Tuesday, April 21, 2009

Bad Apple

Some new Macs have hard lockup problems independent of which operating system is running. That is, Microsoft's operating system and Apple's both exhibit the symptom.

The possibility of low-level incompatibilities between firmware of different components might explain how the problem is OS-independent. I once had a problem with an Advanced Logic Research (ALR was later bought by Compaq) machine whose multiprocessor Phoenix BIOS and the firmware of an Adaptec Ultra-SCSI card were both trying to use some of the same memory segments. Boom. The killer? Everybody involved in building the machine blamed someone else. ALR blamed IBM, IBM blamed Adaptec, Adaptec blamed Microsoft -- and the machine was running OS/2! It took months to track down. Agonizing. Eventually an Adaptec firmware update was created to solve the misuse of a memory segment used by the multiprocessor version of the ALR's Phoenix BIOS, but it sure wasted a lot of time.

This kind of Charlie Foxtrot is exactly what Apple sells buyers as not a problem with Macs because Apple is the last vertically-integrated computer manufacturer and builds the whole widget itself. Because Apple delivers products already baked to perfection, runs the argument, there's less worry some other cook will be in a position to ruin your dessert. Buy Dell, and worry about Microsoft or third-party vendors and the integration of hardware with software; but buy Apple, and It Just Works™. The current lock-up runs counter to the thesis Apple advances, and should be taken seriously.

Incidentally, because it's OS-independent, the problem probably can't be troubleshot with the new-in-Leopard debugging tools. The troubleshooting is likely to require careful observation of what the hardware components' firmware attempts to do, some of which may not involve the kernel at all.

People unhappy with the high-end video cards on these freezing Macs might be interested in this behemoth, due in May.

No comments: