Why hasn't Apple switched to x86?

[ViRGE]

Originally posted by: imgod2u

Not really. As I stated in a previous post, the SSE/SSE2 component of the P4 is 128-bit. It has a register buffer that is able to contain clusters of data (multiple FP or Integer data that combine to 128-bit) and schedule it for execution. The G4's AltiVec component is actually very similar to that of the P4. The reason the G4 is good in distributed computing, especially RC5 is due to its RISC-like nature. Without the limitations of x86 such as only 3 instruction decodes per clock and a massive integer execution engine, it can achieve much better results when performing pure integer math.

The P4 and the G4 aren't nesssarily in the same league. While you're right to the most part(especially why the P4 does indeed do poorly), the P4's SSE 128-bit powers aren't the same. I don't have the exact reason on hand, but Distributed.net's client development team said that they can't go about using the P4 the same way they can the G4 due to the registers in question. In this case, I believe it was the lack of registers on the P4(PPC chips have more registers, especially of the nice 128bit general purpose variety). Either way, excuse my rambling. Whenever someone mentions the G4 and 128-bits, I get long winded. It's not a killer CPU without a clockspeed boost, but there are some cases where its design can make it far faster than any current x86 design.

 

[Sohcan]

The reason the G4 is good in distributed computing, especially RC5 is due to its RISC-like nature

Actually most other RISC designs, excluding PowerPC, don't perform well in RC5. The reason is that they typically follow in the tradition of of the MIPS R2000 by not employing a hardware rotate instruction; RC5 makes heavy use of rotates, so they have to perform at least two shifts and an OR instead. The PowerPC and x86 ISAs, on the other hand, do feature rotates, though the P4 is the one black sheep of these implementations that does not directly execute rotates in hardware, and thus has a four cycle execute latency for rotates.

However, the fact that all RISC instructions have to be of equal length makes writing a compiler for a RISC-like architecture a huge pain.

What makes you say that? The length of the machine instruction encoding is not exposed to the compiler, but rather the assembler. And the simpler load-store architecture and larger number of registers of RISC certainly makes compiler writing easier IMHO...I did after all write a compiler for MIPS (for Toy, a simple non-OO procedure-based language used for the class). Global optimizations, loop unrolling, and software pipelining is more difficult on a strictly register-memory architecture with fewer registers.

RISC doesn't inherently make a processor more efficient, but it does allow simpler designs and whenever a design is simpler, you have less human error in the design that could hurt efficiency.

I don't think the degree of human error can be distinguished necessarily from the ISA design; any modern MPU design is subject extensive human error in the RTL, logic and circuit design stages. The original philosophy was that by relying on a load-store architecture, implementing a larger number of registers, and using fewer addressing modes, pipelining could facilitated. And by keeping the instruction set smaller, faster hardwired decoding could be used; along with the philosophy that a function only should be added in hardware if it adds a substantial performance speedup without a substantial increase in transistor count, die size could be kept to a minimum, design time could be reduced, and clock speed could be higher (especially since CISC designs of the time weren't pipelined). The RISC philosophy was certainly a success; virtually every new general-purpose ISA that I can think of designed since the early 80s has been load-store. But the original RISC pundits probably didn't consider:

- the steady tide of Moore's law would allow x86 to implement many of RISC's microarchitectural features in addition to x86 -> RISC decoding to facilitate pipelining; x86 is at or near the top of the heap in both SPECint2K and SPECfp2K

- Modern out-of-order RISC MPUs with tens or hundreds of millions of transistors would nearly level any design time advantage they have over x86

- The "memory wall" (to use the cliche) would force up to and over 1/2 of the area of modern high-performance MPUs to be occupied by cache

- Backwards compatibility can be very powerful in the software industry, especially for desktop computing.

 

[smp]

If OSX and MacOS had to support x86 .. they would have to support the plethora of hardware that is available as well ...


now, who's used final cut pro on a mac? Tell me that there is a better alternative on windows? If I was into digital video editing, you can bet that I'de pay more for a mac. Not all mac users get them to surf the web.

On a side note, surfing the web is slower on OSX and MacOS .. something about the TCP/IP stack being innificient or something .. ack .. sucks.

