THE CHIA FARM

[T_Minus]

@Rand__ says win big or nothing.
IN reality it's win big or lose big... as your drive is used up doing this long enough.

There's really a cost in terms of electric ($$$$) and SSD\NVME $$$.

You really NEED to get coins, if not you don't just pay for electric but in drives too.
IMO of course.

 

[msg7086]

1 You don't NEED SSD to plot. HDD works. SSD or NVMe is the extra investment that people use to accelerate their plotting.

2 You can't really lose things (unless coin totally loses value). The rest of the year you can just keep your HDD running 24x7 and you'll get enough payback. I throw $5k in drives and my 5 coins already paid it in 3 weeks. Later those drives could be my new Free NAS.

 

[T_Minus]

1 You don't NEED SSD to plot. HDD works. SSD or NVMe is the extra investment that people use to accelerate their plotting.

2 You can't really lose things (unless coin totally loses value). The rest of the year you can just keep your HDD running 24x7 and you'll get enough payback. I throw $5k in drives and my 5 coins already paid it in 3 weeks. Later those drives could be my new Free NAS.

Sure, you don't need SSD\NVME\Fusion-IO.... Except no one is talking about using HDD they're talking about NVME or SSD for large # plots per-day. And of course you can lose everything, those coins WILL be worth less than they are now, this is not sustainable, there is not endless supply of enterprise drives. For everyone like you who got 3 coins or 5 coins quick there are people who got 0 so far... and unless you've sold and cashed in you haven't really paid off your gear either...

 

[msg7086]

Good point, just cashed 6k to really pay off my gear.

 

[Rand__]

Except no one is talking about using HDD they're talking about NVME or SSD for large # plots per-day.

HDDs only work if you can use lots of them at the same time, so that is perfectly fine for most people here, but not for AverageJoe mining on his Gaming Box with a few USB Drives.

And of course you can lose everything, those coins WILL be worth less than they are now, this is not sustainable, there is not endless supply of enterprise drives.

Why is there a need for an endless supply of drives? For storage or for mining?

And it is not that all the drives die after a week or so, somebody calculated a single P3600 2TB does ~10k plots - thats more than most users have in storage capacity.
The problem is those low quality consumer drives which are sold for a totally different purpose (home user, high burst numbers, not sustainable) for roughly the same money, they will take a serious hit in their life expectancy.

Re people who got 0 coins -
there seem to be some 280000 farmers at this point (chiaexplorer, Network decentralisation) , and of these 170000 have received coins (chia unique address count), thats 60% of all farmers - i would call that a pretty good chance to win.
Whether you make money or not depends on if you sell or not, thats correct, but thankfully you get 2 coins, so perfect opportunity to keep one, sell one.

 

[csp-guy]

1 You don't NEED SSD to plot. HDD works. SSD or NVMe is the extra investment that people use to accelerate their plotting.

2 You can't really lose things (unless coin totally loses value). The rest of the year you can just keep your HDD running 24x7 and you'll get enough payback. I throw $5k in drives and my 5 coins already paid it in 3 weeks. Later those drives could be my new Free NAS.

Yep, but how you can convert your chia coins to real money? Sorry for the stupid question, i have zero XP with cryptocurrency.

 

[Rand__]

gate.io trades XCH to USDT, which is relatively widespread. you can trade that for EUR or USD in other exchanges.
O/c always a fee and potentially ID check, but its possible

 

[bash]

A few of us were here years ago in the monero mining threads and I think the XMR I mined and held is worth a great deal more now.

 

[Rand__]

I think its sensible to payout at least a part so you are even with your investment (if possible) and then leave a part to gain (or loose), similar to stock.

 

[ArthurA]

1 You don't NEED SSD to plot. HDD works. SSD or NVMe is the extra investment that people use to accelerate their plotting.

2 You can't really lose things (unless coin totally loses value). The rest of the year you can just keep your HDD running 24x7 and you'll get enough payback. I throw $5k in drives and my 5 coins already paid it in 3 weeks. Later those drives could be my new Free NAS.

I'm not fully onboard with the frantic drive to speculate on huge plotting capacity, it'd be nice to get more plots sooner but eventually you reach capacity and that SSD hardware idles if it doesn't wear out or you don't continue expanding bulk storage. I'm team tortoise with 8x parallel plots (16 core) going on an Unraid 10x raid0 cache and farming from the remaining 14 drive array. If I score some coin I'll upgrade the rust from 4TB each and maybe finally get into Epyc but I don't need a dozen SSDs, if they even have any remaining life, after plotting slows to sustain. Though if it works out better than I expect I have 13x DL360 G6 LFF nodes preventing a pallet from blowing away that I could throw into the mix.

 

[msg7086]

Yep, but how you can convert your chia coins to real money? Sorry for the stupid question, i have zero XP with cryptocurrency.

I had zero xp either, but luckily I figured out. I transferred coins to MXC (who doesn't force you to upload your Chinese gov ID), then sold it into USDT, transferred to coinbase, then sold it into USD, then pull the money to PayPal and transfer to my checking account. YMMV.

I think its sensible to payout at least a part so you are even with your investment (if possible) and then leave a part to gain (or loose), similar to stock.

Right I pulled $6k worth of coin from my wallet and sold, so anything left there is my gain now.

 

[bash]

The only reason(s) to race now are:

1. Make up a greater % of the network as capacity has been growing at a torrid pace.
2. Pools will require a new plot type that has yet to be released. When pooling update happens no existing plots will be able to be added.

