Good point. I don't remember any 486 board that would take more than 64MB ( 8 x 4MB SIMMs).

- D

Some of those older machines are very well built and will probably run for a long time yet, but the biggest drawback is the amount of work done per electricity used ratio. They are slow but sure, but not very efficient. Of course, they still pump out some work.

Cheers

I was reading thru this and chuckling.
8 months ago I took down my dual 3600/2mb/800 Irwindale system as I considered it too expensive in elec to continue to run.
That was app $4500.00 back in 2005.
Gave it to a buddy in Montreal and he uses it as a game server.
Times change and electrical cost per WU done has to be factored in, at least where I am where elec is .145/KWh.
I can't imagine still running a P2 or 3 today.
For perspective my DX3600 rig pulled app 280w at load and did 7000+PPD
My 8 core clovertown machine pulled app 420w at load and did 23,000PPD
My 8 core Harpertowns pull app 320W at load and do between 35-38,000PPD
Haven't checked the draw on the Gainestown rig yet.

Would you spend an unknown amount of time on an issue that may happen 1 in a million (I don't have the statistics, but very few report)? If it happens to your [aging] client, the issue is bound to be specific to that client. Certainly the Work Unit detail will tell you if others get low credit relative to the others in the quorum and time factor of 7 hours CPU time. Obvious, it's of course the number of seeds a unit has completed. Usually the message log has a line near the end to note a special condition.

Probably.