Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

AMD and Apple Rumors Part II

here's a stupid question

I know just enough about Machine and OS architecture to ask this stupid question:

Why couldn't a system have both a PPC and an Athlon?

All applications interact with the upper layers of the system architecture. Those interact with the kernel. The kernel interacts with the processor. Right? Since both PPC and Athlon are Big Endian they could store their information in the same registers in the kernel. Couldn't the kernel direct x86 instructions to an Athlon and PPC instructions to the g4?

Native execution of the instructions would allow for some serious increases in clock speed. Additionally, since there would be a native X86 in the box, wouldn't they be able to exploit the work of the various WINE teams? Or for people who needed more "windows" than WINE, a "native" version of Virtual PC could be released.

<Shrug>

Flame me and list the myriad of reasons I know nothing about anything.

:)

Binky
 
OS X & x86 (repost)

This is a repost of my comments on this subject from the poll a few days ago.

----------------------------------------------------
This whole discussion is far beyond being ridiculous at this point.

There are a few simple reasons why Apple will NOT move to x86.

1. This involves much more than just 'recompiling' applications. There is the endian issue to deal with.

2. 3rd Party developers will not put up with another platform change this soon after the OS9 to OSX migration.

3. This would indicate to the market that Apple is adrift from a leadership standpoint.

4. It would lead to consumer demand for Apple to support every bit of hardware available for the PC, creating the instability issues seen on the Wintel platform.

5. Apple would have to cut prices drastically and compete with x86 hardware pricing, which would reduce margins. This would ultimately lead to layoffs at Apple and a declining stock price.

6. Microsoft would very likely refuse to port Office OS X to the x86 platform, citing lack of demand or some other excuse. Apple depends on Office and that move by Microsoft would kill Apple.

7. It would be a complete betrayal of all their current installed base. All development on PPC based apps would cease, leaving everyone with a Mac high and dry. As mentioned above, 3rd party would migrate to linux or windows development.

Apple uses their software to sell their hardware. Right now, they are in a bit of a slump because of the issues surrounding the lack of a G5 processor. It is not just Apple that is seeing decreased sales right now, it is the entire industry. If Apple is smart, they will position themselves well for an upturn in computer sales, which is turning into a cyclical market. This means getting their product line established by the end of 2003 around the next generation architectures.

The challenges with moving to a non-PPC architecture are significant and would take LONGER than waiting for the PPC970. It does not matter if OS X is ready to run on x86 if there are no applications. The PPC970 will be shipping within the next 12 months. In the mean time, we will all just continue to make do with what is available. We can also be confident that Apple will take this time to continue to improve OS X by adding new functionality and features, which will make the arrival of the PPC970 that much better.

The current G4 systems may be slower than the P4 3ghz systems. Fact is, there are not that many people rushing out to buy top of the line P4 systems, because very few people really need that processing speed. The vast majority of people I know are still running older Pentium II, or Pentium III systems, because they work just fine. A very large number of people are still running Win95 for the same reason. This delay in product cycle WILL NOT KILL APPLE. Doing something rash could.

About future processor fab choices.

AMD - They are to small and losing money. They do not have the ability to start fabbing PPC chips.

MOT - They are moving away from the PPC from all appearances

IBM - The leader in the chip fabrication industry. They are able to provide stable chip supplies and R&D to make Apple successful.

Be patient, enjoy the software updates to Jaguar and the upcoming Panther releases. Look at the other advances in hardware outside just CPU's. I firmly believe that Apple will be bringing it all together to provide the most impressive computing platform ever seen by/at MWSF '04. (Sounds like a long time, but it is only a year off.)

BTW - Go look at their financials. Their revenue levels for 2001 are just slightly off from 1999. Tho 2000 was a great year for them. I am interested to see the 2002 revenue number, but they do not indicate a significant decline in market share. I'm seeing a 10% decline in 2001 revenue from 1999 revenue.

That is why after 14 years of being a Mac-Hater, I've finally become a switcher with the release of the new 1Ghz Powerbooks....

- Kelson
 
Re: Re: Re: Is recompilation really necessary?!?

