It’s a distributed hash table that distributes hashes?
It’ll run down on a Pi4. Unless it’s totally I/O bound, I wonder if it could be turbo’d with some key native code libraries?
We are working on improvements like native Windows binaries
Probably then.
There is plotting (creating the chia plots) and farming (monitoring for wins). Remember this is a “proof of space” system.
Farming is very low resource. Once the space (the 101GB K32 plot file) is created, all it needs to do is sit there on a not-very-fast disk, be accessible via the internet for 
ticket win checks (which aren’t particularly expensive)… and well, that’s it. That’s all. You could easily farm on a Raspberry Pi 4 and a bunch of USB connected 2.5" drives with plot files on 'em.
Plotting is somewhat resource intensive, in terms of I/O and CPU to generate the Chia plot files, but nothing like the constant hardcore GPU and specialized ASIC stuff that “proof of work” systems require. You definitely would not want to plot on a Pi 4 though. I mean you can…
A rough benchmark of a SHA3-256 hash of a 60735349 character file, both in Free Pascal, using the same hash library:
AMD Phenom™ II X4 810 Processor, 2600 Mhz. Windows 64
vs
RPI3B+. Raspberry OS 32.
PC: Ticks (ms): 2922
RPi: Ticks (ms): 10466
Yes, not quite a third the speed. Golly those little Pis are so slow!
Hm. Now how to set up a test to compare that with Python…
eta: /makes a note not to allow anyone to post 61M text messages to ShadowCast.
Try on a Pi 4! They’re quite a bit faster. Here’s a few pics of farming rigs powered by a Pi 4:
Note that IMHO the power drawn by the drives (5W - 8W per 3.5" drive, I’d think) will far outstrip the power of the engine driving it, for any chia farm outside the smallest of small…
Go for density, always go for density…
Sure, I’m not even boosting the clock on that 3B+.
In what program did you make System Resourse Usage graph?
I got that graph from Reddit; I didn’t make it. Source is here. Sorry, I should have linked the original source.
That’s a lot of Chia.
Whooo boy you start to really want faster network connections. I’m glad some machines have finally started to ship with 2.5gb ports… that’s a very solid ~270-280 MB/sec copy in my experience. Quite the bump from 111-112 MB/sec of standard 1 gigabit ethernet. It’s not too expensive to add on, either.
Convert any boring old USB port (A or C) to a 2.5gb ethernet port for $30:
https://www.amazon.com/gp/product/B084L4JL9K
Get a basic 5-port switch capable of 2.5gb for $132:
These 12tb Seagate Ironwolf “NAS” drives have the following performance characteristics:
I’d assume that’s about par for the course on 3.5" 7200rpm spinny rust drives these days? But with the 18tb drives (WD, though I’m not sure the brand matters), I do a bit better:
write speeds may change as the drive fills I’ve observed as the drive fills, the write speeds decline. This above examples are with the drives more or less empty. Once it gets full, I remember writes dipping down from 200mb/sec to 130mb/sec. But reads shouldn’t change with available capacity, I don’t think?!
Such a blast from the past. I remember caring about how fast disk platters spun… 5400, 7200, 10k, even 15k… before the transition to SSD around 2012. Now I suddenly care about spinning rust again, because as a farmer you are gravitationally and inevitabally pulled to the densest possible storage!
(And then there’s striping… for that juice… every added drive in a stripe set is a 1x gain in speeds, so even a simple 2x stripe set would give you double the read/write speed! But if anything happens to those drives, bye bye data… see the raid calculator for more.)
Why this matters – side effects of proof-of-work:
Supposedly ethereum is working on a proof-of-stake approach but I’ve heard a lot of skepticism about whether they can or will get it to work.
I’ve taken most of my Chia talk to https://chiaforum.com/ – feel free to check it out if you take up this hobby. 
The Chia coin goes live tomorrow. Will be interesting to see the price per coin and all that stuff… this is where the proverbial rubber meets the road!
So, very confussed about plots vs farming. I can build the plots and then offload to another PC to do the farming? How long does a plot take to build? I have been reading so much I have confussed the crud out of myself lol.
Yes, you run the GUI to do the farming. Tell that GUI where the files are on your disks via the Plots stab. You can set up as many farming paths (or network drives) as you like. That’s all you need to do:
Chia is hard on SSDs:
I found out that Lazarus defaults to ARM6 code, for compatibility with RPi1s and Pi Zeros. Changing the target to Arm7A gives only a small boost. The Neon/VFP4 registers, instructions should be able to give some GPU-like speed boosts, but that’ll probably require some assembler libraries.
Netspace is almost 4 exabytes now!
Yeah it’s kinda blowing up. Coin price launched at around $600 and is currently $1,466.
This is my first time messing with anything crypto related and it’s… a hell of a drug. This tweet thread pretty much captures it all, well worth a click through if you want a funny absolutely 100% on the nose summary of all the crypto stuff:
I am TL4 on that forum (it’s not mine, but the owner is an agreeable, friendly person) and I got the new user experience when I signed up for it… a taste of my own medicine, dogfooding if you will 
. I actually quite enjoyed it, granted I’m extremely biased, and ended up
Anyway, win-win. Not sure where crypto is going, exactly, but proof of space is a far better way to approach this stuff than infinite GPUs grinding away endlessly and burning through electricity at astronomical rates.
Unless you’ve got one of the Before Times macs with swappable storage; you might want to be careful. I haven’t dug into the details of exactly why there’s so much I/O involved; but apparently doing enough of it to beat at least consumer grade SSDs to death on short notice is a thing.
