that run rings around any current PowerPC chip, and no technical reason that Apple can't migrate to them. AltiVec is a nice too, but AMD's vector processor isn't too far behind. (By the way, I pulled the $100 price out of the air, but the last price I saw on a G4/1000 was two months ago, when it cost $300 in quantity, while an Athlon/2000 costs about half that and gets significantly better performance. At 50W, it's also not that out of line with power consumption per MHz.)

I'd like to know where these chips are. Pentiums and Athlons have the advantage of Quad and Double pumped busses which definitely affects performance. The Xserve shows us that even a quasi DDR setup in a Mac can generate increased memory performance Xserve performance

The Athlon just isn't the right solution for Apple. They need a proc that can easily transition to the Portable market. PPC 74xx allow this 50W processors do not.


No matter how you slice it going X86 is not going to return enough rewards for Apple to attemtp to convince Developers to rewrite their apps.

Intel IA64 is not a given and AMD's 64Bit implementation is actually favored by some.

My thoughts are Apple needs to stay with the PPC ISA. Continue to develop and maintain their OS and apps and we'll all be fine.

PPC has gotten a bad rap unduly.

 

[Fat Tony]

Insert foot in mouth...

Could Apple really come away with using Intel chips. Especially with items like this: http://missingbite.com/details/intelmug.html
floating around??

 

[sturm375]

Originally posted by nuckinfutz


And that arguement would be false. If we take the performance of the G4 chips and isolate them from the rest of the computer we will see that the G4 is a good performing. Check Distributed.net for benchmarks



Athlon and Pentium scores


PPC scores

Not the most scientific but it shows that the PPC processor has more power than many of us typically think.

Doing a bit of investigation into these results, on said website we find this:

http://n0cgi.distributed.net/faq/cache/55.html

Both the front running G4, and the front running AMD Athlon MP are doing this RC5-64 project, which is said by the above link to be not a good benchmark.

Tip: Always be suspicious of application benchmarking. Like Photoshop, this example, DirectX benchmarking, etc.

 

[Snowy_River]

Originally posted by gopher
As much as some people don't like it, in order for Apple to survive, they are going to need to maintain Classic support in one form or another for several more years to come. Sometimes it takes a decade for educators to upgrade their systems. Think for a moment where we were a decade ago? The age of LCs, and some of the II machines. No PCI support, and finding an ethernet card if you didn't purchase one back then is going to be very hard today. If they want those educational institutions to upgrade to a newer Mac, they are going to need to make migration as easy as possible, and as inexpensive as possible. If they manage to pull this off without hurting some people it will be nothing short of a miracle.

This is exactly the problem with the OS X on x86 argument. Yes, OS X can be ported to the x86 architecture, but Classic cannot. Classic needs to continue to be supported for quite a while, or else Apple will cut a significant part of its current user base out. How soon would you buy a new computer if you knew that none of your older applications would run on it? And, then there's the issue of the OS X apps that you'd have to get a x86 compiled version of. Knowing software companies as we do, do you think that these would be free? Personally, I doubt it.

I hope that Apple sticks with the PPC architecture, just gets it properly updated (perhaps by getting away from Moto and relying more on IBM?).

 

[Snowy_River]

Originally posted by topicolo


That may be what you think should happen, but if history is any indication, the forced migration to OS X will happen much sooner. When Apple first introduced the PPC macs back in (April?) of 1994, it began the switch to PPC from 68k macs. By '97-'98, essentially all research and development on hardware and software was focussed solely on PPC macs. That was only a 3-4 year time span. The introduction of OS X is of a similar magnitude as the original 68k to PPC migration and it will probably take a similar amount of time for the mac to migrate to an all OS X platform.

Besides, Steve Jobs has already shown everyone that he intends to migrate the mac completely to OS X soon when he announced that OS 9 was dead. These rumors of Pinot just add fuel to that fire

Yes, but...

I still have some applications that I bought to run on my old Mac LC (68LC020), under system 6, and they still run under Classic mode. Stopping development on 68k code doesn't mean that 68k apps stopped running. There was still "support" for those legacy applications. On the other hand, if Classic were summarily dropped, none of our legacy applications would function anymore, no matter how recent they were. This would be a terrible move by Apple.

All they need to do is to continue to support Classic as an environment to allow us to run our legacy applications.

 

[Sun Baked]

If the majority of the CPUs are bandwidth starved, what's the point of increasing the MHz rating of the CPU yet again?

What's wrong with performance through bandwidth improvements?

The peak performance of some of these chips is amazing but, the actual sustained performance really sucks when the operations are far larger than the cache and end up stored in main memory.

So in reality wouldn't you be jumping from one CPU platform to another to be saddled with a smaller cache, a reversed endian chip, more heat, and higher power useage, and yet still have the same bandwidth problems.

 

[alex_ant]

Re: Re: MacOS X on alternate processor

Originally posted by Booga
I am not arguing that the PowerPC is not a great, efficient, relatively elegant architecture. I'm arguing that PowerPC chips are underpowered, overpriced chips that, barring some breakthrough, will cause Apple to cede all professional work on the platform in a matter of years.

