Open chinabala opened 3 years ago
the -w feature was created to address that issue.
Remember that you are (generally) targeting total throughput, not a small number of seconds somewhere in the log. I generally have embarrassingly slow plot times but have more than one going at a time to get peak throughput. This applied to chiapos and at least for myself and various others to madMAx as well. I have found that even with --rmulti2
I can't keep my system busy outside of phase 1 so I run a phase1 stagger via plotman and end up with three'ish madMAx processes going at once.
So, keep an eye on your daily throughput instead of the plot duration.
I've noticed that my single plot time's 1600s, but when continuous plotting will increases to 1800s-1900s per plot. (I understand to use -w instead, but..)
Wondering would it be the CPU problem or SSD bandwidth or something...
And, what if distribute copying to HDD through out whole plotting phases evenly. Would it help?