Hi every-one. I’m actually running a “small” farm of 4000 nftplots.
I’m not in the whole “buy an old server + 256GB of RAM to replot as fast as possible”:
I don’t really see the point of it “ecologically”;
I don’t really see the point of it “economically” (Contrary to XCH starts, there is no point rushing to replot with low price & not so fast increase in netspace), where I am, it around 800$ to get an old Z440 + ECC RAM+ time to make the new set-up…;
Personnaly, I’m not in a situation where I can do a lot of test & run like in 2021;
(Also, I’m running some forks of interest, but except for 3-4 interesting projects, I don’t see it going anywhere).
So I’m definitely into “slow compressed reploting”, but not into paying anything new to do it.
I’m not in a rush.
What I have:
“low end” GPU(1*1060 6Gb);
“middle end” GPU (1*3060Ti) that I can “borrow” from my son PC;
1TB SSB with decent speed that made my entire farm;
Ryzen 3800X CPU, 32GB RAM;
My 2 “20 cents” questions:
What would you do if you were me?
Anyone with a comparable exotic or “not so high-end” experience could post under this message to share its experience with other “small farmers” : your hardware, software, & time to create X plots with C"Y" compression level, and your hardware for plot decompression.
I am in similar position. 3300 plots. If you are making $200 a month farming Chia if you replotted your existing setup you stand to make $270 or more and your ongoing farming electricity bill would be slightly cheaper too.
You will not lose out from your existing cash flow but you could be earning 30 - 40% more.
Your low end 1060 is actually high end for farming
I say you should plot a few flexfarmer dedicated gigahorse compressed plots and flexfarm them to your current wallet.
You can compressed plot with cpu
Unless you want to change pool to flexpool then you could plot to your own keys instead of using flex’s
I am plotting with an i7-3770 and a i5-2400
that get 90 minutes and two hour plots.
I couldn’t get a quadro 2000 to farm but a k2200 works. C6
32GB memory is insufficient. If u buy anything, it should be to increase memory up to 128GB… 64Gb minimum… if you plan of doing so called ‘compressed’ plots at > C6 level with a GPU. Below that you might get away with CPU ‘compressed’ plotting, but it is slower, if that’s ok.
The more memory u have, the faster u will plot with ur kids game GPU, or even yours, but is is necessary in any case.
Yeah, I understand But 128GB is 550 euros minimum (~590$) and I have to add another 300 to build a decent refurbished Z440 solution. @ actual XCH value/ efficiency gain of compressed plots, it’s like 48 months ROI…
I’m not really in the mood of going further and further…
Also what difference would 64GB make doesn’t it have to be 128GB min?
My actual farmer is actually also a mining rig under Win 10 pro.
It is stable but I have a ~3% time off due to reboots & everything.
I’m seriously thinking about :
buying a Z440 (I found one for less that 200€) as a farmer/JBOD.
cpu plot compressed C7 with my RIG (Ryzen 3900X 32Gb RAM, 1To SSD)
transfer plots to the Z440, farm using the GTX 1660 6Gb.
Do you think I need to go full Linux on the Z440, or is the WSL solution OK (it is planed that it will only do Chia)?
One of my problems is that I choosed the Windows ‘storage space’ route to create bunches of 4 disks, and will have to reorganize that thing deeply with Linux…
currently 70TiB all are gigahorse c8 k32 plots for Chia, my difficulty is 20 and my lookup times are about 15-20 seconds. this is my flexpool : i am using flexfarmer, but in my experience that doesnt make a difference to lookup times because even with a ping of a second it doesnt make a difference when the k2200 is making lookup times 15seconds