Originally posted by GPTurismo
Try using SuSE PPC bud...
Click to expand...
Okay - and I still will have less binary (pre-compiled) app's than x86 SuSE Linux does. I also will find a lot of app's source distributions don't provide a build configuration for PPC.

Originally posted by GPTurismo
PR wise, apple will shoot itself in the foot changing over to an x86 style system.

First, it's CISC, and the only place cisc is really great is in large database applications do to the slow steadiness of the system.
Click to expand...
RISC vs. CISC is an old arguement and Intel/AMD have proven that, on the desktop, it is a moot point. They've, basically, bolted a CISC-like frontend to RISC cores and are achieving great results thoguh brute-force, high MHz. Besides, what on earth makes you say a CISC based database is faster than a RISC one? Tell that to IBM and Sun who sell multi-million dollar database hardware to Fortune 100 companies. All on RISC chip machines.
(And what do you mean by "the slow seadiness of the system"?)

Originally posted by GPTurismo
Fourth, Altivec is an extended VMX, which is a risc only instruction set. Not going to see that on Intel's or AMD's CISC archetectures anytime soon
Click to expand...
What are you smoking? AltiVec is just another vector processer (albiet awesome one) - consider is MMX/SSE done right. It acutally, is kind of "no-RISC" if you ask me. RISC chip instruction sets do very small operations, very fast. AltiVec instructions are more CISC like, doing a ton of stuff at once.

Originally posted by GPTurismo
Fifth, AMD, even though I like their products, isn't highly innovative themselves. Basically everything they have built in the past 6 years has been vastly off of Intels R&D. When Intel had a major chip shortage, they had to go to AMD since they were the only ones that were capapable and willing to help them fabricate. THe big thing Intel had to do is allow AMD have access to all their RnD, and AMD has used all of that to develop all of their processors. AMD as an entity HEAVILY relies on others for their RnD, and that can be bad news in the long run. This is also one reason they don't want to leave x86 behind. Also not having to fund an extensive RnD department is one reason their costs are lower than anyone elses.
Click to expand...
I had not heard this before, do you have any links to news stories on that? (AMD fabricating for Intel) I'd be interested in reading about it.

Originally posted by GPTurismo
I think this speculation is stupid, and uneccessary. Finally, if you are so damn concerned with megahertz and thing it matters, go ahead and jump ship. Your whining is old.
Click to expand...
Frankly, if OS X ran on something other than PPC, I would. Unfortunately, it doesn't, I hate Winblows and I need MS Office applications.

Originally posted by GPTurismo
PS. Megahertz only matters to boys with small pee pees and have to compensate somehow :)
Click to expand...
And anyone rendering DV footage or 3d or large databases (where AltiVec cannot help) , or compiling larger applications (where AltiVec cannot help either)
 
Wow. What a discussion.

There is plenty of evidence toward this move-most important of all I believe is Apple's call to make all new Macs after December boot OSX only.

For all the naysayers who think Apple will never get developers to recompile-have we thought that Apple may have found a way around such an issue? Perhaps the performance gains will be such that the overhead by any form of emulation (I'm talking out of my ass here, anyone may correct me) will be negated?

Regardless, I think it would be a great move. They'll probably try it first on the XServe and then deploy it down the lineup.

Let's face it, kids-no matter what Apple's marketing schtick says about "megahertz myth", it s a load of bull. Sales of PowerMacs demonstrates clearly that Apple is providing little reason to upgrade to new hardware for all but the most power-hungry individuals.

At this point, almost any move would be a good move.
 
Just a question....

Was just reading an article from ATAT...is Amd Support same as AMD Support with MacOS 10.2 features?
 
You've obviously never actually worked in the editing business.

