Slashdot Mirror
Apple Switching To Intel Chips In 2006
telstar writes "According to C|Net, Apple has officially decided to drop IBM, and will use Intel processors starting in their '06 line of systems. This change was rumored last month. The announcement is expected Monday at Apple's Worldwide Developer Conference in San Francisco, at which Chief Executive Steve Jobs is giving the keynote speech." From the article: "Apple successfully navigated a switch in the 1990s from Motorola's 680x0 line of processors to the Power line jointly made by Motorola and IBM. That switch also required software to be revamped to take advantage of the new processors' performance, but emulation software permitted older programs to run on the new machines."
From the report IBM, Intel and Apple declined to comment for this story. How the hell does that make official?
. . . run OS X on whatever Intel system you want, folks. I'm sure that there will be a dozen "I can't wait to put this on my blah-blah-blah Dell blah-blah-blah".
Apple is a hardware company. They will make damn sure that you can only run their software on their hardware.
I'm sure that others have surmised this. There is absolutely no way that Apple will invest the money in an expensive-for-the-consumer line of computers that will be partially obsolete in less than two years; who in their right mind would buy them?
It also occurs to me - another point that I'm sure others have already thought of - that this may be why they are forced to switch to Intel. They can't get chips small enough for a Powerbook G5 line.
You forgot that the Win in Wintel implies Windows compatibility, which this most certainly isn't.
I predict that Apple has gotten Intel to strip down its Itanium line of chips and bolt AltiVec on, as IBM did their POWER4. Remember, Intel does not necessarily equal x86 or x86-64. HP is selling iPods, and is also the premier Itanium vendor. Coincidence?
Intel has the market share so they have the R&D money as well as economies of scale. The PowerPC line is like the Sun SPARC line -- they have a limited market share and they can't afford the R&D to keep up with Intel and AMD when it comes to performance and price. Apple has less than 2% of the PC market. Intel has about 80%. When Intel sells about 40 times more CPUs, how can IBM afford the R&D to stay competitive?
Uhmmmm, ever hear of embedded processors???
This is so blatantly obvious I'm shocked we keep seeing the "Mac OS X on x86!" stories. Intel is not x86. Apple is co-owner of PowerPC. Why would it shock anyone to have Intel making PowerPCs?
Cutie Pi 3.1415926535897932384626433832795028841972....
:)
Actually, the last digit shown should be a 1, not a 2, if you are going to end with "...". The 2 is the result of rounding up because the digit that follows is a 6.
No, I didn't have to look that up.
"but I don't see what Intel has in 2006 that IBM can't match, or AMD, or whoever."
A mobile CPU that consumes 20W.
A dual-core mobile cpu (Yonah).
A dual-core desktop cpu for $240.
There's the Celeron M, which is based on the current Pentium M core (Dothan). A quick Froogle search will find boxed Celeron M processors selling for less than $100.
If the CNET article is correct and the Mac mini is one of the first to adopt Intel chips (in 2006), then I'm sure it will use the Celeron M. By early 2006, the Dothan-based Celeron M will be previous-generation technology, just like the G4 is today. Apple should have no problem fitting the Celeron M into the tiny form factor for less than $500.
TO START
PRESS ANY KEY
Where's the 'ANY' key? I see Esk, Kitarl, and Pig-Up...
I really doubt they can run PPC code on an x86 faster than it is on the PPC. Not only are new PPC close to x86 at native code, but the translation isn't easy at all. I could see a PPC doing a decent job at x86 emulation, but for the reverse there's a problem: registers. If you have a piece of PPC code that uses more registers than the x86 has (I expect this is true of any decent code), then you need to replace registers by memory (L1 at least) accesses. That will cost a lot.
Opus: the Swiss army knife of audio codec
Note that this has been fixed in Tiger. Including the "paper tape" bug.
I know it's like shooting fish in a barrel to find problems in a C|net article, but why not?
Apple has used IBM's PowerPC processors since 1994...
Nitpick: More accurately, "Apple has used PowerPC processors since 1994." The way C|net wrote it, it sounds like IBM is the only game in town until you make it halfway down the page.
The earliest PowerPC chips were from IBM, the G3s were from either Moto or IBM, and G4s were from Moto (and now Freescale). Only with the G5 has it come back to IBM's PowerPCs in a big way.
The Wall Street Journal reported last month that Apple was considering switching to Intel
No, the Wall Street Journal did not. The Wall Street Journal's rumor page -- on par with such publications at The Sun and the National Enquirer, and not intended to be taken as factual -- printed this as a rumor. Not that this stopped Reuters or anyone else from reporting it as fact.
Keep also in mind that the shadowy mystery figures in the rumor are "two industry executives with knowledge of recent discussions between the companies" -- not Apple or Intel employees. Maybe it's Darl McBride and one of his other personalities!
