View Single Post
Old 2005-09-08, 17:17   #6
wblipp's Avatar
May 2003
New Haven

22·32·5·13 Posts

It doesn't really make any difference for the vast majority of situations.

A low value causes more frequent connections to the server, but the traffic per connection is very small, so this doesn't really matter unless you are planning to put 1,000 computers onto the project.

A high value might cause you to work longer after another user has found a factor - but factors are rare enough that this isn't much of a problem. A high value might also cause you to run curves at a low level after another user has finished the level - but the transition points are set so that the effectiveness of lower and higher levels are the same, and the additional low curves will be accounted for by needing fewer high curves, so this doesn't really matter, either.

I've only found one situation that makes any difference, and that requires using the new ecmclient that can share among multiple projects. I have access to a few overnight computers, where people manually start the process at the end of their work day and manually end the process the next morning. I want to use these computers for OddPerfect Most Wanted numbers, which typically take 2 hours per curve. My concern is when they kill the program in the morning, all work on the current curve is lost - this could be up to two hours of work.

My solution was to use 8 hours and multiple projects, so that they work for 8-10 hours on the long curves, and then switch to smaller curves on a different project so that less work is lost.
wblipp is offline