Apple entered the 21st century with a solid line of products and
remarkable growth. The only thing holding Apple back was Motorola -
they just couldn't seem to get the G4 running fast enough for Apple
to appear a contender in the Megahertz Wars.
Apple addressed that in three ways: dual-processor Power Macs
for power fiends, faster G3 models for the rest of us, and a more
powerful and efficient OS, Mac OS X. They finally broke the
500 MHz mark in early 2001 and currently (late 2001) offer a
single-processor G4/867 and dual-processor G4/800.
Where Do We Go From Here?
There are lots of things to look at in projecting the future:
processor speed, drive size, RAM capacity, displays, storage
devices, and new niches are just a few.
Moore's Law says that you can double the amount of data in a
given space every eighteen to twenty-four months. In simple terms,
that means hard drives double capacity or shrink in size, memory
chips store twice as much, processors can shrink in size (needing
less power, creating less heat, and running faster), and prices for
the increased capacity will remain relatively constant.
Thanks to Moore's Law, some parts of the Mac future can be
safely predicted. Here's a quick overview of the past 16 years:
Year Fastest Mac RAM Model
1984 8 MHz 68000 0.5 MB - Fat Mac
1985 8 MHz 68000 0.5 MB - no new model in 1985
1986 8 MHz 68000 1 MB - Mac Plus
1987 16 MHz 68020 1 MB - Mac II
1988 16 MHz 68030 1 MB - Mac IIx
1989 25 MHz 68030 2 MB - Mac IIci
1990 40 MHz 68030 4 MB - Mac IIfx
1991 25 MHz 68040 8 MB - Quadra 700, Quadra 900
1992 33 MHz 68040 8 MB - Quadra 950
1993 40 MHz 68040 8 MB - Quadra 840av
1994 110 MHz 601 8 MB - Power Mac 8100/110
1995 120 MHz 604 16 MB - Power Mac 9500/120
1996 200 MHz 604e 16 MB - Power Mac 9600/200
1997 266 MHz G3 32 MB - Power Mac G3/266
1998 366 MHz G3 32 MB - Power Mac G3/366, rare
1999 450 MHz G4 64 MB - Power Mac G4/450
2000 500 MHz G4x2 128 MB - Power Mac G4/500 MP
2001 867 MHz G4 256 MB - Power Mac G4/867
2001 800 MHz G4 256 MB - Power Mac G4/800 MP
Based on Moore's Law (which translates to 40-50% more power each
year), Apple should reach 1 GHz by Macworld Expo in January 2002.
That looks very likely, and rumors are flying that the next
generation may be announced at 1.4 GHz or even 1.6 GHz.
Taking Moore's Law conservatively, we can expect at least a 40%
improvement per year, or about 20% every six months - which is how
often we can expect new models or upgrades from Apple. Here's what
we could see in the next few years:
Date Fastest Mac RAM
1/2001 733 MHz 256 MB
1/2002 1000 MHz 256 MB
1/2003 1400 MHz 512 MB
1/2004 2000 MHz 512 MB
1/2005 3000 MHz 1 GB
Other considerations:
- The entry level computer will typically have two-thirds the
performance of the fastest model. Today, the 500 MHz iMac offers about 60% the
performance of the G4/867.
- PowerBooks will provide up to three-fourths the performance of
the fastest desktop Mac. Today, they reach 667 MHz, providing
roughly 75% as much performance as the G4/867.
- Prices of entry-level computers will be about 25-35% that of
the speed leader. You always pay a premium to have the fastest
model. And the best value will usually be the model one or two
steps below the fastest. Compare the $3,499 G4/800 MP with the $799
iMac (500 MHz), which offers about 60% the G4's performance at less
than one-quarter the price.
- The premium for the fastest PowerBook will be more reasonable,
because of the higher fixed costs of a portable computer (screen,
battery, case, drives, power supply, etc.).
New Designs
Moore's Law predicts higher speeds, more RAM, bigger hard
drives, and comparable costs. (The first Mac was $2,500, the same
as today's Power Mac G4/450 MP.)
But Apple has always innovated. The Apple II was the first
personal computer with color. Thanks to Steve Wozniak, it also had
a very affordable floppy drive. The Apple IIGS even had a built-in
synthesizer.
The Macintosh itself was a breakthrough, offering a compact
case, an internal monitor, a mouse, and a graphical interface. The
Mac II provided options of color and multiple monitors. The Mac
IIfx had dedicated I/O chips and used a radically different type of
memory. The Quadra AV models included digital signal processors.
Apple was the first to make CD-ROM a standard feature.
PowerBooks made trackballs the norm, just as later models made
touchpads common. And today's 500 MHz PowerBook is quite simply the
most powerful laptop on the market.
Apple was the first company to successfully migrate from one
family of processors to a completely different one, moving from the
680x0 family to the PowerPC in 1994. Then there's the iMac, forcing
an entire industry to support the Wintel-invented universal serial
bus (USB). They've also invented some incredible cases, developed
FireWire, and are perhaps the first to offer DVD-RAM as a factory
installed option.