Barring some breakthrough perhaps, but I disagree that the situation on the PPC front is that bleak. Apple really has a number of options besides moving to x86:

- Hardware coprocessors
- Increased reliance on low-power SMP chips
- Continuing to stick it out and waiting for the fruits of their patience to arrive, in the form of the G5, this new 64-bit chip, and other PPC variants
- Further differentiating the Mac platform from the PC platform in any way possible to shake up the playing field
- Many more I'm sure

All of which would be much less expensive and require much less effort and grief than jumping ship to x86.

Then let me explain. If Apple can save $100 per CPU, and ships a couple hundred thousand units a quarter, they are saving as much money as they currently report in profits each quarter. (ie. double their profits, which is what it's all about.) There are very cheap chips out there that run rings around any current PowerPC chip, and no technical reason that Apple can't migrate to them. AltiVec is a nice too, but AMD's vector processor isn't too far behind. (By the way, I pulled the $100 price out of the air, but the last price I saw on a G4/1000 was two months ago, when it cost $300 in quantity, while an Athlon/2000 costs about half that and gets significantly better performance. At 50W, it's also not that out of line with power consumption per MHz.)

I agree that there are much faster CPUs than current PowerPCs and that most of these are also substantially less expensive. Where I disagree is here:

Could Apple realistically sell a Mac with an x86 processor for so much more than what Dell and Gateway charge? Sure the argument is made that it's still a Mac, and therefore people are willing to pay a premium for it, but I wonder how true that would continue to be once the Mac is, under the hood, basically just an expensive PC. I think this is what would happen:

- Apple debuts an x86 Mac, with a very fast x86 chip, and sells it for $400 more than what Gateway does in order to accomodate their profit model.
- People see these two computers side-by-side in their newspaper ad and wonder why, with the same specifications, one costs $400 more than the other.
- People buy the Gateway because it's a better value.

The Mac faithful for the most part stick with Apple and their new x86 Mac, but where does that leave Apple? ... At nearly the same market share as now.

"But Sony sells a lot of upscale Vaios for more money than Gateways." Yes, but Vaios run Windows. This is actually a big plus to the common consumer. When the consumer asks whether or not all their PC software will run on the Mac, they are told,

"Yes, with the new Connectix Virtual PC for the x86 Mac. ... Which hasn't been been ported yet - expect it in a few months."
"What's 'ported?'"
"Er, it hasn't come out yet."
"Ah, okay. How much is this Virtual PC?"
"$500."
"So if I bought this Mac, I would basically be paying a $900 premium to do exactly what I can do with this Gateway?"
"Yes."

End result of a successful Apple x86 conversion: The Mac is still a niche product. And the niche is barely any bigger.

Who said anything about commodity systems? As has been pointed out repeatedly, CPU != system. Apple could have a motherboard that uses an nVidia chipset, an x86 processor, DDR memory, and whatever else you want, and have it still not run Windows. Likewise with MOSX-- they could still have it not run on standard Wintel hardware, but take advantage of the technological advancement and cost reductions associated with economies of scale.

I don't see Windows running on a Mac as a very bad thing (after all, it already does with VPC). I see OS X running on x86 a bad thing. What is happening now? People are clamoring for OS X on x86. What would happen after OS X were ported over? People would clamor for Apple to release OS X itself in a software box, so they could buy it alone. After all, the port would be so easy, and Apple would sell so many copies, people would say. Apple would not do this, because it would be suicide, of course. And so only the few and privileged would be able to afford a Mac, just the same as today. Those who want price/performance and value would continue to buy from cheap PC vendors. Apple would gain little ground as a result from their massively painstaking landmark multi-year conversion which cost them hundreds of millions of dollars.

My current proposal is for Apple to move the server stuff first, where performance matters most. Then professionals, once the major third party apps are recompiled.

Would they ever be recompiled, though? "Thanks for Carbonizing Photoshop for us, Adobe. Now we want you to port it to x86, and maintain the two separate branches of it for the forseeable future. Cheers." Really, how rude. I could imagine software developers rioting in the streets and a mass panic and feeling of abandonment of current PPC Mac users.

OS/2 failed. BeOS failed. Next/OpenStep failed. SGI tried moving to x86, with their own proprietary "next-generation" x86 architecture (if that's not an oxymoron) that had no BIOS, and that whole effort failed. In history, x86 has been a death trap for nearly all who have set foot on it. It is a dinosaur, a shining example of marketing dollars winning out over superior technology. SPECint 700 or SPECint 7000, I would be ashamed to have an x86 chip in my Mac. It would make me feel dirty. Call me sentimental, but I'm sure other Mac users feel the same way. I cannot envision a better way for Apple to: 1) Stab its loyal customers in the back, 2) Abandon a loyal chipmaking Goliath in IBM and succomb to the fleeting temptations of what's faster at the moment, and 3) drive itself into the ground than to switch its Macs over to x86. Sorry, that's just my take.

Alex

 

[Sun Baked]

Well at least this Forbes writer thinks Andrew Neff of Bear Stearns is clueless.

http://www.forbes.com/2002/08/09/0809apple.html?partner=yahoo&referrer=