Re: Supposed speed of Intel Macs

In article <charlesbouldin-750A95.10124312012006@xxxxxxxxxxxxxxxxx>,
Charles Bouldin <charlesbouldin@xxxxxxxxxxx> wrote:

> In article <110120061801544870%star@xxxxxxx>, Davoud <star@xxxxxxx>
> wrote:
>
> > Charles Bouldin:
> >
> > > On apple's web page, the new iMac Intel is 1.6X faster on garageband
> > > and 1.9X faster on Pages than the g5 iMac. Of course, Apple could have
> > > put the dual core 2.0 G5 in the iMac and achieved roughly the same speed
> > > up....
> >
> > > Granted, this doesn't address the laptop issue, but some of the
> > > "screamer" claims for these new machines are just smoke and mirrors.
> > > It's pretty easy to get a factor of two if you have dual core in one
> > > machine and compare it with a single core system.
> >
> > Well, er, uh, yeah, but a single core system is the only /possible/
> > basis for comparison; it's not as if Apple /could/ have been building
> > faster laptops but had chosen not to do so. The fact is that the new
> > laptop -- only the first generation of its kind -- represents a very
> > significant performance gain over the fastest laptop that Apple
> > /was able to build/ before now.
> >
> > > After years of mocking SPEC, Apple is now touting SPEC as proof that
> > > Intel is faster. I take all these claims with a huge grain of salt; the
> > > real proof will come from users running programs in the real world.
> >
> > Granting all of the above, the PowerPC chip was still a dead end for
> > Apple because of IBM's divergent interests. And the laptop issue is not
> > a small one. Look at laptop sales as a percentage of all CPU sales.
> > Apple was doomed to fall farther and farther behind in this market
> > without Intel chips.
> >
> > Davoud
>
> Dead end? Xbox 360, and all the other game consoles use PPC chips that
> are closely derivative of the chips in PowerMacs. It is cinch that these
> boxes will be adapted to running Linux, and shortly after that we will
> have the comparison of how they run SPEC against the iMac Intel. I
> expect that the Xbox 360 (from MICROSOFT, oh the irony!) will utterly
> smoke the iMac Intel. It is even possible, though unlikely, that someone
> will succeed in a hack that lets OSX run on Xbox.

Derivative, but *very* different. They may do great with SPEC or other
benchmarks, and with the games that are designed to run on and coded
specifically to that processor. I'm no expert, but my understanding is
that to achieve their speed these processors depend on coordinated,
highly parallel processes that need to be executed in software. It's
doable if you're writing one application, if that application is the
only one running on the machine, if you don't need to worry about
being interrupted by other applications, and if you are coding to a
game console all of which have the exact same memory, then you can do
so effectively. If you're writing a general-purpose OS which is
running numerous processes written by many different programmers and
which needs to run on a variety of systems with different
specifications, then you're not going to see the same great
performance.

> But, we must await the data on these comparisons. I still think that the
> fairest comparison -currently- available is the dual core intel iMac
> against the dual core g5 desktop.
>
> As noted by others in this thread, Apple could have gone with the dual
> core G4 from Freescale. I expect that they didn't for two reasons: (1)
> worries about Freescale as a consistent supplier, (2) I'm not sure the
> dual core G4 variant slows the ridiculously slow FSB issue on the G4.

I think (1) is a big deal. Also, there is the plain fact that Intel
has gobs of money to throw at chip design and fab research &
construction, and has great incentive to make the fastest possible
chip for use in personal computers. Freescale doesn't have the same
resources, and if Apple's their only customer, who's to say they'll
remain focused on the personal computing market? That's part of what
happened to IBM, which turned its attention to the Cell/Xbox360
processors.

> I admit to very mixed feelings on this switch. It is another painful
> transition, it makes Macs, to all intents and purposes, just another PC,
> it eliminates much of what made Macs distinct and interesting, etc.

This I don't understand. I will agree that if, in 1990, Apple had
switched to Intel and used standard x86 motherboards, there would have
been a lot lost. NuBus was clearly superior to ISA, and had
plug-and-play functionality that ISA lacked. SCSI had great advantages
over IDE, particularly when it comes to plug-and-play of external
drives. Macs lacked parallel ports, or interrupts, or all those things
PC users had to deal with I'm sure Apple's custom-designed motherboard
controllers were better-suited to the graphics-intensive Mac computers
than its x86 counterparts at the time. It was the combination of this
unique hardware, which made Macs easier to use, and the OS which made
Macs unique.

Now, things have changed on the hardware side. Nubus has been
supplanted first by PCI and not by PCI Express, both of which are much
faster and better protocols that have the same plug-and-play ease of
NuBus. It's also a Wintel standard. SCSI has been replaced by
Firewire, which has even better plug-and-playability. Firewire is also
used in many Wintel machines. USB, and especially USB2 is far far far
better than the morass of serial and parallel ports that preceeded it,
and it is also, again, now a Wintel standard. Apple adopted all those
standards because they were better than what came before. And they are
good standards. They're also used on the Wintel side. Does that mean
Apple shouldn't have adopted them, just because they're commonplace?

So the only distinction left was the processor, and (by necessity) the
support chips on the motherboard. The fact that Macs used PowerPC
processors may have been distinguishing and interesting on a technical
level, but it makes no difference to the user experience. Like I
imagine most users, couldn't care less what's powering my Mac. All I
care about is that it does all the things that Macs do, and that it
does so effectively. And I see no indication to believe that the
Intel-based Macs fall short on that metric, or that it's going to
bring any harm to the Mac system.

> OTOH, it presents the chance for a single machine that runs OSX and
> Windows. It eliminates the cpu supply and performance issues that have
> plagued Apple since the days of the 68020.

All true.

- geoff
.

Relevant Pages

  • Re: MBP wont charge
    ... I have been given many computers - apple and not. ... a laptop I was given is 'crap', I joke with you about it then you ... broken to prove how stupid people are to buy macs, ... I think insisting on a fully working PSU to go with it might have ...
    (uk.comp.sys.mac)
  • Re: Apple/Freescale
    ... >>> Apple told us the conversion would be complete in 2007. ... >>> more PPC Macs? ... And what makes you think they're unreliable - when Macs have the highest ... >>> to no longer need PPC chips? ...
    (comp.sys.mac.advocacy)
  • Re: OS X for the great unwashed
    ... IBM has some very nice multi core post G5 chips that would have served ... Apple well but they would have cost more per unit and in integration ... Apple does lock down the OS to work on only the Macs but not with TCPA. ... not design the hardware and OS. ...
    (comp.sys.mac.advocacy)
  • Re: For those who might want a small PC on thier cart.
    ... burner that my Macbook has will do fine in my laptop PC, I ask an Apple ... FireWire is now 800 (as with many of the new Macs. ...
    (rec.arts.movies.production.sound)
  • Re: FAO Rowland
    ... Which is what Apple did with MacOS X (other firms have ... - it works very well on Macs. ... computers; Microsoft began by selling a bad version of the Basic ... Basic and I was doing that before IBM PCs came out or anyone had dreamt ...
    (uk.people.support.depression)