Steve Jobs passed away!

[Karlos]

RIP Steve.

My general dislike of apple is no secret. However, Steve Jobbs has to be given credit for what he accomplished with the company, taking it from the relative underdog it was to the titan it is now takes smarts, passion, skill and balls. The man made technology "cool". Let's face it, portable media players, tablet computers and smart phones all existed before the iPod, iPad and iPhone but nobody can deny that the latter have become the yardstick by which all other devices in said categories have become measured.

Likewise, NeXTStep was cool, MacOS was atrocious. Nobody really remembers, or more likely wants to remember MacOS before OSX and even though I don't really like the latter I'd be disingenuous to suggest it isn't the ideal OS for many people. So, under his tenure operating systems became "cool" for a lot of people too and again OSX has become a yardstick for many.

On a personal note, having lost several close relatives in only a short space of time to particularly aggressive forms of cancer, my condolences go to his family and friends. And yet, to live as long as he did after his particular diagnosis was again, pretty remarkable.

I don't envy his successor. Whoever has to step up next for apple has pretty big shoes to fill.

[Karlos]

Heresy! Classic MacOS was awesome. Way more character than that funky Unix thing they're using now.

Classic MacOS was awful from top to bottom. NeXTStep crucified it in every way an OS can trump another OS back in the day. There's no way apple could have continued to peddle that cooperative multitasking model and it's amazing that it got as far as version 9. Lack of decent resource management, conflicting extensions, forked file systems and other foibles were eradicated with the move to OSX.

It might not have suited many users at the time, particularly when early versions of OSX underperformed on machines that 9 ran well on, but you have to judge the move on the end result. And the end result is that OSX has seriously propelled apple forwards in the desktop/workstation market. It's very hard to imagine some derivative of 9 cutting any mustard at all by present day standards.

[Karlos]

Disgusting people!

 http://idle.slashdot.org/story/11/10/06/144221/phelps-clan-tweets-intent-to-picket-jobs-funeral-via-iphone

Allegedly tweeted from her iPhone. Oh, the coppery...

[Karlos]

I used to hate Macs, but they really improved after 2000. Expensive still, and until recently rather useless for gaming, but very good hardware - even with the Intel transition.

I'd say the intel transition was one of their best moves...

[Karlos]

Intel Mac==PC

Problem? Macs were already composed of off-the-shelf PC technology except for the CPU for many years before the switch to intel. Graphics, audio, disk and network controllers were all the same stuff you'd find in a PC and were stitched together with the same bus technology.

true Macs are PPC for me and justify the extra cost.

What extra cost? PPC macs have a pretty low resale value compared to their original price tags and with good reason. PPC macs had their day and that day is long gone for most mac users. They're good for us Amiga users as they provide a hardware base for MorphOS 2, one that is comparatively inexpensive since most mac users have gone the intel route years ago.

However, if you are a Mac user, there really isn't anything you can do on some PPC mac that you can't do on an intel one for a fraction of the cost. So, PPC macs might be "true" macs for you, but I expect most of the apple user base would disagree.

[Karlos]

We're probably getting off-topic here, but I agree with Digiman. It's true that Macs used off-the-shelf parts before the switch (though that goes all the way back to the original 128K Macintosh, so I don't see it as any kind of "betrayal" of unstated principles,) but the CPU is rather a key component of a computer, and for the first two incarnations, the Mac had a CPU that was as nice as anything else about the computer, inside or out.

Was a key component, back when operating systems and applications used large amounts of assembler. Now it's a commodity component like anything else, you use the CPU that makes the most sense for the hardware you are building, not because it is a particular family; the software that will run on it can always be recompiled for your target.

Yes, there were probably solid business reasons for the switch,

Yes, that being that Motorola then IBM could not compete in either performance or cost with the relentless advances being made over on the intel side. Intel macs didn't just appear; intel builds of OSX are as old as their PPC counterparts, just kept under wraps for several years. Apple could see the writing on the wall for desktop PPC way back then and wanted to make sure they had a clean exit strategy. Apple stuck with PPC for as long as it was feasible and not a second longer. If anything, given hindsight, I am a bit surprised they never jumped sooner, considering they already had the OS ready.