 

[Dark1]

Just one question why is everyone comparing a dual processor risc system (dual G4) to single processor cisc configuration? Hey the cost of purchasing the dual 1ghz G4 computers is about the same cost as purchasing a dual processor cisc (Athlon MP or Xeon) computer so at least compare those when you?fre talking about a fair comparison. Or at least a single processor based Mac compared to a single processor based PC. Personally I would really like to know,
because right now I don't see any reason to save up 2,000 or more to by a Mac if it performs just a little better then the pc. Are there any benchmarks that actually prove any of the arguments in this thread or is this all just here say?


By the way the argument that G4 is better then the P4 because it performs just as well at a lower clock speed is just like the Athlon is better then the P4 argument the difference is that when people make that argument its because the Athlon is less expensive then the P4 not the other way around.

 

[grrl]

wtf, why would they tell me differently. they had a PC USB keyboard for $26 and they wanted $65 for the mac.

Because Mac buyers are used to paying too much.

 

[imgod2u]

Originally posted by: ViRGE

The P4 and the G4 aren't nesssarily in the same league. While you're right to the most part(especially why the P4 does indeed do poorly), the P4's SSE 128-bit powers aren't the same. I don't have the exact reason on hand, but Distributed.net's client development team said that they can't go about using the P4 the same way they can the G4 due to the registers in question. In this case, I believe it was the lack of registers on the P4(PPC chips have more registers, especially of the nice 128bit general purpose variety). Either way, excuse my rambling. Whenever someone mentions the G4 and 128-bits, I get long winded. It's not a killer CPU without a clockspeed boost, but there are some cases where its design can make it far faster than any current x86 design.

That I'm aware of, the PPC does not have 128-bit general purpose registers. If it did, it would truely make it a 128-bit CPU because the general purpose registers are what determines the addressable memory size and the ability to handle larger basic data types (integers being the current limitation). The PPC does have a lot (and I mean a lot) of 128-bit SIMD registers (each capable of storing 1 SIMD data cluster). The P4, on the other hand, uses the normal FP register stack (each 80-bit) to store FP data. It is capable of forming and executing a 128-bit cluster (1 SIMD data Vector) however, it cannot store as many SIMD instructions at once as the G4 can (in fact, it cannot store as many of any data type as the G4, that's one of the advantages of leaving x86 in the trail). Does this provide better performance when it comes to working with SIMD? I don't know, but I'd say yes. Does it take up a whole lot more die space? Of course it does.

Actually most other RISC designs, excluding PowerPC, don't perform well in RC5. The reason is that they typically follow in the tradition of of the MIPS R2000 by not employing a hardware rotate instruction; RC5 makes heavy use of rotates, so they have to perform at least two shifts and an OR instead. The PowerPC and x86 ISAs, on the other hand, do feature rotates, though the P4 is the one black sheep of these implementations that does not directly execute rotates in hardware, and thus has a four cycle execute latency for rotates.

If I'm not mistaken. Intel dumped most of the hardware execution features in favor of a more generalized method of doing everything. Is there any particular instruction (perhaps an SSE instruction) that it has in to perform something similar to a rotate (possibly shift all the integer data types in an SSE vector at once)?

 

[VirtualLarry]

I wish Intel and Microsoft would just make a Quantum Leap(tm) and DUMP all the old Windows code and x86 chip "compatibility" and start over from scratch. But most people would whine they can't use their 10 year-old software if they did.
I've been saying it for a few years now - a total "do-over" with a dual-boot option right out of the box would do Windows wonders!

Maybe what we need is a new platform. New chip/archetecture, new OS, but it'd have to backed by a HUGE company to have the push to make it popular enough to have people abandon the old. Not gonna' happen.

You mean, like the VLIW (sorry, "EPIC") Itanium IA-64 chips? Backed by Intel, one of the largest multinational corporations on the PLANET?

You can see how well that platform has done on the market so far. Things should pick up though, when WinXP-64 release version is up and running on commercial Itanium 2-based servers.

 

[bluemax]

As an interesting historical note, IBM chose the 8088 processor because it was cheap and there was a glut on them on the market.
They were used in vending machines.

It was NOT chosen for superior performance or upgradability. If they knew then what we know now they would NOT have chosen the x86 processor.

