Raj Rijhwani wrote:
[...]

a significant drop-out/restart rate over a year, and for each one that's a LOT of wasted resource which would not have gone to waste with a smaller work unit.  If nothing else it's a very wasteful design decision.

On CPDN, the volunteers are clearly warned about how much and how long a computation is required for a WU.

This poor machine here sufferring my hackings will maintain some applications unchanged for over a year despite 'upgrades' elsewhere within the system. CPDN has a very good chance to complete intact.

As for their very long "WU"s... The present coarse grained solution is the quickest way for them to get some results. A finer grained model would add considerably to their system complexity and implementation difficulty.


The present CPDN seems to work well enough and is quite a good challenge for /real/ crunchers to chew on! A real good test for reliability and fortitude!!


How long is your commitment?

(;-))
Martin