I just got into Chia a couple of weeks ago. I thought I had the process down but I think I made a big mistake. Ok, let me start by giving details.
I downloaded the latest version of the Chia GUI for Windows 1.7.1. It took about 5 days to sync to the blockchain. I bought (1) 16tb hard drives and I had (5) 6tb drives laying around. So ~ 46tb or so.
In the GUI I clicked on Add Plot and selected MadMax Plotter. I have a 32 thread processor so I selected 32 threads. I have an nvme for temp drive. For the past two weeks I have been plotting to all of these drives and finally finished today.
So I thought, now I decide whether I want to solo or join a pool… That’s how typical mining works. I decided, well, I will just try a pool first. So I joined a pool and entered that url into the GUI. It doesn’t seem to detect any plots.
Then I do a Google search and I am reading that you have to decide before you even make the plots whether you want to solo farm or join a pool. Is that correct? Please tell me that isn’t still the way this works. Hopefully I am wrong and just reading old posts in the search results.
In the GUI when you go to make a plot it just reads “Enter a pool (Optional)” so I thought well that can always be added later. If it truly matters on how you waste your time and hardware I would expect that part in the GUI to really state more obvious details.
I just joined this forum today and already saw an interesting post on FlexFarmer. This seems like the way I want go, but if all of you that have gone through this have any advice please let me know if you like FlexFarmer. Also, even if I plan to grow I can’t imagine going over 100tb. So would you chose a pool over solo? What if you want to switch pools? Are your plots “tied” to a specific pool?
And the biggest question of all… are all the plots I have created garbage unless I want to solo farm?
Did you make OG plots, that you farm solo. Most pools use NFT plots and when you created the plots you had to select which type of plot you wanted. You needed to have a contract key to make a NFT plot.
Yeah you make a good point, it is really not clear from the GUI how important this can be and I think it also should default to pool plots.
Basically, yes If you didn’t select the pool option when plotting, then you can’t join most pools.
Flexfarmer is perfectly fine (very lightweight) I’m not sure if that fixes your problem, but let’s ask @Chris22 who should know.
Other option would be foxypool who I think is still running and use the old style plots.
100TB is possible to go solo, but I would prefer a pool with that amount I think
Ohh wow, I see how it works now. This is a bad design in the Chia GUI. I mean, for a new person I saw “Plots” and clicked on it, added new plots and started plotting. If you click on “Pooling” you can add a Pool and then you get the option to add an NFT plot. So basically, you have to decide from day one if you want to solo or pool farm… and I didn’t.
Mine are all OG plots so they are wasted. What makes me the most sad is the life of the nvme that was wasted. Half of it’s life is gone.
Ok, back to square one. I hope I’m not in for any more surprises.
Can you use the same NFT Plots if you change pools?
You can’t farm plots that aren’t on our pool nft with FlexFarmer.
Given that there are compressed plots now though, you may be better off just replotting for them.
If you use this guide Gigahorse Ubuntu Simple Plotting Guide you can plot and farm (using FlexFarmer) without ever syncing the node then run the client in light wallet mode. However, those plots will be locked to Flexpool.
Thank you to everyone for replying. Ok based on all of this info…
I wished I had 128gb ram or 256gb ram to do Gigahorse. I only have 32gb ram. I think I am going to have to plot it in Chia GUI. But I am going to move the drives over to a mini desktop 35 watt machine when done plotting. I had Chia GUI installed on that also but I just uninstalled it and am going to give FlexFarmer a try. (Already set up the config file yay)
@ChiaJoe That’s awesome! I sure do hope the same happens for me. I’ve never even come close to 50% before. I don’t know what’s supposed to happen when it reaches 0. Anyone else have an nvme that keeps going and going and going?
Ok I’ve got 2 plots down and about… Ughh to go. haha
The TBW (Tera Bytes Written) value is used by the manufacturer to end the warranty on the NVMe drive. Otherwise, the NVMe drive should keep working with no signs of fatigue. For how long is hard to say. I doubt anyone knows, especially with so many different drive models.
I know of no studies for scores of different model NVMe drives. So there might be some NVMe drives that wear out shortly after hitting the TBW value.
