Dan Knight
- 2007.05.15
One More Response to the G3 Debate
Nathan Hill weighs in on G3 support in Leopard:
The G3 support in Leopard conversation is interesting.
I don't think there is any value in Apple keeping support for G3
machines, but the issue that might muddy the water is that the G3
and G4 are closely related chips. I remember on Mac OS X beta, you
could get the system to run on 7500/8500s and such, even though
that wasn't supposed to be possible. I figured it was because they
were already sharing a family of assembly/architecture code. I
figure, even if Leopard doesn't officially support G3 chips, some
of the system will still be useable, maybe with a few tweaks like
XPostFacto, on old machines because the G3 and G4 aren't worlds
apart in design.
Of course, I agree with other comments that the ultimate
question will still be - why run Leopard on such an old machine
anyway?
It will be interesting to see how it all shakes out in the
end.
Nathan J. Hill
www.disciplesx.com
Nathan,
Apple would have to deliberately compile the
Leopard installer to prevent it from running on a Mac without a G4
or later CPU and would also have to deliberately compile Leopard
itself to require the presence of a G4 or later to keep hacks like
XPostFacto from working. I can see them doing the first (and hope
they won't), but can't imagine them doing the latter.
Dan
Other Legacy Hardware Issues
Following up the conversation in Completely Flawed Logic, Joseph Burke
writes:
I would like to see the G3 continue to be supported as much as
anyone, but there come a time when one has to face reality. It is
not just the G3 chip itself that is the problem, it is also all the
other legacy stuff that goes with it.
Should PCI video continue to be supported in the PCI-Express
era? Do you really think that the ATI Rage or Radeon 7000 chipsets
are truly useful anymore? Is USB 1.1 still useful when everything
uses USB 2.0 now? How about machines without Firewire support?
There's plenty of people out there that would like for Apple to
release all their classic OSes to open source and allow the
community to continue development so that older machines can
continue to be supported, but will Apple do it? Not likely. It
would be nice to run an updated version of Mac OS 9 that still
supports 68k machines, but that's not going to happen.
You also have to think, does it really serve Apple's interest to
keep legacy support going forever? That is part of the reason why
Windows is as bloated and inefficient as it is. Too much legacy
support. If Apple didn't obsolete it's older hardware by dropping
support in later versions of the OS, then X would rival Windows for
lines of code and be slow and cumbersome to use, just like
Windows.
Also, Apple is a hardware company first, and a software company
second. If they kept supporting the old machines forever, how would
they sell the latest models? Everyone would just become a
Frankenmacer and continue to upgrade their old machine, secure in
the knowledge that they will never be unsupported in the
future.
Another point I forgot to make concerning Core Image. Core Image
allows the application that supports it to choose between the CPU
and GPU to perform graphics operations, whichever of the two is
faster. The G4 has Altivec; the G3 does not. Given that there are
no officially supported Core Image video cards for PCI based
machines, it is reasonable to expect that graphics would be carried
out by the CPU much of the time on a G3. Without Altivec to help it
along, this will mean huge slowdowns in graphics intense operations
as the CPU struggles to keep up with everything else it needs to be
doing at the same time. As Core Image continues to evolve and
graphics related functions become more taxing, the G3 without a
sufficiently fast graphics subsytem is just going to be too
impractical.
A line needs to be drawn somewhere, and I think it is quite
reasonable to expect that Apple will cut off the G3 and it's
associated legacy technologies.
Joseph,
It's a real mixed bag. On the one hand, we have
people who are content with their older Macs but want the latest OS
so they can run the current versions of Safari, Mail, and so forth.
On the other, we have an OS that increasingly demands far more than
these older Macs can comfortably provide.
Apple is a solutions company that makes hardware,
software, and operating systems that are all designed to work very
well together. Unlike the Windows and Linux worlds, Apple provides
"the whole widget" with the Mac - hardware, OS, and apps. I can't
call Apple a hardware company, because they make very good money
selling OS X and Mac software. Nor can I call it a software
company, because they make very good hardware and turn a nice
profit on it as well.
In the best of all possible worlds, Leopard would
draw a line between hardware that will provide adequate support for
its technologies and those that may be able to run it but can't do
so efficiently. I'm leaning toward graphics as the dividing line:
Macs with AGP video will be officially supported, while those with
PCI video won't. PCI support will be part of Leopard for those who
have added additional video cards to their G4 Power Macs, but the
installer will probably look for AGP graphics.
