Macintosh, a Long Term Solution

This letter was written in response to news reports that the University of Waterloo in Waterloo, Ontario, Canada, was considering phasing out Macs in favor of Windows computers. This letter should be appearing soon in Imprint, the university newspaper.

Macintosh, a long term solution

Any move by the University of Waterloo away from the Macintosh and toward the Intel-Windows platform cannot be considered a long term solution. The facts speak for themselves.

1. The Macintosh operating system has been evolving since 1984. Resources permitting (memory, drive size, etc.), a 1986 Mac Plus can run many applications in use today.
2. The Macintosh OS is already a 32-bit operating system, which is exactly what Windows 95 hopes to be, but fails at because of the required backward compatibility with 16-bit Windows software. The Mac OS has always been a 32-bit OS.
3. Apple has already made the transition from a sophisticated CISC technology to the more powerful, more efficient RISC technology - all this without loss of compatibility with older software and hardware.
4. Today's Power Mac is prepared to run tomorrow's OS, whether the next version of the Mac OS, Apple's version of NeXTstep, Linux, or BeOS.
5. With programs such as SoftWindows and Virtual PC, Mac owners have the option of running both Macintosh and DOS/Windows software. For those seeking the most performance, there are also PCI cards which provide a full Pentium II processor.
6. The Mac OS has a history of being easier to use, as does Macintosh hardware. Despite the best intentions of Microsoft, Intel, Compaq, and others, the competition has not yet caught up.
7. Microsoft has a history of obsoleting otherwise viable computers via OS upgrades. For all intents and purposes, Windows 3 abandoned the 80286 and Windows 95 runs so poorly on an 80386 as to make using it a futile exercise. We can anticipate Windows 98 and Windows NT 5.0 leaving the 80486 CPU behind.
8. Intel has made its intentions to move from the 80X86 chip family to a new VLIW RISC chip code-named Merced. This 64-bit CPU is intended to ship in 1999 and replace the Pentium family, obsoleting millions of computers.
9. Apple has been building Power Macs with 64-bit* PowerPC processors for several years - without obsoleting old equipment or disrupting the ability to run legacy applications.

Clarification: The PowerPC processors have a 64-bit base architecture. The versions used in Power Macs to date are 32-bit implementations of that 64-bit architecture. Because of this, future migration to 64-bit implementations of the Power PC design should be fairly painless.

The Apple Macintosh has been and remains the best long term computing solution. They are the only computers that can run the Mac OS, versions of Unix, and Windows, something no Intel-based computer does.

Other Macintosh advantages include ease of use, ease of expansion, lower support costs, and ease of networking.

It would be very shortsighted of the University of Waterloo to abandon Macintosh support - unless it wants to create an information systems department with more staff and a larger budget.

Dan Knight
editor in chief, MacTimes
webmaster, Low End Mac

* Since first posting this editorial, I have been in correspondence with a gentleman who questions the validity of claiming the PowerPC 601, 603, 604, and 750 are 64-bit CPUs. Here are the facts:

1. The PowerPC family is based on the 64-bit architecture of the IBM RS/6000.
2. These chips all have a 64-bit data bus.
3. These chips use both 32-bit and 64-bit internal registers. (64-bit registers are for floating point numbers.)
4. These chips have internal data paths of 32 to 256 bits.
5. These chips use a 32-bit memory space.
6. Only the 603 and 603e support a 32-bit data bus, which is very inefficient.
7. Motorola and IBM promote the PowerPC as a 32-bit implementation of a 64-bit architecture.
8. 64-bit implementations of the PowerPC design will work like the current 32-bit implementations.

Based on the underlying architecture and the 64-bit data bus, I believe the PowerPC can legitimately be called a 64-bit CPU, even in current 32-bit implementations. Others maintain that the use of 32-bit registers and 32-bit memory space justify calling it a 32-bit CPU.

I think both positions are legitimate for historical reasons.

• The Intel 8086 was a 16-bit CPU. It had 16-bit registers and a 16-bit data bus. However, it only had a 12-bit address space.
• The Intel 8088 was considered a 16-bit CPU, although it was essentially an 8086 with a narrower, 8-bit data bus.
• The Motorola 680x0 family is generally considered a set of 32-bit processors, yet the 68000 had a 16-bit data bus, 24-bit address space, and 32-bit registers. (All subsequent 680x0 chips were 32-bit across the board.)
• The Intel 80386 was a 32-bit CPU, whether in its normal or SX (16-bit data bus) version. However, until the OS and applications were rewritten for 32-bit operation (something Microsoft is still nailing down with Windows), the 386 spent most of its time in a 16-bit compatibility mode.
• The Intel 80486 and Pentiums are enhanced versions of the same 32-bit architecture first used in the 80386, making them 32-bit implementations of a 32-bit architecture - even though the Pentiums have a 64-bit data bus and some 64-bit and 80-bit registers.

The simple fact is that the Intel CISC architecture is at the end of the road, forced to use increasingly powerful technology to support an increasingly outdated architecture. This is one reason Intel will be introducing the 64-bit Merced chip in 1999 - to overcome the 16-bit and 32-bit legacy of the 80X86 architecture.

The PowerPC architecture is at the start of the road. When it becomes feasible to move personal computers to the 64-bit level, migration to full 64-bit implementations of the PowerPC design will be relatively painless. We won't be moving to an entirely new architecture, as Wintel users will be forced to do.

Dan Knight, March 14, 1998

Additional Resources

• Article from The Chronicle of Higher Education