 

[Sohcan]

The P4, on the other hand, uses the normal FP register stack (each 80-bit) to store FP data

You're confusing SSE1/2 with MMX and 3DNow...the latter two SIMD sets alias their 64-bit vector datatypes onto the 80-bit registers of the x87 FP stack. SSE in its implementations on the P3, P4 (including SSE2), and the Athlon XP has a completely independent 8 x 128-bit register file...this is akin to Altivec, although it has 32 architectural 128-bit SIMD registers instead of 8. Altivec, aside from any particular microarchitectural implementation, is also a bit more powerful than MMX/3DNow/SSE/SSE2 since it has four distinct operands: two source, one destination, and one filter/modifier that is handy for permute operations. The latter x86 SIMD extensions unfortunately follow in the tradition of x86/x87 by only having two operands: one source, and another that doubles as both a second source and the destination.

One important distinction between the implementations of Altivec and SSE is that while the G4 has a full 128-bit datapath to and from its vector units, the P3, P4 and Athlon have a 64-bit datapath, probably since their vector units share execution resources with their FP units. Thus to perform a 128-bit vector operation, they must be divided into two sequential 64-bit operations, therefore halving the potential throughput. These are just microarchitectural features not dependent on the SIMD instruction set specification; it is possible that a future x86 MPU could feature seperate SIMD execution units with 128-bit datapaths.

however, it cannot store as many SIMD instructions at once as the G4 can

I assume you're referring to the SIMD datatypes? Because instruction words are not stored in the register file, but rather directly fetched from the memory address space.

If I'm not mistaken. Intel dumped most of the hardware execution features in favor of a more generalized method of doing everything

I definitely wouldn't say that at all...I should clarify that the shifts/rotates on the P4 aren't necessarily performed by any sort of "software emulation," but rather the P4 does not feature a fast barrel shifter that previous x86 MPUs have had. Shifts and rotates on the P4 are handled by the third ALU rather than by one of the two double-speed ALUs; its higher latency, 4 execution cycles vs. 1 on previous MPUs, is just an artifact of the circuit design decisions and higher clock speed employed by the P4 design. Despite the increase in execution latency, the throughput (cycles between issues of the same instruction) remains at 1 like earlier designs. Likewise, some of the P4's x87 FP instructions increased in execution latency from the P3, but this is just an artifact of the the circuit design required to achieve higher clock speeds. For the most part, the instruction execution throughput remains unchanged relative to the P3, except perhaps for the simpler arithmetic and logical instructions, which effectively halved in latency and doubled in throughput on the P4 with respect to the P3 due to the two double-speed ALUs. With out-of-order execution, throughput is generally more important towards performance rather than relative instruction latency.

Only on occasion will you find on most MPUs that, except for the simpler arithmetic, logical, and memory operations, the instruction execution latency is 1 cycle. More often than not the hardware must be pipelined to achieve the desired clock cycle; even when the hardware for a complex arithmetic operation can be implemented in parallel, such as using a Wallace tree to implement multiply, it may need to be pipelined into 3 to 5 cycles or more. More complex (though less frequently used) operations, such as divides and transcendental functions, need to be executed in a more serial manner using a finite state machine (akin to, for example, long-hand division, but with techniques to speed up the number of steps) that may take 40 to over 200 cycles to execute.

 

[n0cmonkey]

Originally posted by: smp
If OSX and MacOS had to support x86 .. they would have to support the plethora of hardware that is available as well ...


now, who's used final cut pro on a mac? Tell me that there is a better alternative on windows? If I was into digital video editing, you can bet that I'de pay more for a mac. Not all mac users get them to surf the web.

On a side note, surfing the web is slower on OSX and MacOS .. something about the TCP/IP stack being innificient or something .. ack .. sucks.

Not the tcp/ip stack, but the gui... I think. They took a lot of stuff out of FreeBSD, and probably took the tcp/ip stack which is one of the best out there. I think Linux is the only real competition of the free unix-like systems. They both beat the crap out of Microsoft's tcp/ip implimentation.

 

[n0cmonkey]

Originally posted by: bluemax
As an interesting historical note, IBM chose the 8088 processor because it was cheap and there was a glut on them on the market.
They were used in vending machines.

It was NOT chosen for superior performance or upgradability. If they knew then what we know now they would NOT have chosen the x86 processor.

IBM is also a member of the group that designs the PPC chips. Motorola, Apple, and IBM are the main designers I believe. IBM's g3 processor is the better between them and Motorola. Other companies like Cisco also purchase PPC chips (g4 from Motorola, although they may also get more). Other versions of the PPC chips are embedded. The PPC chips use less power and are less hot than comparable x86 chips, which is always a plus

 

[n0cmonkey]

Besides the plethora of technical reasons keeping Apple from switching to an x86 architecture, Apple cannot change the architecture *again* because developers would be angry.