Time will tell where Apple decides to draw that
line in Leopard - and beyond.
Dan
Leopard's Backward Compatibility
Pete Gontier says:
This whole discussion of Leopard's backward compatibility is
moot, because Apple is much better acquainted with the constraints
of its own development process than anybody on the outside of the
firewall. But one glaring factor the discussion you recently posted
ignores is testing.
Every moment the organization spends on older
hardware is a moment which cannot be spent on newer hardware.
Setting or clearing a few check-boxes in Xcode is really not the
issue. If the code compiles and links, that doesn't mean it runs.
It has to be exercised through a full suite of tests for each
supported hardware configuration. And there are a lot of those,
especially considering the hardware tweaks Apple makes to improve
quality without advertising them. Should any test fail - and some
will - engineering needs to get involved to diagnose and fix the
problems. Every moment the organization spends on older hardware is
a moment which cannot be spent on newer hardware.
And the reality is that the mass market doesn't upgrade
operating systems; it buys new computers. Making Leopard run on any
old hardware at all is of considerably lesser value to Apple than
making it run on new hardware. Having high expectations for
Leopard's backward compatibility is probably not a good bet
regardless of how passionately one might wish one's beloved G3 box
to last forever.
- Pete Gontier
<http://www.pete.gontier.org/>
Pete,
You make a very important point about testing, but
we can't ignore how sales of Leopard to existing Mac users impacts
Apple. Piper Jaffray estimates that Apple will sell 2.6 million
copies of OS X 10.5 during its first quarter and 9 million
copies its first year. That points to about 3 million units sold as
upgrades and represents perhaps 20-25% of Mac users who have
compatible hardware.
There are millions of G3 Macs out there running
Tiger, and the people who upgraded to 10.4 two years ago are likely
candidates to upgrade to Leopard later this year. Apple has to
weigh the income from 1 million more copies of OS X sold
against the cost of testing. I somehow can't imagine them leaving
the bulk of those users behind.
Dan
Our Emotional Ties to Older Macs
Doug writes:
I agree in theory to Apple supporting earlier systems as long as
possible, but you always have to weigh the cost vs gain. Trust me -
as an owner of a 20th Anniversary and
a G4 Cube I understand how Mac users
can get attached to their computers and want to bring them along,
oftentimes longer than they should. There is no way to argue away
that emotional tie, and for that reason I hope that it is
reasonable for 10.5 to continue support for older G3 systems. Also,
your point about Mail and Safari are good ones, but there are solid
alternatives to these programs that run on earlier OS versions.
Some counter points:
- 10.5 only applications: With the exception of iLife it is
doubtful that many 10.5 only applications will show up for 6 months
to a year (making the G3 systems even that much older still). And
many of these applications will rely so heavily on GPU support that
they would be unusable on the G3 anyway.
- Hand downs: This one hurts because 10.5 iChat will be
compelling for the hand downs for support, but I think 10.4 has
matured enough that major interface changes are not in the cards.
So basic support should not be a real issue.
- 10.4 will be better on their hardware: You say this as well.
Sometimes an OS upgrade downgrades your system. 10.4 may well end
up making a better user experience for those systems.
The bottom line is that is it not easier to just leave
compatibility as you continue to innovate new features. Apple needs
to balance the cost to the number of users it will "Really" benefit
to maintain support with. The reason we are all talking about this
is that it is questionable that the G3 systems make this cut.
As I mentioned, this is becoming less of an issue as systems
continue to drop in price and become more disposable. The real down
side there is that right now Apple forces you to purchase a display
with most systems. This artificially raises the cost of simply
replacing a system (but makes hand me down's really nice). I really
wish Apple would bring out a mini/micro tower that had an entry
level price closer to $400. Use built-in video but have a slot for
PCI Express and room for two 3.5-inch drives (Time Machine will
benefit from it). Not only would this box reduce this kind of
debate. It would grow their market share 3-fold in a single year.
Making them a real threat again.
IMHO
Doug,
Your last point it your best one. If Apple were to
sell a headless Mac for even less than the cost of the Mac mini, a
lot of Windows users and a lot of Mac users with older hardware
would gladly jump onboard. Ditch the costly and generally sluggish
2.5" notebook hard drive. Use a lower cost tray-loading SuperDrive.