Virtually no one in this forum has complained about wearing out their NVMe drives, and if ever there was a group that pounds on their NVMe drives, it is this one.
Also, unless the manufacturer pounded on their NVMe drives (pre-release), they might not know their own drive’s actual endurance. I doubt that they would hold off on releasing their new models for months (or over a year), which is how long it would take for them to conduct such tests of 24/7/365 writing to the drives.
Drives based on QLC NAND flash will wear out the fastest. That does not mean they will wear out fast. They will simply not take as much of a pounding as TLC NAND flash based drives (or MLC or SLC based drives).
One more reason for the TBW value assigned by the manufacturer:
Many people will run a tool that shows that their NVMe drive is at end of life, and so they go out and make a new purchase – not realizing that they could have kept using their supposedly worn out drive.
Thank you everyone. I always thought it meant after the TBW the drive would fail. So that is great news! Yessssss
As far as the big storage HDDs originally I thought I would just keep buying WD external USB drives but those things are costly and I am poor. I bought two of them but hopefully can find something cheaper. One cost me $200 on ebay and the other was $259.
I know you all probably have some monster systems. I found an 8 bay DAS USB enclosure on Amazon and I just got it yesterday. So far I love it. But I haven’t tested it yet obviously because of my plotting mishap. I have the five 6tb in there (I know, not ideal space, but I already had them here) and I would like to fill the remaining 3 spots with some 14tb-16tb drives.
I’m going to do a search on here and see if someone has some suggestions for cheap hard drives.
Definitely… I had them here at home not being used anymore. Eventually, fingers crossed, they will all be replaced with 14-16tb drives. I believe that is the max for this external usb enclosure.
You know what else I can’t wrap my head around? If I am able to parallel plot. This stuff is complicated (for me). Right now I created a que to just plot one at a time. But I am only doing one every 30 mins or so. Since I have multiple final plot drives that part is fine.
But I only have one nvme temp drive. So does that ruin it for me plotting in parallel?
Assuming you are creating K32 plots, they require approximately 250 GB of temp space.
So if you have a 1 TB NVMe drive, you can run 3 plotting jobs, concurrently (probably 4).
While plotting one plot, check Task Manager to see how much RAM you have free. Also check your CPU usage, to see if you have idle time.
You will probably have enough RAM and CPU cycles to accommodate running simultaneously plotting jobs.
The temp drive is a 2TB nvme. I have an AMD 3950X 16c/32t cpu and 32gb ram.
When you go to create a plot it asks how many threads you want to use. Since I have been doing single plots I just entered 32. But if you do parallel plots do you divide the threads?
So, if I do 4 plots do I still enter 32 or divide it by 4 = 8 threads?
I imagine once the plots are done things should get easier but this beginning phase is blowing my mind.
I have had 3 SSDs (not NVME) die on me so far, all in different systems. All 3 were from Samsung 256GB TLC, 512GB 850 EVO and 512GB 870 EVO. All were only used as system drives of a Windows OS, not for plotting or similar stuff. None of these drives even came close to the TBW.
In contrast, my plotting NVME SSDs have all lasted so far without a single failure (4xGIGABYTE GP-AG42TB with 3.6 PB TBW) and they are all currently at almost 9PB TBW.
Available Spare: 100%
Available Spare Threshold: 5%
Percentage Used: 100%
Data Units Read: 19.885.076.633 [10,1 PB]
Data Units Written: 16.980.110.872 [8,69 PB]
Host Read Commands: 34.071.742.298
Host Write Commands: 30.988.308.686
This ultimately shows how negligible these TBW figures are.
Flash memory is sold to keep data while powered off. With each TB written, those cells are wearing off (silicon degradation) and the time without the power is shortened. So, those TBW numbers really don’t mean that after that number will be reached the whole thing will collapse, but rather that if such drive will sit without power for whatever is time expected, the data will start expiring. On the other hand, in a computer that runs 24/7 such wear will be much less visible.
So, for plotting, the worst thing that will happen is that after some power off, the drive needs to be formatted, and will be happy to do the plotting job. For a computer with such drive as OS drive that is powered off from time to time, whatever was on that flash most likely will be corrupted.