You're right. I have never worked in the editing business. I'm still a student.
We use both Avid Xpress DV and Final Cut Pro 3. Most of us find FCP to be more "ease of use", but still equally as powerful and it is really improving rapidly (can't wait til' #4). I know Avid make some really heavy editing solutions, but thats further up the ladder. Apple just have'nt climbed that high yet, but when they do.... maybe they will buy Avid and integrate it's technology into FCP(Not likely though), but they did buy Emagic and scrap Logic for Windows and Linux. That was one of the most preferred soundstudio programs.
I'm getting away from what this thread is really all about.
I'm afraid that AMD are also falling behind in the microprocessor race just like Moto did(remember when we actually had the fastest cpu's). AMD are loosing money and cutting jobs. Dell where supposed to use AMD Opteron cpu's in their servers, but have instead opted for Itanium 2 from Intel. That is a major blow for AMD. They are also having a hard time convincing customers that their processors are equally as fast as Intel's. Naming a processor 2800+ when it clocks at 2200 mhz seems even more desperate then what Apple has been doing. Plus the fact that it's hard to even get your hands on a 2800+(someone mentioned AMD being able to produce at higher rates than IBM?).
I'm hoping that "Big Blue" will be able to deliver next summer, so we can have some G5's in time for MWNY. I'm really excited about the 970 and so are the people at ArsTechnica and that's got to say something because they seem to know a thing or two about microarchitechture. You should check out their in-depth review. It might bring new faith into the fact that IBM is Apple's future choice of cpu's.
 
Re: OS X & x86 (repost)

Kelson, a few things:

I cannot discuss the software issue as my knowledge is not up to par in that department. But I do question some of your assumptions. First of all:


3. This would indicate to the market that Apple is adrift from a leadership standpoint.

I don't necessarily agree. Why would the adoption of another chip vendor (assuming that the inconvenience to is minimal) be seen as Apple being adrift? I see it more as being very pragmatic and intelligent in the face of Motorola's inability to provide needed performance gains.

4. It would lead to consumer demand for Apple to support every bit of hardware available for the PC, creating the instability issues seen on the Wintel platform.

Once more: how do you come to this conclusion? The boxes would still be Apple-only, not 3rd party clones. Just because it runs an AMD chip doesn't mean that people will *think* it should support the Wintel hardware world. You can be certain that Apple will make sure people do not equate AMD with a Wintel box. Just as the G4 is known as the "Velocity Engine Chip", you can be certain that any AMD chip will be named something else to keep peoples' minds off AMD and on Apple.


5. Apple would have to cut prices drastically and compete with x86 hardware pricing, which would reduce margins. This would ultimately lead to layoffs at Apple and a declining stock price.

I disagree entirely, and your argument here is baseless. As I stated before-the addition of AMD does not require Apple to become a member of the Wintel universe. Apple would push very hard to keep the brand recognition strong that they are unique, yet compatible with the Wintel universe. Just as Apple uses chips from IBM, it doesn't mean that Apple competes with IBM! The addition of AMD would permit Apple to compete with the Wintel world in means of raw performance, little else.


7. It would be a complete betrayal of all their current installed base. All development on PPC based apps would cease, leaving everyone with a Mac high and dry. As mentioned above, 3rd party would migrate to linux or windows development.

Possible-once again, I don't know enough of the technical aspects to make a commentary.


Apple uses their software to sell their hardware. Right now, they are in a bit of a slump because of the issues surrounding the lack of a G5 processor. It is not just Apple that is seeing decreased sales right now, it is the entire industry. If Apple is smart, they will position themselves well for an upturn in computer sales, which is turning into a cyclical market. This means getting their product line established by the end of 2003 around the next generation architectures.

Read the numbers, please. Apple's sales have been hardest hit. Dell has managed to increase its sales through this entire "downturn". The sales are out there, but Apple's not getting them. The market is consolidating (ergo Gateway, HP/Compaq are losing sales-but more due to those going to Dell than to a bad market). Apple does well to implement the latest technologies and architectres-and they are well positioned. But the performance issue is dogging them terribly today.
 
ClockWork

Good post. It proves that you are thinking about the isuue, instead of what other people are doing and just having a knee-jerk reaction to the possibility that Apple is switching processors.

About Avid, I was simply pointing out that Apple doesnt necessarily have the market cornered with FCP, and the reason is because they dont have horsepower to back it up. I would have to say that the vast majority of the editing market would prefer FCP over Avid anyday, if they could only see more performance. With HDTV becoming mainstream, and larger resolution become attractive to the consumer (Imax, HD-DVD's, 2k Film Quality at home...etc), Apple hardware is going to have to step it up to compete.