Put the power supply inside the box. Offer a single PCI Express
slot for video (or other) upgrades. Provide support for up to
4 GB of RAM.
Macs already have plenty of expandability thanks
for FireWire and USB 2.0, making it easy to add an external hard
drive for backup (and thus easier to store your backup offsite if
you so desire), so I don't see a need for a 2nd drive bay.
Include OS X and iLife, which are worth over $200,
include Boot Camp, and price the basic box at US$499 or less, and
the world will beat a path to Apple's door. I'd buy one today and
pay for a copy of Leopard in October.
Dan
Apple Will Leave Some Users Behind
Folowing up on Why Apple Must
Continue G3 Support in Mac OS X 10.5 'Leopard', Kris
Finkenbinder says:
Hi again, Dan,
Yes, I think you've pinpointed the core of the issue. We are on
the same page regarding the need for continuing G3 support. What we
don't agree on is the likelihood of Apple not dropping G3s.
I can't boast the long history with Apple products that yourself
and many Low End Mac readers probably can, but just in the last few
years I have noticed several points at which Apple have made
certain decisions in the interests of "moving things along". For
instance, I don't know if you've noticed, but at a certain point
all new Macs became unable to boot OS 9, not for any technical
reason but just because Apple felt it was time to put the spurs to
users and developers who still hadn't completely shifted over to
OS X. There may be a technical reason for not booting
OS 9 on G5s, but it started in the middle of the G4 line of
products. The only reason they didn't do it earlier is because they
knew they couldn't get away with it until the number of laggards
was low enough that their outcry would be ignored by the rest of
the community.
Then when they moved to Intel processors, they dropped Classic,
the ability to run OS 9 applications on top of OS X. Once
again, there is no technical reason they couldn't still allow
Classic to run on Intel, since they pulled off the magic of
Rosetta. If they can run OS X/PPC on Intel they should be able to
easily run OS 9/PPC on Intel. There is already at least one piece
of software, SheepShaver,
that shows it is technically feasible (but it only runs up to OS
9.1, so it can't take advantage of the improvements in 9.2.2).
Again, they simply chose a convenient point to make the decision,
and the switch to Intel gave them the opportunity to do so without
looking too arbitrary.
Apple made business decisions at each
juncture to advance their transition strategies, and users got left
behind.
As I'm sure you're well aware, there are still a lot of people
who want or need to continue running some critical or favorite
pieces of OS 9 software, but they are no longer able to do
that on new Macs. Even with Classic on OS X there were some
applications that just didn't work, yet they went ahead and
disabled the direct boot option. Apple made business decisions at
each juncture to advance their transition strategies, and users got
left behind. At each point the cut-off was made long before many
users expected it or found it acceptable, judging by the amount of
complaining each time. And once Apple removes something, it
generally is impossible to put it back without their support.
Things that don't exist can't be tweaked a la XPostFacto style.
I have no reason to believe this can't happen again. They
already refuse to discuss the matter of G3 support, and then there
is the removal of any mention of G3s in documentation on the
website and in ReadMe's for the Leopard betas, which is I think
fairly ominous. Not to mention the fact that the Leopard installer
already refuses to install on a G3. Why would they go to all the
trouble of modifying documents if they weren't planning on dropping
G3s? Why would they put out a beta with the intent of supporting
G3s in the final version, when the beta itself can't be used to
test G3 support? Usually the whole point of a beta is to let end
users test it on their own hardware and report bugs. But the users
are already forced to use hacks just to get it installed on a G3
machine, and they encounter many problems even after they do get it
installed. Of course they have no reason to report these problems
since G3s are no longer mentioned in the documentation as being
supported.
These are very bad signs. In fact, I wish I'd remembered to
refer to these examples from the beginning. I think in combination
it points to the strong possibility that Leopard simply will not
support G3s in any functional way, which is why I began writing you
in the first place.
I love Apple hardware and software to death since the advent of
OS X (not a big fan of the classic Mac OS, never was), but I
generally try to maintain an arms-length separation between myself
and any computer system, a sort of OS-agnostic viewpoint. I do my
best to run cross-platform software whenever possible that would
allow me to move freely between Windows, Mac OS X and even
Linux. I always try to take the long-term view when choosing
software or hardware, looking at least ten years down the line.
