Madmax -w option

Are you perhaps plotting using the GUI? That does it one at a time, so I assume you don’t see a “-n” argument in that case. I haven’t kept track of the GUI side so I don’t know if a fix is coming in 1.3

I do not use the GUI.
I do see the -n option, via taskmgr.

I see the entire command line with all of the options that I applied, and see no additional options added that I did not specify. What I specified shows up, and what I did not specify is not there.

Strange. I don’t get the CTRL-C to work, it just keeps whatever I had before.
But by making the Command Line column wider I can see it passing the wrong value

Command line that I run (but X-ing out some values for obvious reasons):
chia plotters madmax -n 2 -k 34 -d W:\Chia\ -t W:\Chia\temp\ -2 W:\Chia\temp\ -r 16 -u 512 -v 256 -K 2 -c XXX -f XXX

Visible in Task Manager, Command Line column:
chia_plot_k34.exe -f XXX -t W:\Chia\temp\ -2 W:\Chia\temp\ -d W:\Chia\ -c XXX -n 2 -r 16 -u 512 -v 256 -w 0 -K 2 -k 34

The problem being the “-w 0”. It also changes the order of arguments, but that is not a problem.
Are you sure you could do CTRL-C in task manager?

Yes.
First click on the row containing the data in question.
Then CTRL-c to copy it to the clipboard.
Then paste it into notepad.

By the way, I did not catch that it added a “-w 0” to the command line.

That looks to be the culprit. But I do not want to interrupt my processing to find out.
If anyone has run chia_plot_k34.exe directly, and madmax did not wait for the copying to complete before starting the next plot creation process, please let me know.

If someone confirms that it will not wait, then I will interrupt my plotting to run it directly via chia_plot_k34.exe

Thank you.

The MadMax version included with Chia is annoying and always waits. This is why I use latest MadMax GIT and compile it which is easy since I’m on Linux.

2 Likes

Discovery and solution:

Discovery:
When running madmax via:
chia plotters madmax -k 34 etc…

The above kicks off:
chia_plot_k34 (with a bunch of options). One of the options that it (inadvertently?) includes is “-w 0”.
As far as I can tell, that is causing madmax to wait for the competed file’s transfer, before starting its processing on the creation of the next plot.

Solution:
I ran:
chia_plot_k34 -k 34 (with a bunch of options – being sure to not use the “-w” option).
Viola! When madmax completed the plot, it immediately started its creation of the next plot, while simultaneously copying the completed plot to its final destination.

I see this as a bug with chia.exe, for passing along a “-w 0” argument that was not part of the user’s command line.

If chia.exe would do everything that it was doing, except for adding in the “-w 0”, then:
chia plotters madmax -n 34 etc, would work properly (in terms of file transfers).

So the work-around is no big deal, now that I know what to do and have the lengthy list of options/arguments and their respective values all set.

This will save a significant amount of time, by avoiding delays that were due to large file transfers.

1 Like

Maybe you could write about it on the github Issues page. Chia monitors that, and most likely they will fix it. Maybe some other people will give your post thumbs up, drawing Chia’s attention to it.

By the way, how long does it take you to produce one k34 plot? Is that ~4x of k32?

Some other user of this forum posted a link from github about this issue (although I do not recall if they identified the erroneous “-w 0” that gets added to the command line). I do not remember the name of the user. Also, I never used github.

With Chia, it was taking 16 hours.
With madmax, it is taking 6 hours.

However, I am creating two at a time, as I have two NVMe drives that I use for temp processing.

Since madmax uses 100% of my CPU time (for only one plot’s processing) for a good portion of of each phase (seems like phase 2 is the least CPU hungry), it means that when I run two at a time, they are competing for starving CPU cycles.

My point is that running 2 at a time take 6 hours each.
But if I were to run only 1 at at time, it might cut an hour or more off of the time?

The file sizes are sort of a multiple, but the larger files (K33 and K34) seem to have extra “padding” (for lack of a better noun, because I imagine the additional data serves a critical purpose).

K32 = approximately 108,xxx,xxx,xxx bytes.
K33 = approximately 224,xxx,xxx,xxx bytes.
K34 = approximately 461,xxx,xxx,xxx bytes.

I had originally plotted most of my plots as K32.
One day, those will be retired by Chia. So I figure, why wait for the announcement and not know how long of lead time we will have to re-plot.

So I am jamming as many K34 plots as my drives will hold, and filling in the remaining space with K33 or K32 plots. And it is complicated, due to different drive capacities. The Chia calculators give plotting results/recommendations based on using as much space as possible on the drive. That results in having too many K32 plots. Yes, you do end up utilizing 98% or 99% of your drive space, but it is loaded with K32 plots – and I am trying to get away from K32 plots.

1 Like

Wow it passes “-w 0” when the “-w” argument doesn’t require any number… how stupid. I hope they fix this.