 

[n0cmonkey]

Originally posted by: BFG10K
Three main reasons:

(1) Steve Jobs is a retard.

Proof? There are several reasons I think he is quite intelligent.

(2) Apple can continue charge black market prices for PPC processors.

And you think they would charge less for x86 hardware? R&D costs money. Im sure VIA has to spend quite a few dollars developing a north/south bridge and whatnot. I dont think Apple would be able to make the money back selling cheap cheap (x86 ) systems.

(3) The RDF coming from Cupertino has convinced Mac users that the dual 1 GHz G4e system is much faster than any x86 system, which of course is utter garbage. In fact the Palomino is almost clock for clock competitive with a G4e.

I personally dont think they are faster all of the time, but in some instances they are. The RC5-64 or whatever DC project proves it. I like the architecture. I like the fact I dont have to put up with stupid x86 incompatibilities. I like the OS. I think its worth spending the money on. Doesnt mean its bad

 

[cmdrdredd]

well, let me say that most PC users know exactly crap (want to say the s word, but you know...) about macs and Apple. The fact is RICS CPUs are superior in almost every way. Clock speed does not equal better performance take the AMD/Intel war for example.

Now to be more to the point...Apple sticks to the PPC arcitecture because it's better for what Apple's market does. 1)Graphics 2)Video editing 3)sound editing. Alot of top artists in their field use Macs for their work.

tony hawk uses macs for editing his videos. Moby uses his mac for his sound editing.

Again there is no point arguing with single minded people that think what they have is always better than what everyone else has. it seems every PC user says the x86 platform is better because they can play Q3 at 300fps and the best mac only 130fps. Sure...drop them on the PC and the best player in the world on the mac and it won't matter who has the faster FPS.

 

[OneOfTheseDays]

Mac users will never understand. It doesn't matter if your CPU is more advanced and can perform better at same clock speeds. BRUTE FORCE and sheer power will always win. Look at the AMD/INTEL battle, at the moment Intel is winning because of brute force and the sheer speed advantage that they have. The Mac cannot do everything the PC can do, but the PC can do just about everything the Mac can do BETTER. I'm talking about Gaming, Photoshop, Video Rendering, you name it, the Mac is simply outperformed by the PC. This age old argument that Mac's are better for editing and video rendering is complete crap and you know it!

 

[n0cmonkey]

Originally posted by: Sudheer Anne
Mac users will never understand. It doesn't matter if your CPU is more advanced and can perform better at same clock speeds. BRUTE FORCE and sheer power will always win. Look at the AMD/INTEL battle, at the moment Intel is winning because of brute force and the sheer speed advantage that they have. The Mac cannot do everything the PC can do, but the PC can do just about everything the Mac can do BETTER. I'm talking about Gaming, Photoshop, Video Rendering, you name it, the Mac is simply outperformed by the PC. This age old argument that Mac's are better for editing and video rendering is complete crap and you know it!

It can do *EVERYTHING* better? Like run Mac OS X? I dont think so.

 

[cmdrdredd]

and a dual 1ghz Mac in OSX version of Photoshop will outperform a PC anytime. This is what the mac is made for.

everyone immediately thinks Gaming and says my PC wins, but that's not what you buy a mac for.

 

[cmdrdredd]

forgot to mention...APPLE that's right APPLE makes the toold that allow you to master DVDs. Which means the best compatability you can get. And what PC comes with Firewire built in? You almost always have to search hard for a Mobo with it or buy the card. That makes it great out of the box. FinalCut Pro and DVD Studio Pro are the best tools available for video editing and DVD mastering.

 

[BFG10K]

Proof?

The proof is in the crap he says and how it doesn't translate into the real world.

And you think they would charge less for x86 hardware?

No they wouldn't and that's why nobody would buy them and hence they'd go out of business. With PPC Apple has the desktop monopoly on that processor plus they can spew garbage about how it's faster than x86 systems and they can trick 2.4% of the world's computer users that what they're buying is superior, when in reality it isn't.

I personally dont think they are faster all of the time, but in some instances they are.

RC5 and three cherry picked photoshop filters are about the only advantages. But hey, if you want to spend $3000 for a dual 1 GHz system instead of an PC costing half the price and being twice as fast in everything else, go right ahead.