FCP I love---Motorola G4, I dont.
 
this is starting to sound more like the usual "Apple is dead" talk than anything else. In all honesty, who cares what processor they use? As long as people can still turn on their Mac and use it however they choose, thats whats important. Apple gears most of it's products toward the general consumer, the switch ads are clear evidence of this. Apple will do whatever is best for consumers, period. Don't try to figure out what Apple (or Steve) is doing, just sit back and enjoy the ride :cool:
 
Oops forgot to add this.

I heard in another forum within MacRumors that AMD chips were based on RISC, but you CISC emulation. This isn't the first time I've heard this. It would be interesting if true.

With Apple and AMD both being part of the HyperTransport consortium, maybe they'll use AMD to fab ppc based chips for a HyperTransport based system. However, this is extreme speculation, but that's why I'm at macrumors.

For anyone to say that Apple purchasing chips from AMD wouldn't mean much to AMD, they're wrong. AMD needs all the business they can get. Apple would be a great source of revenue for them. If you want to weigh in cost benefit then you may come up with an argument, but they wouldn't do it unless both parties come out as winners. That's just the way you do business, unless, you're Micro$oft.

Hmm... Maybe the 970 will be for the XServe and AMD PPC 64 will be for consumers. Maybe it's the other way around. Maybe AMD is designing PPC 64 chips that will be fabbed for Apple in IBM's NY plant based on HyperTransport, Alti-Vec and genetically engineered hamsters.


From someone's earlier post, I have NT for ppc.
 
Originally posted by blueBomber
this is starting to sound more like the usual "Apple is dead" talk than anything else. In all honesty, who cares what processor they use? As long as people can still turn on their Mac and use it however they choose, thats whats important. Apple gears most of it's products toward the general consumer, the switch ads are clear evidence of this. Apple will do whatever is best for consumers, period. Don't try to figure out what Apple (or Steve) is doing, just sit back and enjoy the ride :cool:
Click to expand...

Very lively and interesting thread.

I tend to fall in-line with blueBomber on this one. It makes little difference to me what processor is under the hood. So long as they keep the OS, peripherals and software running smoothly I'll stay with the Mac platform. Macs have never been the fastest or cheapest, but they have long remained the best and most innovative when you examine the platform as a whole. In past years I upgraded for speed. In recent years, when I upgrade to a new Mac it's largely been driven by Apple's addition of new features (OS X, DVD-R, iTunes, iMovie, iDVD, etc.). So, on the consumer side of things, Apple needs to do what's right to keep delivering innovative machines that work and work well. If this means switching chips, then do it; so long as it remains essentially transparent to us.
 
Larger Databases need more processor power? Then why do tell are people using large clusters of Sun systems using 700 > megahertz processors. BEcause in largescale databases you want

Reliability, easy scalability for storage, and consistancy. Larger logic requiring tasks like heavy indexing will bring any system down unless pushed off to another machine.

DV? I now people using linux and higher end apps on gigahertz p3's. Same as 3D. That's all a pipe dream that you idiots swim in. Welcome to the real world kiddies. The only reason you HAVE TO HAVE megahertz is to play games. If you are in an industry that requires such power you need to go get real equipment from Linux clusters, or even the big dogs like SGi/Irix. In clusters, megahertz per machine isn't that super important, as long as you have controllers than can maintain them which requires better coding.

It's like in cars, I don't care if your car has 1000 HP, but if it doesn't have the design and engineering and transmission, to torque conversion etc. to use that power and to handle turns, all you have is a land rocket and that is good for only one thing. Going fast. Nothing else.

GPT
 
One word makes a big difference: "server".

Yes, no matter what people will tell you, applications do have to be recompiled when you change to a CPU with a different instruction set. SUSE comes on 7 CDs, plenty of space for multiple binaries. PPC binaries will definitely NOT run on an X86 CPU, nor the reverse, without an emulation layer (think "like watching paint dry").

