Question about generation times

[rshepard]

Should the time to develop each generation increase? It seems that after running for a while, the time between generations gets longer-- as much as 2 hrs on an 1133 MHz box--
Is this normal? Everything looks OK, (i.e. the process is running, CPU is maxed, got memory to spare, etc.), I'm just seeing long generation times on some of my boxes.

[michaelgarvie]

It does get longer, but 2hrs seems a bit excessive. Are you not running anything on top of it?

[rshepard]

Nope- it is the only thing running-- I had a second machine hit a 3 hr interval, but it seems to have worked its way back to ~15 minutes at the moment.

[rshepard]

Apparently the client can work itself into a "dead spot"-- I restarted both, and they are running normally.

[michaelgarvie]

The more time spent per generation, the better this generation is. This is because circuits are evaluated in stages, like job interview rounds. If a generation is taking long it means most circuits are getting to the final round which takes longer to evaluate.

[rshepard]

oh.......... oops :blush:

ok, next time I'll just let it run