Bummer.

For good or ill, that machine is in production, and I don't get much
opportunity to mess with it.  It's "backup" twin is waiting for 2.1 to
come out so I can see if that's better.

The production machine is still running the stock CPU, and if I
turn off powersave, it does a number of Bad Things, just like
before.  _That_ I've verified.

If I run the same SW on the machine with the 1.0 Ghz CPU, with powersave
disabled, it's similarly stable (OK, but not great).  Either CPU with
powersave off is not stable, but the faster CPU won't even boot
reliably, where the stock CPU will at least require a bit of load
to freeze up.

The production machine still
freezes up on me every few months to remind me that all is
not entirely right.  It does so when crunching web statistics at 2 AM.
(heavy disk and memory/swap traffic seems to trigger it)  My suspicion is
that it drops an interrupt in the ATA driver once in a while, but
that's just a guess.

-dgl-

>> Whoa!
>>
>> This problem sounds suspiciously like the one I was seeing some time
>> ago with CPU upgrades on my G4 machines.  See thread:
>>
>> http://mail-index.netbsd.org/port-macppc/2005/03/19/0001.html
>>
>> Is this the fix for my problem, too?
>
>I doubt it. Your problem was something else, and was something you saw on older and newer CPUs. This problem mainfests itself differently.
>
>> Can I patch it into 1.6.x?
>
>1.6 has a completely different toolchain, so I can't see it being necessary.
>
>Also, turning off powersave might've made things better for you on your machines, but on, for instance, my 2 GHz 7447a, running the CPU at 100% means about 40 watts more power consumption and the heat that goes along with that. So I'm glad at least for that reason that the problem is fixed elsewhere.
>
>I remember reading about your problem, but I couldn't offer any help because I couldn't replicate the problem myself. Is it something you still see?
>
>John