The fact is RICS CPUs are superior in almost every way. Clock speed does not equal better performance take the AMD/Intel war for example.

You don't seem to understand that an IPC advantage means nothing if the lower IPC processor has more than enough of a clock speed advantage to overcome it. The fact is that a 2.53 GHz P4 or a 2200+ XP will totally cream a 1 GHz G4e and cost about 1/4 - 1/2 the price too.

In fact as I said before, the Palomino achieves almost an IPC parity with the G4e - that is a 1 GHz Palomino (if it existed) would almost be as fast as a 1 GHz G4e. The obvious situation of course is that the Palomino is at 1.8 GHz while the G4e is at 1 GHz. And the P4 @ 2.53 GHz is faster than the 2200+ XP which means that it's even faster than the 1 GHz G4e.

AND both x86 processors are far cheaper than the PPC processor.

 

[cmdrdredd]

but you also fail to recognise the importance of a GOOD OS!

Xp is good, but the Unix core of MacOSX is soo much better. Especially with dual CPU usage...the fact is that Windows holds us back more than the hardware.

 

[imgod2u]

Originally posted by: cmdrdredd
well, let me say that most PC users know exactly crap (want to say the s word, but you know...) about macs and Apple. The fact is RICS CPUs are superior in almost every way. Clock speed does not equal better performance take the AMD/Intel war for example.

And as you see from the Athlon/P4 example, total power counts regardless of IPC (more work per clock) or clockspeed (higher MHz). Except for a SMALL group of photoshop filters, similarly priced x86 platforms would destroy any Mac in the same price range. Sure the RISC design is better than x86. But Apple and Motorola simply do not have the devotion, expertise and sheer money for developement that Intel has. What good's a nice base design if your company doesn't have the resources to pump out newer and better chips very very fast? Since AMD entered the picture, the x86 market has more than tripled in speed. How far has the PowerPC gone since the release of the 733 MHz P3? Their chips have improved in speed at most 70% in special circumstances.

Now to be more to the point...Apple sticks to the PPC arcitecture because it's better for what Apple's market does. 1)Graphics 2)Video editing 3)sound editing. Alot of top artists in their field use Macs for their work.

And? Your average consumer buys an Intel based system because they think MHz = performance. Just because the market gobbles this BS up doesn't mean it's true. Graphics? Have you seen the After Effects benchmarks? Do you know the plethora of opensource video software available for the Windows market? (Flaskmpeg and VirtualDub being 2 of the most significant that comes to mind). Whenever you has a huge base, you will have a huge array of software. At this point, software developers are porting their software (that's right, porting) to the Mac platform simply for "old times sake". 5% of the consumer market certainly doesn't justify pouring a huge amount of resources to develope it for Mac to begin with. Photoshop 7 was released for the x86 market about a month before it was available on OSX.

tony hawk uses macs for editing his videos. Moby uses his mac for his sound editing.

Again there is no point arguing with single minded people that think what they have is always better than what everyone else has. it seems every PC user says the x86 platform is better because they can play Q3 at 300fps and the best mac only 130fps. Sure...drop them on the PC and the best player in the world on the mac and it won't matter who has the faster FPS.

And practically every rendering farm to date, not to mention office, server, IT department and software developer use x86 platforms. Sure, some weirdos may use an iMac at the desk to look perty, but what do you think they send the information to to render their 3d artwork? A PowerMac or a dual or quad Xeon? Face it. There is little advantage to using the PowerPC design and all the world to gain from switching to x86. One of which being Apple won't have to spend money on R&D just to help Motorola keep their PPC chip in line with x86 processors (which costs a lot less). They make great computers and if they were to embrace the mass consumer market instead of keeping themselves in a niche, they could be quite competitive with companies like Dell or Gateway. I know if I could play all of my current software (which runs on Windows) on a Mac, I would go for a PowerMac long before I'd go for a Dell. Of course, I don't buy pre-built systems to begin with but if I were to, I would choose a Mac IF it were able to run all of my software INCLUDING all of the games and run them well.

