CRAZY Low Spec Chia Plot and Farm Guide for NOOBS: 8GB RAM = 10min plots

today we're covering setups for small farmers and we're going to be looking at 8 gigabyte 16 gigabyte what kind of plotting times you might expect on systems that go all the way down to potatoes and if you are a potato farmer this is for you and today we're going to be testing all the way down to an R5 3600 which is not at all an expensive CPU we're also going to be looking at a variety of incredibly cheap things even stuff like quadros that are very old some 1070s we've got some additional 307s and 30 60s that we'll be looking at and we're going to be even checking out the performance of our favorite which is the P4 yes the P4 is making a comeback here and I'm going to show you some stuff today that's pretty exciting so we've got two different test systems set up here and I did this so that we would be able to get wattage statistics between the two and so if you see here you can see kind of the idle 7075 Watts that this 550x system with 16 gigs of RAM and a 3070 is running and over here you can see the R5 3600 with another 16 gigs of RAM exact same type of ram in both of these systems we're going to look at a multitude of different systems and we're also going to test out multiple graphics cards some of them surprisingly cheap and some of them surprisingly low-end also and the findings as far as what kind of wattage you can save while you're replotting are substantial first off I want to toss a big shout out to madmax who made all this possible with amazing engineering frankly and the gigahorse 2 platform which allows you to get very good compression levels all the way almost up to double the Savings in space so let's get started I'm going to take you through the steps for Linux today now if you are interested in checking out any of the chapters below you can just use those to replay or jump to any topic that you're interested in as well this guide goes in hand with the windows guide if you're new to you should definitely check out both of those because things like setting up your windows wallet and setting up a desktop gooey so that you can get your plot contract will not be covered in this I'm just going to assume that you've already watched those you can find that timestamp down in the description below as well as some of the items that we're going to be covering here and there's one Piece One Piece of Special Sauce here and this is what's going to make everything just fly and that is the fison nvmes these little dudes are capable of amazing performance I think you're going to be really pleased especially if you've got anything just anything that's men spec your plot times are going to be really good let that sink in and let's get started so I'm going to cover the hardware that you need after that we're going to cover installing a GPU then we're going to look at installing Ubuntu dirve in madmax's gigahorse package next we're going to go over a script that will set things up for you that is on my website that you'll be able to copy and paste line by line as well the website has descriptions of what each line does and we're going to look at installing and running the complete system that will include a farmer and also a plotter so after that we're going to get to benchmarks of speed now this is where I think the most interesting things are so you definitely have to watch until this part because there is so much good stuff in there after that we're going to be checking the pool that we're going to be joining we're going to be joining foxy pool again here today and make sure that they are registering the partials that's how you know that you have a functioning end to-end system and then we're going to go over some conclusions and some next steps for you so let's dive in so the hardware that you need to consider for this is pretty Basic Hardware these are low-end systems even a $100 system is very capable here because literally the only thing you need to have is about 16 gigabytes of RAM your CPU can be pretty much anything out there and you do need to have an nvme slot preferably available on it or if not available accessible with enough pcie slots on your system that might be one of the biggest constraints that you have because you also need a full PCI e6x Lane for the graphics card that you're going to be putting in there but we're going to be looking at a couple of systems today a PCI E4 nvme based system and a pcie 3 based nvme system and we're going to see some times that are surprisingly comparable on both of them and the next thing that you need as far as Hardware is whatever you're going to be plotting to so that can be in vme like the fison drives that we're going to be demonstrating here today you can also use consumer things but if you do use consumer things like ssds or nvmes or even some high-speed spindles make sure that you put them in a rate array so that they can gain a little bit of extra speed because they're going to kind of blow through their C and then tank on performance and you'll see that your times plotting on them are going to be many many many many times higher next we're going to look at just some basic USB drives that I have laying around for demonstration we're going to be putting plots on those and the final thing that you need is to make sure that you have a good cooling solution for your computer taking a look at the fan that I'm using it is important to keep good air flow on nbmes especially when they're running sustained workloads for potentially days and days on ends week on weeks and weeks on end they can get really hot and if they get really hot that can degrade their performance and impact their lifespan let's go ahead and install our mvme SSD and graphics card into this system this is the R5 3600 it's only got 16 GB of RAM only has two channels filled right now and is running on a b450 motherboard and going even lower spec we've got the quadro p 2000 this is a 5 gigabyte card and what this means for you if you're a Chia farmer is that if you have something like a 1060 that is a 6 gigabyte version you're probably in luck and if you check locally you can probably swoop those up for a really good price check the links below for some affiliate links to those as well and this