Hello. I have 2 plotters. #1 is Threadripper 3960X, 128GB RAM and 4 x Gen3 2TB NVMEs. #2 is Dual Xeon Platinum 8160, 128GB RAM and 4 x Gen3 2TB NVMEs.
#1 Does 20 plots in parallel in around 10 hours.
#2 Does 20 plots in parallel in around 12 hours.
Each plot is saved in its own HDD.
I do not set a delay or anything in the GUI. Will I get any kind of better plot times if I set some delay?
Also, would a SATA SSD set as secondary temp dir help improve times?
Short answer yes, but its probably better to run a cli plot manager to do that, you can then time via phases optimally as oppose to just guessing with GUI.
lots of good guides on youtube and here to get you going. i would assume you could up your 40 a day on both machines , by how much I’m not sure.
If I run 20 in parallel won’t there be an overlap at some point? If I set a stagger time of 4 hours every 4th plot that begins will be at the same phase.
Just staggering like that will not do what you think. The I/O bottleneck causes all the jobs to basically sync up over time. In a couple days to a week, all the jobs will be running on the same phase together. Or at least close to it. With a plot manager like swar’s or plotman, it keeps them staggered properly so they benefit from running at different times.
Yeah, this is true, with some caveats:
you can reduce the effect somewhat by having fast offload media (which helps overall plotting time, too, since less time writing = more time plottin’!)
many hour staggers take a LOOONG time to fall back in sync. I stagger by 4-6 hours sometimes.
you can use the Windows “select text from console” method to pause plotting, or suspend from task manager if you need to slow a plotter down for a bit so it doesn’t collide in writes!
The slower the offload media, the worse this effect will be.