 

[mooojojojo]

wow, there are so many uneducated responses here it's sad.
first.. about the macs being more expensive. well they are but not to the extent some people here imply. for starters the new powermacs (which are the expensive ones) come with 3level cache and dvd-burners, all the powermacs come standard with 64bit pci slots, gigabit ethernet, firewire. things you don't see standard on an average pc, and things you have to pay premium for. then comes the style of apples machines and their cases. people always pay more for style (of course when they want it).. it's like buying designers clothes.. well not exactly the same. but they have better looks. the os also.. believe it or not but the looks of the working environment can do wonders to the users mood, and if you're a designer for example, being in the right mood is pretty important for creativity and getting the right ideas. some one said that it's the same price or even cheaper to build a dual proc pc (athlon mp or xeon) as it is to buy a powermac. while this can be true for the athlon, there's no way you can build a dual xeon system for the price of a dual powermac. of course it's bad selling a standard keyboard for $65 just because it's apple, but when they know that people would still buy it.. why not. I mean everyone wants to make the highest profit (blizzard with war3 for example.. selling it for $60 because they know that it will sell well even with this high price).

now.. bfg10k said what.. jobs is a retard.. actually jobs is kind of a genius. he brought an almost dead company back to the living and even managed to make it profitable. thinking out the imac and setting trends all across the computer industry. I bet it takes more than a retard to do that. before you say that it wasn't him that did that but apple's design department, it's not so. he's notorious just for that, that he has the final word on any apple product and he's involved closely with the design and creation of each of them. and the palomino being clock for clock faster than the g4.. come on..that like saying what.. a athlon 1.2 is faster than a 800-1000mhz g4?

sudheer anne. you think that the PCs being faster (which is debatable) than the macs in video and image editing makes them necessarily better? you should think about it more before making such statements.. think about applications, ease of use, integration, traditions. there's a reason why the majority of artists (video, image, audio) use macs and not pcs. like someone said there's no real alternative (at the pricerange I guess) of finalcutpro for the pc. likewise with colorsync and a whole bunch of other applications and features. but, yes, the pc is way better for gaaming.

 

[MadRat]

*thinks back to the Intel ads with the BMG - even though they had a show run off Mac choreograph software*

 

[imgod2u]

It's amazing how many people love to rant on about their perty Macs. This thread isn't to address the benefits or weaknesses of a Mac. That would be pretty useless because religious fanboys can never be convinced. They're simply too stubborn. This thread is to address whether it'd be beneficial to Apple to switch the PROCESSOR out of their machine to an x86 processor from Intel or AMD. As I pointed out already, there is very little benefit from stubbornly sticking with a processor from Motorola when 2 companies who are practically dedicated to making CPU's are available to put in their machines. The PPC, and I'd hate to burst the bubble of many people, do NOT perform better than equivalent x86 processors currently in any task with a few exceptions such as 1 or 2 Photoshop filters and possibly some distributed computing projects. Apple has to spend money helping Motorola develope the chip just so it can keep up (and it's not doing that very well either) with x86 chips from either Intel or AMD. It'd be much better if they'd just give up trying to be the "weird computer makers" for the pure sake of being weird and focus on what would make their machines better. And currently, that is to switch to using an x86 processor like the P4 or Athlon in their computers. I think they would take a very significant chunk out of Dell and Gateway's market if they were to switch to a Windows-compatible platform and OFFER Windows on their machines. They'd also keep their current market because, let's face it, most Mac users do buy it with some rediculous notion that it's "stylish". Apple wouldn't give that up by switching processors and it certainly doesn't have to give up its closed-form design. After all, just because it works with other x86 platforms doesn't mean it has to work on a Mac. OSX doesn't HAVE to support all the hardware out there. That's what we're talking about when we say advantages of Macs right? OSX and the way Apple builds their computers (i.e. include firewire, etc.). What does any of that have to do with the processor?

 

[thorin]

I wish Intel and Microsoft would just make a Quantum Leap(tm) and DUMP all the old Windows code and x86 chip "compatibility" and start over from scratch. But most people would whine they can't use their 10 year-old software if they did.
I've been saying it for a few years now - a total "do-over" with a dual-boot option right out of the box would do Windows wonders!

What do you think Itanium and it's EPIC instruction set are? They're the Quantum leap that the Wintel world needed to make and that the consumer is scared of. But Apple proved that it could be done when they switched from the 68000 series to the current PPC line (do you know what Fat Binary is/was ?? Do you recall when all Mac software vendors sold 2 or 3 different compilations of the same software?? Did any of them fail to weather the change??" ... NO!).

Thorin