Re: I hate Linux -- Apple is leading the standards charge on the

["Bryan J. Smith" <b.j.smith@ieee.org>]

On Sat, 2004-01-31 at 10:34, Mike Connor wrote:
> ** From my casual observation, Apple doesn't adopt PC buses and
> components until it's clear that the bus will be a standard

Yes, once PCI was standardized, Apple _quickly_ moved to it.

> and that the public will no longer buy a $100 mouse and such.

Give me a break.

BTW, Apple's serial bus wasn't really any worse than what USB is.

> ** FireWire seems to be their only bus that was adopted by the PC
> world-- definitely the other way around on PCI and I think USB and
> CardBus/PCMCIA (I think I saw a Mac laptop with it once).

PCMCIA is ISA.  No way is Apple going to adopt ISA!
PCCard/CardBus is PCI, but still PC-centric.

USB and FireWire _were_ supposed to complement each other.  USB was
designed as a simple host-to-device Programmed I/O (PIO) interconnect,
and FireWire was the intelligent device-to-device bus.

Unfortunately, Intel barked at Apple's licensing, even though the IEEE
had no issue with it.  So Intel prevented FireWire's widespread adoption
throughout the PC world, by blocking it at the 1st tier PC OEM on its
mainboards.

Internal engineering memos leaked from Intel show that even Intel
engineers feel the marketing gain from this was a huge mistake, because
of USB's lack of protocol standards.  USB was designed for one thing,
to make Intel (host IC) and Microsoft (OS host driver) jobs easy.

That's why it took 3 years for USB devices to hit the market.
That's why if you have more than 1-2 USB devices on a bus, they
start to conflict (because the "brains" is in the device driver, not
the host driver).
That's why USB2.0 is much slower than FireWire between devices (because
it must go to the system's main memory, not directly device to device
like FireWire).

> ** If you buy a laptop based on what you're seeing in the linux laptop
> compatibility lists, you'll be fine.  True, it'll probably take a driver
> patch here and there.

Microsoft is trying to make laptops as incompatible with Linux as
possible.  The PC OEMs -- or, more specificaly, their Tawainese
manufacturers -- are more than happy to accommodate this as Microsoft is
willing to pay them to do so.

Luckily Microsoft underestimated the number of Linux enthusiasts (let
alone paid Linux developers) out there willing to hack and reverse
engineer the specs.  But it's still a PITA.

But it's still 10x easier with a Mac portable, no contest.  I like to
do _real_ "plug'n play."

> ** If Linux is your main OS, you don't have to pay anything at all :)

If MacOS X is your main OS, you pay for a solid BSD implementation --
one that has _more_standards_support_ of the _latest_IETF_ (among other)
capabilities.  I'm willing to pay for that, but you may not be.

> ** I imagine if you're on this list, you'd be using the unix features of
> OS X all the time anyway, so there's not much point in buying OS X and a
> $2500 (for something with a screen) one button laptop to run it on.  The
> new Apple laptops don't seem as rugged as the old ones either.

Okay, the "one button" thing has to go.  I agree.  Jobs is just stupid,
we need 3 buttons as _standard_.

But as far as "price," get off the $2,500 crap.  You can get a
_comparable_ MacOS X notebook to a PC one these days.  Most Apples are
ultra-lightweights, so compare them to PC ultra-lightweights.  And you
_can_ get a sub-$1,000 Apple notebook.  The G5 will start to change
all this too (see below).

And let's compare _peripherials_!  A _lot_ of PC OEMs are charging an
arm and a leg, some are using proprietary upgrade components (even
memory still!), whereas Apple has gone standard -- more than even select
PC OEMs.  Port replicators and docking stations are rediculous!  PC OEMs
design their USB to be purposely incompatible with many of the
"universal USB" ones out there.

Lastly, let's talk about "power v. performance."

With the G5 (IBM PowerPC 970), it was _built_ for low-power -- *UNLIKE*
the AMD Athlon32/64 and Intel P4!  While everyone drooled at the fact
that IBM's original 0.13um PowerPC970 only used 42W at 2GHz -- most
didn't stop to realize that it only used 19W at 1.2GHz!  Deaaammmnnn! 
Now that 0.09um PowerPC970s are shipping, it's even better!

With AMD Athlon32/64, you have to "slow it down" to 800MHz to get
sub-20W operation.  With Intel, the P4 has the same problem.  The Intel
alternative is to use the Centrino, which is a P3 core, which is about
the same, ~800MHz performance for sub-20W operation.

I'm sick and tired of seeing benchmarks of a PC notebook running at
"full power" wasting away 60W+ compared to a G3/G4 using less than 30W.

With the G5, 1.5GHz of 64-bit IBM PowerPC performance is sub-20W.
So my next notebook will be a PowerBook G5.

> **  The cheapest laptop PC is ~2.5GHz+

Which runs at only 800MHz when your on battery power.

> and the top-dog Mac laptop is 1.3GHz.

The G3/G4, yes.  Motorola has done a piss-poor job of performance.

That now changes with the IBM PowerPC970 -- aka Apple G5.

> I have yet to see a case where a processor could make up for a
> 100% difference in clock speed,

Bull!  The P4 is 1/2-2/3rds the speed of a P3, MHz for MHz.

AMD's Athlon32/64 now use a "speed rating" because the Athlon
benches "equivalent" against the P3, _unlike_ the P4 which is
_slower_ than the P3, MHz for MHz.

That's just talking x86.  When you move away from x86, it all
changes.  IBM's PowerPC970 is twice as simple to design for
equivalent performance.

Furthermore, you don't even need a clock!  Clocks are the _worst_thing_
introduced into ICs.  CPUs will be losing clocks very soon, because
it's a massive source of EMF and suseptible to EMI, let alone the
fact that the speed of light is too slow to travel even 1/50th of
the length of a chip in a clock cycle now.

That's not me talking.  The semiconductor trade associations have
_repeatedly_stated_ that clocks _must_ be removed by 2006, or design
failures will occur everywhere.  Timing closure in IC design due to
clocks are the #1 issue right now for IC designers.

> regardless of it being CISC/RISC/32bit/64bit.

Don't compare Motorola's lackluster G3/G4 designs to IBM's new
PowerPC970 (Apple G5).  It is an awesome design.

> Actually, I've never seen a >30% difference made
> up for; perhaps if you do broadcast grade video / extreme number
> crunching would a 2GHz G5 catch a 32bit 3GHz+ AMD/Intel.

You just wait as the 64-bit OSes and apps come out.

Maybe you should see some of the benchmarks IBM has on PowerPC970
running _Linux_ versus P4 running Linux.

You see, PowerPC970 isn't just for Apple.  It's for AIX workstations,
Linux workstations and servers as well as embedded systems.

As "economies of scale" of the PowerPC970 increases, the price will
start to be competitive.  Just wait another 12 months.



-- 
Bryan J. Smith, E.I. -- Engineer, Technologist, School Teacher
b.j.smith@ieee.org



-
To unsubscribe, send email to majordomo@silug.org with
"unsubscribe silug-discuss" in the body.