It's no surprise either that they made the move to ARM for portable devices so easily too - they obviously put a fair bit of effort into making sure that OSX was platform-independent. Their OS team could probably compile it for MIPS if they were bored enough. Again, the family of CPU doesn't really matter these days - it's whatever is best for the device you are building.

but the fact remains that x86 is just plain ugly, and the only reason it's faster is because of all the revenue that can be poured back into R&D on it.

It's 2011, not 1992. I used to hate x86 too. That's all it really was, just hate for the sake of it. The truth is that like it or not, the "x86" has risen above every reasonable technical criticism that's ever been levelled against it. First it was too slow and would never survive the RISC revolution. Which it did, just fine. Turned out that all the main architectural features of RISC don't actually require a reduced instruction set in order to implement. Then it was all "it will never survive the 64-bit revolution". Erm no, if anything, it's the most popular 64-bit platform in existence, likewise the most popular multi-core platform. Then it was "too power hungry" but again, in performance per watt it's holding it's own just fine, certainly a lot better than the last PPCs that saw desktop use. Of course, ARM are still better at this game, and that's why they are used so extensively for mobile devices. However, if an x86 part with superior performance per watt and cost appeared, most hardware vendors wouldn't care about switching because, again, today CPUs are just another system component and the system software can be recompiled. The fact that for most mobile devices the rest of the software base tends to be Java based should give you some idea just how little people care about the real CPU these days.

You can get a more powerful machine for cheaper with x86, yes,

Which is what the vast majority of consumers want and hence what any business that wants market share will aim for.

but for those of us whose horsepower needs are still around what a high-end G4 or a G5 can provide, it's still nice to know that there are computers that are actually nice under the hood.

You should probably look up current x64 designs rather than thinking back to your 386-era days. There's very little not to like about them. They are clean, rational and well thought out designs. The fact that they can still run code designed for 386 era devices whilst being so radically different is a testament to how well engineered they really are.

So, ugly? Hardly.

[Karlos]

It was lulzworthy after spending a decade bashing Intel to see them whip around and give big gay hugs to x86 processors, but yeah, the Intel transition was without a doubt their shrewdest technological move.

Yeah. I was watching the keynote presentation in which the intel news was finally dropped. At the same time, they were still advertising dual G5 machines on their website with the usual litany of why PPC over x86 stuff. I had to chuckle at that.

[Karlos]

Back on topic, his showmanship certainly improved over the years but nobody can deny his enthusiasm for technology, even at the start.

Here, Steve demonstrates the original OSX
[youtube]j02b8Fuz73A[/youtube]

[Karlos]

You say that like it somehow makes an ugly architecture un-ugly. Yes, compilers can make any high-level language essentially usable on any Turing-complete architecture with sufficient memory. That's beside the point. Kludgey is still kludgey and elegant is still elegant, whether or not most people care about it. Some of us still like to use assembler, if only for hobby purposes. Some of us still care about these things.

No, I say it as if it makes an architecture irrelevant. Which it does. However, since you seem to be persisting in this "elegance" angle, there's nothing un-elegant about the x64 superset, regardless of how one feels about the old x86 ISA. In fact, you know what? If you want to go there, there's nothing particularly elegant about the PowerPC architecture. Sure, it has some nice features, particularly 3 operand instructions but a pure load-store architecture is, quite frankly, a bit of a pain in the ass at times.

Besides, architectures do still make a difference today, if less so than in the past - or weren't you paying attention when Android x86 became a full-fledged project rather than a simple cross-compile, because the original was heavily optimized for ARM?

The difference they make is far more significant in terms of the hardware they are used for and not the software they will run. Of course the software will be optimised to get the best out of the hardware, but the software itself doesn't drive the hardware choice like it once did. That's the point.

