You can call this a success, ignore the wise decision of Jobs to dump IBM as their chip provider, and hope IBM eventually develops some way of better accessing the multiple cores on their own chip, but right now the fact is Apple is soaring and if they develop a way of better programming multiple cores than IBM has been able to despite all the endless Cell hype, IBM will have even more egg on their face. And considering HP has already blown past them as the biggest computer company in the world, despite IBM's ballyhooed partnerships with Chinese and Japanese, they're already wearing quite a bit.
Here's an honesty challenge for you: Use the same vitriol to describe the situation with Microsoft and the XBox. They've lost about $4 billion so far, and were hoping to finally be profitable in 2008, over six years later. Producing poorly designed, tested and manufactured 360s didn't help the picture much.
Meanwhile, Sony's gaming division losses are being cut dramatically due to much lower hardware costs and massively increased sales. The PS3 will pay off a lot faster than the XBox may eventually pay off for Microsoft. Sony's initial problem was only that the PS3 was too advanced for its time, and thus it was far too expensive to manufacture and produced sticker shock among the consumers.
IBM is “clearly on the decline?”
You obviously haven’t the slightest glimmer of a clue about which you’re mouthing off. A company that grows its earnings 25% year-over-year, as IBM did in the last year, isn’t “on the decline.” A company that has over 10X as many employees, about 5X the revenue of Apple, that produces the number of bleeding-edge patents as IBM does (over 3,000 last year) isn’t about to decline any time soon.
Apple’s decision to not use the Cell processor makes no difference to IBM’s plans for Cell or other Microelectronics division. It was the next generation, the future direction of the PPC product line towards creating a “standardized CPU” for products of interest to IBM’s partners - Sony & Toshiba. The Japanese have plans in a lot of products for a range of Cell chips - not just the game consoles. You’re quite clearly suffering from tunnel vision, thinking that the desktop market is the be-all, end-all indication of the success of a processor product. This is as silly (and stupid) as I would be if I insisted that the Pentium line was going to flop because of how many times we at cisco rejected Intel’s latest product offerings. There was nothing wrong with Intel’s products, they just weren’t a fit, they cost too much for our application and they spent too much die space on features we didn’t need (eg, floating point).
For Apple, it was smart to go with Intel’s Core line - doing so allowed them to use virtualization at native speed and remove the Windows software compatibility issue as a barrier of entry into the corporate desktop market. IBM had reached an inflection point in their PowerPC product and they wanted to go in the direction of many cores on one die instead of playing the catch-up game against Intel. The larger number of cores on a single die is a good idea for video, signal processing, military applications and so on. This is a market that no one else is addressing.
There’s plenty of applications for this type of product, just not on the desktop. There’s a huge embedded CPU market out there, and Motorola used to (NB the past test) to own this. Moto has fallen on really hard times, Intel never really understood this market well, so IBM has (rightly, IMO) decided that they could pick up quite a bit of business by creating a standard library of chip designs, coupled with a CPU architecture. The Cell CPU isn’t the beginning of this product effort, and it won’t be the last, and IBM has plenty of other plans in this area which I won’t disclose because I don’t know whether they’re still covered by the NDA’s I’ve signed.
Now as to Jobs “never switching to IBM chips in the first place.” Again, you don’t know what you’re talking about.
IBM wasn’t the source of chips for Macintosh computers when Apple shifted from the Motorola 68K family to the PPC. Moto had licensed the “Power” architecture from IBM. IBM has held the patents on the “Power Architecture” since the first RISC machine, the 801. If you trouble yourself to actually study computing history, you’d see that the PowerPC’s instruction set and RISC architecture bears a very close resemblance to the 801. Where it differed was mostly in the memory management and FPU, because there was no IEEE floating point back when the 801 was designed.
So Apple, when they got the word from Moto that the 68K was reaching the end of the line, they stuck by their agreement and partnership with Motorola, went with Moto’s PowerPC products, the core design of which was licensed from IBM. At cisco, we were using 68040’s at the time and we got the same word, and we instead went with MIPS CPU’s, starting with IDT’s R4400 version of the MIPS R-4000. cisco still runs a lot of IDT/MIPS chips today for general purpose CPU’s on CPU boards in the mid-range and high end. The low-low end uses the PowerPC “QUICC” “system-on-a-chip” products.
Fast forward to the late 90’s and early 2000’s: Motorola decides that their future is in cellular phones, and following the fad at the time, they decided to spin out their Microelectronics division into Freescale Semi. Freescale announces to all and sundry that they’re no longer going to do bleeding edge CPU development, they’re going to concentrate on mass-market, low-power embedded products, especially embedded “system-on-a-chip” products.