Because of this detachment, I can't really give Apple the
benefit of the doubt based on their previous actions in similar
circumstances. I'm not trying to imply that you've been affected by
the Reality Distortion Field™, but I do think you give Apple
too much credit when you say you refuse to believe they would drop
G3 support already. All the signs point to the fact that they plan
to do so. They may not be an Evil Monopoly®, but they are still
a business, and businesses can't be trusted no matter how lovable
and cuddly their company dictator may be. Their goal is selling new
hardware, not helping people continue to use old hardware.
This references a point that I brought up previously but didn't
state very clearly. It is instructive to ask ourselves a few
questions in this type of situation:
- What are the consequences if my assumption is wrong?
No big deal. G3 support remains, everyone's happy.
- What are the consequences if your assumption is wrong?
Disaster. Resale value of G3s plummets. Consumer confidence in
making any extensive investment in the Mac platform suffers. Cats
and dogs living together. Mass hysteria.
- What is the probability that the community voice could affect
the decision either way?
Greater than zero.
Thus, I choose to prepare for disaster, in the hopes of avoiding
it. Even if I thought it was highly unlikely that they would drop
G3 support so soon, I would maintain the same attitude toward
making sure Apple knows what we want. I believe they call this
"pessimistic optimism". Assume (and plan for) the worst while
hoping for the best. When one knows the gods are actually
listening, it behooves one to do plenty of loud praying for good
fortune.
Anyway, thanks for the lively discussion. I guess we will see in
a month or so who is right.
Cheers,
Kris F.
Hello again, Kris,
As I'm typing this in Claris Home Page, a
10-year-old app that runs beautifully in Classic mode on my
2002 Power Mac G4, I have to admit
that the lack of official Classic support is my primary obstacle to
going Macintel. I have received positive reports about SheepShaver,
but I'm not ready to go there quite yet.
I agree that Apple has made some very arbitrary
decisions. Mac OS 8.1 can run on 68030 Macs, but the installer
won't let you install it. Tiger runs on upgraded pre-G3 Power Macs,
but you have to use a hack like XPostFacto to install it.
Apple made the switch to OS X-only Macs in 2003,
and every G3-G5 Mac running OS X supports Classic, which runs
almost every classic app beautifully. The X-only Macs have new
hardware, and things like USB 2.0 are not supported in any version
of OS 9.x. As Apple's hardware and operating systems move forward,
they have to make cuts. You can't run System 6 on a G3 Mac, nor can
you boot any recent Mac into the classic Mac OS.
But what is the bad thing that happens if Apple
drops G3 support with Leopard?
- All of these older Macs will continue to run everything they
can run today. They will probably also be able to run future
versions of Firefox, Office, and other third-party apps.
- While the G3 may not be supported officially, it may still
work. Tens or hundreds of thousands of Mac users will use
XPostFacto 5 to put Leopard on their increasingly outdated
computers.
- If Leopard absolutely cannot run on G3 Macs, the value of this
equipment on the used market will drop, creating great
opportunities for those who don't need Leopard. Also, the value of
older G4 systems that are supported by Leopard will increase
temporarily. A lot of people will move up 2-3 steps in upgrading to
newer, used, low-end Macs.
As
the late Tera Patricks often said, "Nothing improves without
change." If people want the full benefits of Leopard and 10.5-only
apps, they may need to upgrade to newer hardware. And it shouldn't
be that hard, as Apple's Migration Assistant was designed for that
eventuality.
Yes, we'll mourn the loss of support for some
older Macs, including ones that shipped long before OS X first
shipped. There's still a lot of life in that old hardware, but
those who choose to continue using it will be lcoked into using a
"outdated" version of OS X that's far less outdated than their
hardware.
All in all, I expect Apple to be reasonable in
their cutoff point for OS X 10.5. We should know in June.
Dan
The Time of the G3 Is Done
Joseph Burke writes:
Just a brief note. You state that only the newest iMacs and
eMacs powerbooks and the G5 support Core Image. I
have a Digital Audio 533 that
disagrees with you. Any Radeon 9500/9600/9700/9800 video card is
Core Image capable, and there have been several of these made in
Mac versions. There is also the possibility of using a PC video
card with a flashed ROM, but since these are officially
unsupported, I don't think it's fair to count them. Also, the
Nvidia 5x00 and 6x00 cards are also Core Image compliant, but
again, those are mainly modded PC cards. So anyone with a G4 AGP
tower can get a Mac Edition Radeon 9800 Pro for officially
supported Core Image compatibility or get a modded card from the PC
side and get unofficial compatibility.