You are correct though people who see this is a marathon will still end up doing ok in the long run. People who see it as a sprint are taking a gamble on my points above. I literally had 99% of the hardware sitting in storage and my garage. I have been buying and flipping used enterprise gear for years as many in the forum do.

 

[gb00s]

I'm super confused about how Chia behaves.

First the size of the max size plots. The documentation says 254GB. When I'm plotting 10 plots in parallel on a 2.8TB SX350 after overprovisioning, the plotting fails with the 10th plot as the drive is full. Well ... 10x 254GB is for me ~2.6TB, not 2.8TB.

Then, plotting only 4 plots in parallel with like 40min staggering, is much quicker for each plot compared to plotting 10 plots in parallel with normal RAM size (3390MB) and 4 threads for each plot. So with a 48 threads and 128GB RAM setup I should be very fine with 40T for 10 parallel plots. The used 2x SX350 in RAID0 are not even on the limit of I/Os with 10 plots.

Also, don't know why, plotting the regular GUI-client is much slower than on the 1.1.5 experimental version.

I will start plotting from CLI. May this makes more sense.

Btw, 631 plots here and not a single coin ... LoL ... Plotting with 8 harvester machines and store them on filer. Is this a mistake?

 

[Rand__]

First the size of the max size plots. The documentation says 254GB. When I'm plotting 10 plots in parallel on a 2.8TB SX350 after overprovisioning, the plotting fails with the 10th plot as the drive is full. Well ... 10x 254GB is for me ~2.6TB, not 2.8TB.

Basically you have to add the generated file's size to it as the tmp file is assembled on the temp dir and then moved to target dir.

Then, plotting only 4 plots in parallel with like 40min staggering, is much quicker for each plot compared to plotting 10 plots in parallel with normal RAM size (3390MB) and 4 threads for each plot. So with a 48 threads and 128GB RAM setup I should be very fine with 40T for 10 parallel plots. The used 2x SX350 in RAID0 are not even on the limit of I/Os with 10 plots.

It behaves weird, I give you that, not consuming all resources it could get.. Probably suboptimal implementation...

Also, don't know why, plotting the regular GUI-client is much slower than on the 1.1.5 experimental version.
I will start plotting from CLI. May this makes more sense.

Cli and Linux is the recommended platform... but I also use the slow Windows gui... which often misbehaves, does not not want to start new Plots and so on. Should migrate too

Btw, 631 plots here and not a single coin ... LoL ... Plotting with 8 harvester machines and store them on filer. Is this a mistake?

If you have a lot of plots that are being created per day this might not change for a while, since chance to win is basically #plots x time divided by total network size, so while you increase chances every day probability (luck) just has not favoured you

 

[gb00s]

Basically you have to add the generated file's size to it as the tmp file is assembled on the temp dir and then moved to target dir.

According the the docs 254GB is the max tmp file size one plot should consume. The end-plot is 101GB.

... Probably suboptimal implementation...

You mean ... ? My setup is suboptimal or .... ?

Cli and Linux is the recommended platform... but I also use the slow Windows gui...

Yeah, Windows I read at the beginning is around 10-15% slower. Not sure this is still valid as of today.

If you have a lot of plots that are being created per day this might not change for a while, since chance to win is basically #plots x time divided by total network size, so while you increase chances every day probability (luck) just has not favoured you

A friend of mine here started a week ago and had yesterday 171 plots and already owns 2 coins ...

Also, I thought the much longer plotting time for a higher count of plots may come due to the file transfer of the plots. So I switched from directly saving them in the final filer to a 1.2TB SX350 and then renaming them and copying them to the final filer. Yes, it improved like 20min, but .....

 

[Rand__]

According the the docs 254GB is the max tmp file size one plot should consume. The end-plot is 101GB.

Well i only can tell you what I observed... 256 did not work out for me.

You mean ... ? My setup is suboptimal or .... ?

no, Chia implementation

A friend of mine here started a week ago and had yesterday 171 plots and already owns 2 coins ...

Luck/probabilities...

Also, I thought the much longer plotting time for a higher count of plots may come due to the file transfer of the plots. So I switched from directly saving them in the final filer to a 1.2TB SX350 and then renaming them and copying them to the final filer. Yes, it improved like 20min, but .....

I resigned to just plot without much optimization for now since i have not seen actual hard facts that i could use to optimize...
too lazy to actually analyze logs and monitoring values

 

[msg7086]

GB or GiB? Maje sure you don't mix units. If you use decimal based unit, it's 269 or something I think.

You can stagger the plotting to avoid hitting the same peak point at the same time.

 

[gb00s]

GB or GiB? Maje sure you don't mix units. If you use decimal based unit, it's 269 or something I think.

... quoting official CHIA-documentation ...

A k32 will take up 101.3 GiB of space once completed but will need a total of 239 GiB of temporary space as it is being created. A single k32 plotting process never needs more than 239 GiB of space. One needs to be careful here as 239 gibibytes uses 1024 as its divisor where GB or gigabytes uses 1000 as the divisor. That means you will need 256.6 GB of temporary space and the final plot file will take 108.8 GB.

 

[gb00s]

Just adding current situation. Reduced to 8 plots and still afraid I'm running out of space. Is this maybe a filesystem issue? I'm using XFS. XFS uses 4k blocksizes which fits perfectly to the SX350's ....

 

[Rand__]

I still think you need to add the tmp file (or part of it). Just check how much space a single plot currently uses in sum ...

And re chances, with current 6EiB chances to win are ~ 1:126.000 per plot ...

Edit ... so 1:195 for your ~650 plots
Thats not correct, as that would imply all your plots pass the Pre Filter, which they dont, so rather 1-4/126.000