That's a huge win!

This raises an interesting question for me.  Is anyone doing any
serious performance work on the 4c ?

The reason I ask is that someone just pointed me at a paper on
"Hardware Profiling of Kernels", Andrew McRae, USENIX Jan 93.

A precis is that to minimize sideffects of the profiling instrument
on performance, Andrew built a simple piece of hardware which plugged
in the boot rom socket of a WD ethernet card.  Using a hacked gcc, he
was able to recompile 386BSD to automatically generate function entry
and exit events to write through the socket to a RAM based data logger.

All for $100.

It crossed my mind that this could also be done quite easily via the
NVRAM/TOD socket on the 4c machines.  I have enough HW skill to build it.

Any comments/interest ?

Regards

Paul