Where will Apple go from here?
A small but vocal portion of the Windows community will move to
Mac OS X, which will offer the power and stability of Unix and
Linux with the ease-of-use of Windows and the traditional Mac OS.
This momentum will move Apple past 10% market share.
Someone will port OS X to Wintel hardware, where it will become
a viable alternative to both Linus and Windows, perhaps reaching
10% of the market. The hardware will not be as elegant as Apple's,
but the OS will make a lot of friends.
As market share continues to improve, Apple may once again
consider clones, but in a different light. Instead of allowing
several companies to compete with Apple's own product, they will
codevelop a handful of niche Macs bearing the "Apple Approved"
logo. This could let the Mac OS grow to 25-30% of the market by
2005, by which time the Wintel consortium may finally have settled
on a single OS (they are currently divided between consumer and pro
versions of Windows, 95/98/Me and NT/2000 respectively).
Somewhere around 2002 flat panel displays (LCD and newer
technologies) will achieve price parity with traditional monitors.
Compared side by side, the crisp, thin, low power displays will
quickly dominate the display market. At the same time, portables
will drop in price due to the incredible growth of the flat panel
market.
Apple will be the first to capitalize on this, offering a
next-generation iMac in a whole new enclosure with a 15-16" flat
panel display. With reduced shipping costs and less chance of
damage in transit, the Flat Mac will be an attractive, affordable
hit.
Another development may be MacSlate, a combination
Newton/PowerBook. The MacSlate will come in several sizes, ranging
from pocketable (like the Palm) to large enough to incorporate a
15-16" flat panel display. All models can be used in either
portrait or landscape mode, include a stylus, have USB and FireWire
for peripherals, ethernet and AirPort for networking, and work with
a wireless mouse and keyboard (optional).
Over the course of a year, the MacSlates may diminish the
PowerBook line and begin to make serious inroads against the Flat
Mac. After all, for a little more money, you have a battery and
portability.
By 2005, we'll probably have two lines: MacSlates and Power
Macs. All will probably be based on a G5 (or possibly G6)
processor, run at 2GHz or faster, and have 1 GB of RAM for best
speed and reduced dependence on the hard drive.
Gigabit ethernet will be the norm, AirPort will be in the 30Mbps
range, and FireWire will transfer data at up to 3.2Gbps. The
typical hard drive will be around 100 GB, and desktops will
typically include a multi-drive RAID array, so even if one drive
crashes, the computer will keep on going.
CD-ROM will seem as quaint as 5.25" floppies, although the
higher capacity DVD drives of the future will still be backward
compatible with that old medium.
Even 100 MB and 250 MB Zip drives will seem quaintly small, just
as floppies are today. Files in that range can easily be spurted
over the information superhighway, which makes the internet of the
1990s look like a telegraph system. For backup and high capacity
data transfers, we'll probably have something like a 4"
double-sided archival derivative of DVD-RAM capable of holding 20
GB on a single disk.
Or maybe spinning drives will give way to solid state high
capacity storage cards, data cubes, or some sort of gel pack
technology straight out of Star Trek.
Physics
Although Apple and others are funding research involving quantum
computing, which could overcome the speed of light barrier,
instantaneous communication remains a dream.
The speed of light is roughly 3.0 x 1010cm (3.0 x 108m) per second. Early personal computers,
like the Apple II and Commodore PET, ran at 1 MHz. Electricity
travels 30m (about 100') in one-millionth of a second. That's not a
problem unless your computer is huge (e.g., the one on Star Trek's
Enterprise).
At 10 MHz, that drops to 3m (10'). At 100 MHz, to 0.3m (about
12"). With today's 100 MHz motherboards, all components must be
within 30cm of each other if they are to remain in sync at 100 MHz.
In fact, because the speed of electricity in a conductor is a bit
lower than the speed of light, and to allow the design to exceed
spec, it would probably be best to build everything that must run
at 100 MHz within perhaps a 25cm diameter. (Good thing not all
components need to run that fast!)
Moore's Law predicts faster and faster processors. Today's
processors already run several times as fast as the motherboards in
our computers, with the G4 operating at up to 10x motherboard
speed.
Motherboards will also get faster. Expect a jump to 133 MHz in
2001, then 200 MHz in 2002 or so. By 2005, we may well see 2 GHz
and faster processors on 200 MHz or faster motherboards.
On a 200 MHz board, all 200 MHz components must be within about
12.5cm (5") to maintain coherency. The 2GHz processor itself must
be even smaller, with the longest internal trace under 1.25cm
(0.5") in length. This includes the on-chip level 1 cache. A
backside cache clocked at half CPU speed must be within 1" (2.5cm)
of the CPU.
Following Moore's Law, tiny powerful fast CPUs and compact
caches in near proximity shouldn't be a problem. It's just a matter
of making smaller and smaller dies.
But the tiny motherboard will present new design problems,
problems which will probably be solved in three dimensions.
Time will tell.