Well, now Apple is in a lurch. They need to do one more spin of the Mac on the PPC. They needed a faster machine without changing the target architecture. So Apple has to go back to the only CPU house working on PPC architectures, and that’s IBM. Apple works with IBM to create the chip used in the G5. IBM was also the only game in town to fab the chip, and that’s how Apple ends up with IBM as a CPU supplier.
For all that effort, tho, it is pretty clear that the PPC has lost the edge in ultimate price/performance as a desktop/laptop CPU because of some very clever things that Intel’s Israeli design group did with the Pentium-M and the follow-on Core products. Once Intel did this, the prime advantage of the PPC architecture, the greater efficiency of the instruction set per clock cycle, was largely lost.
IBM wasn’t able to deliver a CPU at the price points for Apple, because Apple would have been one of the very few buyers of a CPU line with multiple chips (ie, a hot one for desktops, a low-power chip for mobile apps, etc) because everyone else was going to Intel or AMD’s 64-bit products. In chips, volume means lower prices, which means margins. Low-volume chips are margin killers, unless you can price your product in a very narrow market.
As long as Apple has been around, they’ve been fighting the “clock speed comparison” between their non-x86 CPU’s and x86 PC’s. By going to the same sort of CPU that commodity PC’s were using, all the issues of trying to explain the difference in useful throughput to customers were gone. Apple could just say “We’ve got a X GHz CPU clock” and that is that.
And that’s how Apple arrived where we are today, on Intel’s Core architecture.
re: your claim that IBM sold off the Cell. Bullcrap.
IBM has not sold off their Cell chip plants. They licensed the first Cell product to Toshiba to produce for the market. IBM has just announced (this spring) that they’ve shrunk the Cell design down for their 45nm fab. That’s not exactly something that a chip outfit does if they’re selling off the product completely. As I said previously, IBM is one of the two bleeding edge fab lines in the world. The significance of that clearly went sailing over your head at Mach-1+. When you have IBM’s bleeding edge fabs, you don’t piss away your resource pumping out commodity chips. You do what IBM now does - you produce bleeding-edge chips, and you license the older stuff to companies with less advanced fab lines. IBM and Intel’s bleeding-edge fabs represent investments of billions of dollars, and they have to be kept running at maximum capacity on chips that have the profit margins commanded by being produced at the highest speeds and highest gate densities.
As for HP: HP is noe nothing but a hollowed out shell of what they once were. The truth is that HP is a printer company that makes PC’s as a hobby. The printer line has been been HP’s real profit cow for years, and looks to remain such until one of the lower tier PC manufactures goes out of business. Did Fiorina get what she wanted? Yes. But she destroyed one of the top-of-the-line R&D and product companies in the market to turn it into a commodity PC manufacture. They’re now riding high. Big whoop. So did Dell once. Before they became a commodity PC company, they were at the top of their markets for decades, and they had a diversified product mix in computers, printers, test equipment, medical equipment, etc that allowed them to weather the various cyclical downturns and spending shifts that happen in the computing industry. HP survived the shift away from minicomputers (like their 1000’s and 3000’s) to PC’s - because they had a base of the test/medical/graphics equipment to fall back upon. No more. The next time there’s a shift in the market, they’re going to be going through the same problems as Gateway or Dell. And since they’ve lost much of their bleeding-edge R&D staff by becoming a commodity company, they have very little prospect to create completely new technologies. They’re now in a race to the bottom against the Chinese to see who can create the cheapest PC.
There’s a reason why the HP employees were dancing in their hallways when Fiorina was tossed out by the board. If McCain were to pick her for a VP candidate, you can pretty much write off McCain’s chances.
Back to the discussion at hand: antiRepublicrat brought up the Cell as an example of what can be done with the multi-core, specialized architectures of chips that are like GPU’s today. That’s all he did. You accused him of flacking for Cell, which upon re-reading the thread, is clearly an invention of your imagination. AR did no such thing, he was merely carrying on a side-discussion about the Cell as an instance of parallel processing chip with special purpose side-processors. You equate the entire computing market with the success or failure of IBM in the commodity PeeCee/Windows market, which is a profit sinkhole for any company that wants to maintain higher profit margins. IBM was smart to get out of that market, because there isn’t anywhere to go but down for profit margins in the market, and there is no point in trying to slash margins to compete with the slave labor of third world countries in slapping together commodity PC’s.
And now I’m done with you. I’ve done my best at educating you with the facts. It is up to you if you want to continue to ignore them.