However, I think I heard someone say the welded-shut machines run Mac OS X Server. This makes a huge difference! Nobody runs applications on the server, they run them on client machines. When you're running Mac OS X Server, all the software you run comes from one supplier - Apple. There is no reason whatsoever Apple couldn't make an XServe box with an Athlon or two in it, and an Opteron next year. It could be cheap, fast and very good. X Server already comes with Apache and MySQL, right? Add PostgreSQL or maybe even Oracle, and you've got a box that corporate America could rack up by the thousands. Who cares what instruction set it runs inside?
 
To the people who were talking about Linux on PPC binaries/sources/whatever...

PPC Linux binaries indeed won't run on x86 Linux and vice versa.

BUT, MOST of those open source programs can work on either processor, just a compile away.

To the guy who mentioned "packages." Those are binaries. And guess what, binaries are produced from sources.

KDE works great on Linux/PPC as it does on Linux/x86.

Same concept can be applied to Mac OS X, especially with Cocoa and FAT binary Apple has been doing with 68k/PPC. Compile for both PPC and x86, then distribute.

To that dude who said Mac OS X applications 100% wouldn't run on x86. You're wrong. In fact, there is this one clone of NeXTstep's Mail program (now Mac OS X Mail), which is open source, and is done in GNUstep, works well natively with Cocoa on Mac OS X and natively with GNUstep on Linux. Same GUI, but beauty of it is menus are different in how it is implemented (vertical menus in (NeXT|GNU)step, horizontal menus in Mac OS X), but they're from the same source.

So yes, a simple recompile is possible if the developers put in time.
 
Sorry I've had it

After reading some of your comments, I'm more and more convinced that Apple blew it. The switch doesn't seem likely on their end, but it does on mine. It's really a shame that so many of us loyal Mac users are switching due to low morale!
They had the chance to get on Moto's a*s about the G5 matters but didn't. Too bad. I can't afford to wait until MWNY; I'll give them till MWSF then make my decision as to whether or not I'll invest in XServes and PowerMac towers. Until then, looks like this tired Pismo wants to be the administrator console for the servers and then a new 1Ghz PB is in my hands as my personal management computer. Now Apple, get me to your towers so it can function as my primary work machine and not a 3.06 Ghz Dell.
 
Originally posted by bluecell
I give up. Most of you seem to be missing the bigger picture. Apple only has 5% of the market. That's not going to change unless they make some changes. The sad reality is that Apple is losing out with PPC. They need OPTIONS. Anyone who has questions relating to recompiling and OS optimization should check out the article on Unsanity.org entitled MACH-O ABI. OS 9 is now a legacy OS. OS X is a whole new start for Apple. Are you going to stop using Mac when Apple starts using AMD processors? No, you won't. You'll just have a faster machine running an amazing OS.
Click to expand...
Yes, and we'd have a faster machine running an amazing OS with the PPC 970 as well. And in that case, we'd even be able to run all our old software on it!
 
Some 80 posts up...

Some 80 posts up somebody asked if programs would have to be recompiled to run on a hypothetical (and highly improbable) x86 version of Mac OS X. Some other people have stated that the answer isn't known.

The answer is:

Yes. All software (but Java apps) needs to be recompiled.
All software that talks to the hardware (drivers, anyone?) needs to be rewritten from the ground up. (Most of this is done by the OS.)
The OS itself needs much more than a simple recompile.

HP haven't managed to come up with a (functional) driver for my scanner since the introduction of OS X, but they "are working on it". Do you really believe all third-party drivers will be rewritten AGAIN?
 
Originally posted by wrkalot
Would some one please pass the popcorn?
Click to expand...

Would a mod please get rid of you and wt_brains useless comments?


It would be very nice if this rumor came true because I think we
all want a fast processor come January. But I think All the reasons
Kelson stated make good sense, but I also think BlueCell makes
a good point. I think that if apple has a simple way to migrate
to the an x86 AMD chip then they should do that, but there are
many reasons why this wouldn't be easy. But if possible, then
they should do it. I don't know what to make of this AMD support
in Jaguar stuff - but it looks promising for those that are pro-switching
to AMD.
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back