You also mention Altivec as not being relevant. Again, I have to
disagree, and it goes back to Core Image compatibility. Part of the
Core Image technology is to allow the operating system to determine
whether the GPU or the CPU would be the most efficient at rendering
graphics and goes with whichever would be the fastest solution. In
a G3 system with neither Altivec nor a Core Image compliant video
system, graphics operations will crawl. In a G4 machine you can get
away with a weak video card to a certain degree, because you have
Altivec to fall back on. In a G3, you have no such safety net.
Evolution of this same technology also works against the G3. As
Core Image continues to evolve and new technologies emerge that are
dependent on it, the G3 just isn't going to be able to keep up.
Let's also look at the overall power of most G3 systems. The
fastest G3 based machine was, I think, the iMac DV. 600 MHz is not a lot of
power. Rumor has it there was an extremely rare G3 upgrade at one
time with a 1.1 Ghz G3 chip, but the Loch Ness monster and Bigfoot
have been seen more times than it has.
Many G3 machines have already been upgraded to G4 or even
scrapped. Let's face it, you can't even give these things away
anymore, except to collectors who will tuck them away on a shelf
and never use them. The G3 iMac is a very stylish conversation
piece (especially if you have one of each color) but not much more
than that.
Now for the iBook. The G3
clamshell iBooks are the same story as the iMac. Machines to
look pretty on some collectors shelf, but not for using. Just too
slow and limited.
This brings us to the snow G3 iBook. While it is still usable
machine today for some, the snow is going to keep falling further
and further behind. It maxes out at 900 MHz.
I just looked up the requirements for Adobe Creative Suite 3.
Hmmm . . . that's funny, the G3 isn't listed. It is
supported on a 1 GHz G4, not much faster than a 900 MHz G3, so why
not on the G3? Could faster CPU clocks, Altivec, and fast graphics
cards really be so important? Yes, they are.
Nobody bemoans the lack of G3 support to Adobe, so why complain
if Apple does the same? Software requirements are only going to
continue to rise as time goes on. It is inevitable. The G3 is going
to be dropped by software devs, so it makes no sense for Apple to
continue supporting it.
The time of the G3 is done.
You also have to question the logic of continuing to release OS
version after OS version for a machine that is based on a chip that
will only get slower with each update. The point of keeping an old
Mac going is to keep it useful. If all your system resources are
devoted just to run the OS (and slowly at that), then how does this
in any way keep the older Mac relevant or useful? Demanding G3
support in Leopard goes against what (I thought) LEM stands
for.
Joseph,
We believe in options, and just as some people
choose to run newer versions of OS X on unsupported hardware,
we'd like to see Leopard able to run on G3 Macs. Not every program
needs Core Image. Not every user benefits from it. But every user
benefits from being able to run the latest version of Safari or
Mail.
The fastest G3 iMac ran at 700 MHz, and the
fastest iBook at 900 MHz. Each has AGP video with at least 16 MB of
video memory. The iBook G3/900 even has Radeon 7500 graphics on a
2x AGP bus. Yes, it's not supported by Core Image, but it works
nicely with Tiger, as thousands of users will attest.
As for your Digital Audio G4, it's disingenuous to
argue that it supports Core Image without noting that you have to
add supported AGP video card to do so. The oldest Macs that shipped
with hardware that supports Core Image are the ones I listed.
I'm hoping that Apple isn't ready to compile
Leopard so it can't run on G3 hardware, although I'm willing to
accept that they may choose not to officially support it. Then give
Mac users the opportunity to see how it fares on their older
hardware; they may choose to go back to Tiger, just as some who
installed 10.4 went back to 10.3.
Telling Apple it's okay to abandon five-year-old
Macs with Leopard goes against what LEM stands for, as it would
make it impossible for users to try the new OS, decide whether they
want what it offers, and choose for themselves whether they want to
pick up newer hardware to run it properly.
Dan
Dan Knight has been publishing Low
End Mac since April 1997. Mailbag columns come from email responses to his Mac Musings, Mac Daniel, Online Tech Journal, and other columns on the site.
