For Ubuntu, they’re probably around the 525 drivers, as that’s the newest in the repository.
@madMAx43v3r Doing a clean installation of 531.41 Studio drivers seems to have fixed the issue, I was using 531.61. During plot checks GPU usage goes to around 20%
1 Like
Not all physical 16x PCI-E slots are ELECTRICAL 16x.
Many are 8x or 4x or “depends on what else is using some of the lanes” on many motherboards.
It was going well until this error.
2023-05-29T11:15:09.352 chia.plotting.check_plots : ERROR <class 'RuntimeError'>: badbit or failbit after reading size 8 at position 21056465890 error in getting challenge qualities for plot \\CHIAWORKSTATION\Plots\Drive Bay 1\Chia GH NFT Plots\plot-k32-c8-2023-05-28-13-39-b14dea764f16636d86093b65e06676b42dfca150f5f3c08a3f928de04da36215.plot
2023-05-29T11:15:09.353 chia.plotting.check_plots : ERROR Proofs 5 / 30, 0.1667
Then it just exited back to cmd line.
Edit: That plot must be damaged, comes up bad when checking on the plotter as well.
Yeah that’s usually an incomplete copy.
1 Like
Thanks, something else I need to get my head around, it was bad enough when plots got produced in 15 minutes, but now its 2.5 minutes, so much faster than I can shift them.
Any recommendations for an automated plot mover for Windows to move to multiple drives?
Edit: Read about Plot Sink, but can’t seem to find a Windows version.
GPU plotter supports multiple -d folders.
Plot Sink for windows can be found here: chia-gigahorse/plot-sink/windows at master · madMAx43v3r/chia-gigahorse · GitHub
1 Like
And this is how I run it and works great, sample text for the folks.
.
In one CMD window run below
chia_plot_sink – H:\mmxk32c\ I:\mmxk32c\ J:\mmxk32c\ K:\mmxk32c\
In another CMD windows run below
cuda_plot_k32 -n -12 -x 11337 -M 128 -S 4 -C 7 -t f:\nvme\ -d @localhost -p pppppp -f ffffff
1 Like
Thanks, but having too many issues at the moment 
As well as the previously posted issues, which seem to have been resolved I still have the following.
If I copy across my 10G network, for some reason it affects TV recording on the server (HD Home Run / Mediaportal), so I use a USB dock, which worked faultlessly last time I plotted, this time it keeps dropping out. Not sure if that’s the dock or drivers, mind you I plotted on Linux last time, so probably windows, although I still used the dock.
For some reason the blockchain stopped updating several hours ago as well, coupled with the problem of moving the plots I’ve gone back to standard Chia for now, just need to replot 37 k32’s to refill the space I cleared.
I’ll regroup, make a plan for moving plots and then try again another weekend, back to work tomorrow.
Hi! What you use for -t ? -2 is the 3 x 1tb nvme in raid 0 right?
Yes!!! Exactly like that!
Hello guys,
I have a small farme about 32TB, and I’m planning to add double space and go up to 64TB with compressed plots.
Actually I’m running farmer on a RPI4 with flexfarmer, uncompressed plots.
Which level do you think is the max to run on a RPI4 4GB with compressed plots? C2-c3-c4?
And with a Mini-pc with Intel Alder Lake N95 + 8GB Ram? Maybe C5-C6?
Thanks!
1 Like
If it’s who I’m thinking of, they’re using a stripe of eight (8) 980 Pros… not just one.
1 Like
Plotting K32/K33 plots with -r 2 is slower than running two instances of Gigahorse using NUMA and plotting two plots at a time AND even slower than plotting one plot with one GPU. Why is that? Maybe “streams” argument should be raised, e.g. to 8 from the default 4?
Of course.
Because half your plotting traffic has to traverse the interconnect.
It’s already streams-per-GPU. If you notice a speed improvement by manually raising or lowering it please share your findings with us!
1 Like
Anyway.
I’m using flexfarmer on compressed plot, in a WSL virtual machine since 2 months.
It justs works flawlessssly,.
My only concern is about reporting “space”.
My farming rewards are as expected.
Anyone with a good PS script (different of C- GPT proposals that do not do the tricks 
) to
erase a minimal amount of space into a certain list of folders oldest stuff
, versus a certain amount of disk minimal availability to reach , continuously,
will get my reward.
It would avoid me to erase old plots manually continuously, it would be enough to get you 0.25 XCH.
YOU:
wtf
1 Like
Such undocumented configs requires tests.
*are you surprised?"
What do these different errors mean? Do I necessarily need to replot these plots? Because running them through plot_check gives me around 85% of passed challenges. Maybe these plots won’t be getting errors all the time and I can just ignore these errors?
ERROR worker: Failed to get qualities, the plot is likely corrupt ch=0c0cdc708f7c error=ANSDecodeDeltas() failed with: Src size is incorrect (t=C3, pos=4200000000)
ERROR worker: Failed to get qualities, the plot is likely corrupt ch=0c0cdc708f7c error=Invalid size for deltas: 20274 > 845 (t=6, pos=2261586100)
ERROR worker: Failed to get qualities, the plot is likely corrupt ch=0c0cdc708f7c error=Corrupted plot data (t=3, pos=263649366)
ERROR plots: Failed to initialize plot; skipping it error=create new mmx prover backend: badbit or failbit after reading size 4 at position 79044862040
The first three can happen sometimes and are in every plot. The last one is def a bad / interrupted copy or failed HDD.
2 Likes
You back from holidays mate? Github looking a bit quiet!