So what is the —check-plots option for? And how do you find out which plots are bad when it is done? I’ve been running it and it has found bad plots but unless you are willing to sit in front of the screen for hours or days how do you find all the bad plots? Where can I find the log file?
Or better yet it would be great If the client had an option to auto delete bad plot files. As you run plot check.
When you make plots there is a check and if it doesn’t pass it’s deleted
bladebit_cuda -f -c -n 1 --compress 7 cudaplot --check 100 --check-threshold 0.8 g:\plots\
This is the NoSSD thread 
Did I post info in the wrong thread again, another pint added to my owe list 
1 Like
You do not need to check new plots for errors. Our software is good enough to not produce damaged plots to begin with. A --check-plots switch is made to verify copied plots or bought plots and to identify a damaged disk.
I am sure your software is very good not meaning to throw any shade. I had a farmer that was running unstable and would lock up while moving plots so I am very sure that that is the reason I have some plots that are bad. I have since then fixed my farmer and it has been running along and farming very stably for about 4 days. But I decided to check my plots (about 2000) and I see that while watching for the first 20 minutes as it was checking the plots it flagged at least 2 as bad. I don’t really want to waste space on bad plots and would like to not have to replot my attached drives. This is why I asked if there is some way to weed out my bad plots other than watching the farming client as it checks each plot? Not trying to complain about your software. I hope this clears up what I am saying. As always thank you for taking time to reply.
I’ve never used NoSSD, but if it runs in the CLI can you pipe it’s output to a text file?
How to use a computer basic steps (for babies)
./startclientexample.sh >> ~/logfile.txt
2 Likes
Thanks, I want output on the screen session and to a log file to grep for errors and such.
Pipe your command’s output to the “tee” command.
Run:
man tee
for help.
I am assuming you are running Linux, based on @rallbright’s reply to you, where he used the “/” and not the “\” character.
Windows does have a “tee” command, via powershell.
There are likely 3rd party downloads available, too, for Windows.
1 Like
This is the Man TEE
2 Likes
Quiet here. No acknowledgement from NOSSD Developers on Drplotter………I think it will be interesting to see how farmers realign after this.
I wish his Video card needs would be trimmed down…
NoSSD is just so much easier, especially at scale. Yes, my plots can’t be used elsewhere but I’m ok with that.
4 Likes
NoSSD is 10x faster at plotting on top of being way easier to set up and run. DrPlotter is only going to appeal to a small target market (4090 gamers with 200TiB of attached storage).
7 Likes
How is this calculated? Number of cuda core? Thanks.
All values are obtained from the benchmark NoSSD and collected in a table
1 Like
I don’t think that I would agree with that. I cannot imagine a whale that runs any compressed plots and is using 1060 GPUs. I would assume that all those farms are somewhere around 3090 or 4090 already (especially those that run GH thanks to recomputer server running multiple cards per box). With the best previous compression (NoSSD or GH), they are running such farms at 2x. I think that it would be a no-brainer for those big farms to start switching to DrPlotter to run at 4x.
2 Likes