I don't give a damn about historical turf wars. I've looked at the architecture, both in its original "some day I'll be a real 32-bit chip!" incarnation and in its later forms. Too few registers and an almost-but-not-quite orthogonal approach to using them that's never entirely disappeared, for one thing. (At least they seem to have finally ditched the last vestiges of memory segmentation in x64.)

...snip...

There are 16 64-bit GPR for the x64. Which as it turns out, is about ideal as compilers have trouble making really deep optimisations for more registers than that. It's not bad for humans either. I would have loved it if the 68000 had completely general purpose registers rather than the 8 data and 8 address registers it had. There was always that one time when you ended up needing an extra register and had to temporarily juggle them. Legacy x86 instructions still use their assumed registers from that set, but you aren't forced to use them, especially when writing 64-bit code. Clean x64/SSE assembler is a far cry from the old x86/x87 code, which I'll grant was pretty nasty. You are being disingenuous to suggest otherwise.

Aaand when did we start judging quality by commercial success? If we judged movies by their box-office, Transformers 2 would be a masterpiece.

I'm sorry, I must have missed the point when equating popularity of movies to the quality of hardware design was declared a valid argument move.

The architecture got where it is now because it has adapted and always managed to beat the competitors in whatever metric actually mattered at the time, be it cost, clockspeed, power-efficiency, core count, machine width, whatever. If anything, it's track record demonstrates that in the end, the instruction set is the thing that matters least of all.

[Karlos]

True but Commodore went long before PPCs became mainstream computer processors. Apple were still on 68k when CBM went.

I think naturally, Commodore would have moved to PPC, even if it was a couple of years after Apple. It's the natural progression...

People always seem to say this and back in the day I probably thought it too, but in retrospect, is it really?

PowerPC is an implementation of the POWER architecture developed by IBM. It has nothing whatsoever to do with the 680x0. In fact, other than sharing the same native endianness, they have little in common. For example, just a few key differences:

680x0 allows arithmetic/logic operations to be performed directly on memory addresses (via indirect addressing) encoded in to the of each instruction. The PowerPC is a strictly load-store architecture.

680x0 separates address registers from data registers and segregates the operations that can be performed on them. The PowerPC treats all integer registers as general purpose (a few of which of which may have special meanings in particular contexts, but otherwise they're all the same).

68020+ defines a rich set of addressing modes, including lesser-used (though they were useful in printers apparently) memory indirect operations. The PowerPC supports only basic indirect addressing with the usual increment and decrement operations, and of course, only for load/store operations.

68881/68882/040/060 FPUs typically supported long-double types in addition to the regular float and double precision. The PowerPC FPU typically supports 32/64-bit IEEE754 only.

Bit position increments LSB -> MSB on the 680x0. On the PowerPC it's the opposite way around.

There are more differences too. Some of the above can be considered as a logical consequence of being a RISC processor (though PowerPC is not as "reduced complexity" as many RISC architectures), but others are just outright differences in design.

In all honesty, other than the endianness, the 32-bit 80x86 series (ie not running in segmented mode) probably has more in common with the 680x0 family than the PowerPC does.

Back in the day, I welcomed the appearance of PPC as an upgrade path for our systems and I still get fun out of using them but with even greater hindsight, perhaps it was not the "natural" choice. Or at least not the best choice. Not that it makes any difference now

And just a closing thought, Commodore did make expansion boards for the Amiga with non-680x0 processors on them. Guess what they were...

[Karlos]

Not sure what that has to do with Apple or Mr. Jobs......

We're 4 pages in. That's 3 pages more than the average when it comes to topic meandering

[Karlos]

Great as Jobs achievements were, his works will be forgotten by both the masses and the nerds in 30 years time.

By contrast, the achievements of Dennis Ritchie have never been known by the masses yet in 30 years the nerds will still remember him and his works.

True and I suspect the same will be true of Donald Knuth when he eventually passes away :-/

[Karlos]

Odd the article focuses on Unix... when IMHO C was his greater gift to the world... this is not to put down Unix, but rather to emphasise the importance of C.

I totally agree, but it's also worth noting that C was developed in order to help get Unix away from it's assembly language origins, so without Unix, arguably there'd probably be no C, either.