Yeah, you're right - it's 6h, but I did say around, and I doubt that a 700MHz system is going to make the 60% mark in 7h let alone 6.

AFAIAA, HCMD2 targets 6hrs CPU time before returning results. If the job is at least 60% complete at the 6hrs CPU mark, the limit is extended to 12hrs CPU.

Given AVERAGE use and OS overhead, I see 7hrs wall-clock as a fair estimate - certainly more digestible the the full story with all the caveats and disclaimers.

Yes and No. For me the issue is not if you own already the equipment or not. It is what you are doing with it that counts. If it is a device you use daily for personal work or other taks, then fine, when unused switch to WCG. But if the idea is to run exclusively 24/7 WCG then the ratio WU/watt would be terrible. If you live in a region of the world were energy cost is minimal fine, but if you have high costs then simply do a small calculation and maybe you can afford a low cost more recent system with the difference in energy consumption.

I share your opinions about usefulness. While I do have a few powerful crunching systems, any lesser systems I have are either being used or they don'€™t even get attached. There is no way I could stomach using a computer that does 5% the work of an i7 but uses half the electric. My fence sitter is the IC2D laptop; it only crunches when in use, which varies in frequency. I do keep an eye on it and let it run for a few hours at least once a week.

If you only have one or two computers then I can see the attraction of adding another device, however I just would not be able to get away from the idea of replacing 3 devices with one that does more work and costs less to run.
The more you crunch, the more important efficiency becomes.

This is an example of one system returning a HCMD2 result after completing about ¼ the amount another system:
INFO: No state to restore. Start from the beginning.
Finishing early because max runtime has been exceeded.21780.406250
called boinc_finish

CMD2_ 1435-1IW0_ B.clustersOccur-1FLT_ W.clustersOccur_ 1_ 62094_ 65747_ 1-- 615 Valid 05/02/11 10:16:45 07/02/11 00:31:10 6.00 103.9 / 101.0
CMD2_ 1435-1IW0_ B.clustersOccur-1FLT_ W.clustersOccur_ 1_ 62094_ 65747_ 0-- 615 Valid 05/02/11 10:16:38 05/02/11 20:53:19 6.05 27.5 / 28.3

In this case most of the task will have to be resent; both were cut off at 6h, so both only completed part of the task. Got to be very messy for the server.

This is an example of one finishing and a wingman only getting about 20% through the task,

CMD2_ 1379-1GR0_ A.clustersOccur-2BAQ_ A.clustersOccur_ 6_ 55258_ 57131_ 1-- 614 Valid 28/01/11 08:55:15 07/02/11 05:28:50 8.68 71.8 / 105.4
CMD2_ 1379-1GR0_ A.clustersOccur-2BAQ_ A.clustersOccur_ 6_ 55258_ 57131_ 0-- 614 Valid 28/01/11 08:50:49 29/01/11 21:01:23 6.02 25.9 / 19.6

There seems to be a lot of lesser computers attached to the HCMD2 project. A good thing in that it allows lesser computers to be used at WCG, but I have my doubts about the merits of this for the projects servers. I can'€™t see this being the norm for any new project.

Another factor to consider is the cost of air conditioning . . .

Legacy hardware might make sense in February but not so much in August.