Queuing up work in the GUI (ghetto-style)

I’m still learning my way around how plotting actually works - I’ve started and stalled out on full disks probably 3x already and thrown all those 30% complete plots out each time.

I’m thinking the right number of simultaneous plots to at least keep things moving around for now is 6:

  • It’s an 8-core CPU
  • I have 2x Samsung 980 Pro 1TB drives - so I figure no more than 3x plots per drive.

If I’m right, then ultimately what I want is to queue up plots for days/weeks and just let it run and do it’s thing.

I know everyone (cool) is using the command line or a management front end - I just haven’t dove into those yet and I’m wondering if it makes sense to use the GUI to setup:

  • Plot 1 - SEQUENTIAL - Count:29 (max) - tmp:/nvme1 - dest:/sata1
  • Plot 2 - SEQUENTIAL - Count:29 (max) - tmp:/nvme1 - dest:/sata1
  • Plot 3 - SEQUENTIAL - Count:29 (max) - tmp:/nvme1 - dest:/sata1
  • Plot 4 - SEQUENTIAL - Count:29 (max) - tmp:/nvme2 - dest:/sata2
  • Plot 5 - SEQUENTIAL - Count:29 (max) - tmp:/nvme2 - dest:/sata2
  • Plot 6 - SEQUENTIAL - Count:29 (max) - tmp:/nvme2 - dest:/sata2

If I kick all of those off, I’ll have 6 simultaneous plots queued up and running for the next X days, fully utilizing the SSDs and slowly dumping files to the SATA drive.

Anything wrong with this if I’m willing to forgoe some performance to just getting this goddamn process moving?

I’m sure as I learn more I’ll want to come back in here and mess with staggered starts and really care about the plot phases, but at least for now I was thinking it’s an 80/20 solution without needing to get more in depth with more tools before I can “be productive”.

Am I nuts?

I really, really urge you to use the CLI for plotting. It’s very straightforward. One shell window per parallel plot, repeating over and over. Use a sleep command so they don’t all start on top of each other.

Or get fancy and use a plot manager script, but really, you can go far with a simple shell and sleep command.

Trying to unpack this a bit (figure out how it applies to my setup).

I believe my 2x 1TB drives will limit me (from a size perspective) to no more than 6 parallel plots occurring on the host machine at a time - otherwise the drives fill up.

The CPU is 8-core, so it probably doesn’t want to do more than 7-parallel plots (or 8 if I’m pushing it).

So lowest common denominator is 6 here - so let’s say 6 parallel plots.

Do you mean TERM-WIN-1 running 3 and plotting to /dev/nvme1 and TERM-WIN-2 running 3 more and plotting to /dev/nvme2?

Or were you thinking of many more than that and me getting fancy with overlapping phase start times so it’s potentially much more parallel plots getting kicked off 1-2 hours apart when the first batch moves into a different phase and isn’t pounding the same resource (i.e. CPU or Disk) in the same way?

I hope to eventually get there, I’m just trying to wade into this puppy a foot at a time.

1 Like

Yes, the simple way is one console / shell window, doing 999 plots. Stagger the start time by ~4 hours, and use the appropriate number of shell windows for the cores / drives you have. Each shell window is 1 set of parallel plots.

1 Like

i do both of these…

1 pc which is the main farmer is plotting like the op in the gui.
4 other pc’s (which are prebuilts/eth miners) modified with extra nvme drives etc running multiple cli/powershell windows

Got it - I’ll pull the command line that the GUI is setting up and see if I can fumble together something easy in a bash script for now.

1 Like