"I don't know that Apple's market share can survive another architecture shift. Every time they do this, they lose more customers" and more software partners, he said.
Apple has changed architectures once, from the 68K to PowerPC. This change was, for the most part, completely transparent to users and developers. Why would they lose customers over something so painless? Next thing you know Detroit will be losing customers because their latest cars have a V8 and anti-lock brakes where last year's models had a V6 and a dashboard Jesus.
Even if you count OS 9 to OS X as an "architecture" change, nobody was forced into it and OS X did and does still run OS 9 -- and earlier -- apps.
Apple shipped 1.07 million PCs in the first quarter, and its move to Intel would likely bump up the chipmaker's shipments by a corresponding amount, McCarron added.
In other news, transferring $1.07 from your checking account to your savings account is likely to raise your savings balance by $1.07.
WiMax? Sure. ARM? Sure. Hell, might Intel even be getting into the PPC biz? Stranger things have happened.
If Steve Jobs stands on the stage at the Worldwide Developers Conference and announces Apple's moving to x86, Satan will rise up from the underworld and devour the souls of every innocent puppy and kitten. And then emit the fart that ends the world. This is, of course, completely unlikely to happen, as we all know Satan prefers chunky peanut butter to the souls of small animals.
Well, one reason is that the PPC is based on the POWER architecture--which was invented by IBM in the first place.
Cthulhu for President! Why settle for the lesser evil?
The 680x0 emulator wasn't all that fast, and much of the operating system was running in emulation mode for years. Early PPC chips didn't have enough cache to contain the translation tables of the emulator, which resulted in cache trashing. And the change in floating point formats (the 68x00 floating point units could do 80-bit arithmetic, but the PPC only had 64 bits) broke all the engineering applications. Many of them never bothered to convert to PPC, and Apple exited the engineering market.
And that time, they didn't face an endian change.
If the L1 was as fast as registers, people wouldn't bother putting more registers on their chip. While the L1 can give you a throughput of 1/cycle, this doesn't count latency (~3 cycles on recent Intel chips I think), the fact that at least one x86 instruction has to be a register operant, and the fact that a PPC would *already* be using the L1 for other things.
About renaming, the PPC does it too (so it has even more registers), so you still have less registers. Also, the renaming is mainly there to allow the pipelines to work correctly. You still only have eight "logical" registers to put stuff in.
Last thing, AltiVec must be pretty hard to do efficiently on x86. First, it does twice the amount of computation as SSE does per cycle, but also because it does a MAC, which would have a 9-cycle latency if implemented in SSE.
Opus: the Swiss army knife of audio codec
How many can they support at once?
All of them.
Basically, NeXT solved the multi-architecture binary problem over a decade ago. NeXT's development tools could build for MC68K, HPPA, SPARC, and x86, just by checking the boxes for which architectures you wanted to include.
If Apple finally does ship their OS on x86, it won't be a switch, it will be an addition. This is a Solved Problem.
The experience with the G5 has been that it runs far too hot, and simply can't be clocked up as well as originally anticipated. As far as portables go, it's a dead end. Assuming Apple really does make this switch, it's the last nail in the coffin for G5 Powerbooks and Minis.
Have you tried Intel's compiler on your code? It's free (as in beer) for Linux, and often significantly faster than gcc. The auto-vectorizing is particularly good compared to gcc's efforts.
Sure:
- AMD64 performs better, in general.
- AMD64 consumes less power.
- AMD64 runs cooler.
- Socket 939 is compatible with dual-core Athlon64 chips.
Hope that was sufficient...Galileo: "The Earth revolves around the Sun!"
Score: -1 100% Flamebait
FWIW, Windows NT 3.51 (a.k.a. "Daytona") was the first PowerPC release.
A) The Mac never had a twenty percent market share, the highest penetration was just under 12%. Perhaps you are referring to the Apple II
B) If you look at the historical market share you'll note the biggest drop in share occured when Microsoft introduced Windows '95. (market share dropped slowly, about a percent a year when the Power PC was introduced and got nearly cut in half when Microsoft unleashed '95) While causality is generally difficult to assign I think this case is an exception to the rule.
But for these minor factual issues, you're all over it.
First of all, Intel is not going to do anything that could jeopardize the x86 line.
What?
The whole point of Itanium was to ditch the legacy x86 kruft and go into the future with a clean, modern architecture. Intel would love to get away from x86 if the market would let them, because it's become a huge PITA to deal with, what with all the kludges and workarounds that have been tacked on over the years.
You seem to be under the impression that x86 is all Intel does, and manufacturing some other architecture would be some huge burden for them to take on. News flash: Intel isn't a one trick pony. Ever heard of something called the i960? How about StrongARM? Xscale? Intel does all those, and a host of other, non-CPU, chips as well.
Under capitalism man exploits man. Under communism it's the other way around.