plotting performance let me know what you think its times are going to be like in the comments below right now I want to wow you just right off the bat so I think this is a good one for us to start with going to actually recommend that you watch the first about 25 to 30 minutes of the Chia video that I just released here you're probably going to want to have the graphical interface with chia's wallet and so that's why most people most Farmers even if you're plotting in Linux even if you're farming in Linux that's going to be a dedicated separate machine and then you can use your Windows machine to terminal in and when I mention terminaling in we're going to be using the command line and I recommend you get the windows terminal which is going to make that just a little bit easier for you you can install that from the Microsoft store in Windows 11 and I believe in Windows 10 as well next you need to join foxy pool and the steps for that are outlined in the prior video as well link to that in the description below and you can also find it right here and while I realize that Linux is definitely not for everybody this is a great way to dabble we're going to download a piece of software called Vinoy go ahead grab the Windows executable and download this to your desktop from Source Forge and once it's finished downloading you're going to need a USB key that you can write over it will format and lose all the data on that so make sure that it's empty to extract it here you can go into it and ventoy to disk is what you need to click and it's going to look for a device that may be plugged in I'm going to go ahead and plug in our USB key and you can see that it pulls up the 32 gab dis that I've got here we're going to click on install It'll ask us to validate that we are about to destroy all the data that's existing on this device downloading the iso Ubuntu into here at this point so we're going to go ahead and grab the server image and after this is done downloading it's literally as simple as pulling this USB out after you unmount it and then installing it and booting from it inside computer that you're going to be using for the [Music] plotting so once vinto boots up go ahead and select ubun to and normal mode and we're going to select the top option which is to install Ubuntu Server you can use the Tab Key to select Ubuntu Server minimized and click tab tab done it's going to automatically find an IP address write down this IP address ours here is211 can skip through the next couple options where it's talking about updating next go down to where it says setup dis as lvm group since it's going to just be a single dis and we're going to partition the entire thing and continue it's going to ask for a name I'm going to call ours 3600 when it gives you the option to reboot now at the bottom go ahead and tab tab down to it and go ahead and run that it's going to ask you to unmount so go ahead and pop out your USB vent toy key now when it brings up the prompt here you should be okay to go ahead and enter your username and password which we're going to do here and then we're going to type a command to validate that we have the IP address that we think we should have and that is IP space a okay we can actually go back to our Windows machine that we're using to terminal into this machine and make all the edits and set up all the plotting at this point we've done everything else that we need to do physically on this machine now if you haven't taken a chance yet to go ahead and in Windows download get the blockchain synchronized get the faucet create your contract and join the pool then you're going to want to do that now so go to the video that I just released and the link to that timestamped is right below here and that'll get you up and running as fast as possible probably a couple of hours to get the blockchain fully synchronized and everything squared away so I'll see you guys back here in just a little bit if you haven't done that yet okay so now that you've got your blockchain synchronized and up and running we're going to go back down to pooling here and if we hover over gray Jackal you're going to see that we have our xch contract address we're going to copy this and bring it over here to a notepad and I'm going to walk you through every one of these pieces of information in the notepad and what it's doing and if you look here this is a plot command so this is the CLI execution of how a plot gets kicked off where it lives how fast it goes What compression level it's at based upon the nvme is going to be also something that's in here as well as the number of them that you're planning on creating so the first thing that we're going to see here is the C directive and the C directive is our plot contract that we just grabbed there from the hover over and so you go ahead and paste that in the next thing that we're going to do is we're going to grab our farmer public key and you can find that just by logging out really quick going to your wallets three little dots go to details and your farm Republic Key is this right here now Cuda plot k32 this is a binary that we're going to download and we're going to be using for creating the plot this is the gigahorse 2.5 uh the new one the gigahorse new one this is what you use to create the higher sea level compression plots The Hyphen in here is the number we're going to create one is for just one negative one would be to create an infinite amount of them this is most likely once you get up and going what you want to use however for our testing purposes we're going to keep it at one the capital c that we have here is 17 we're going to go with a compression level of 17 on this P2000 which is a 5 GB card here I did test 18 and 19 on it as well and it is capable 17 seems to be like about as far as you would want to go if you were planning on farming it after you had plotted it though so we've got our C directive we have our F directive next we have the hyphen T this is the temp path so you'll note that in Linux you don't have a drive letter here you just start with the for Slash and a mount so this is basically a path on the system then we have the hyphen 3 this is also the same directory both of these two directories are living on one of these small 320 gigabyte in vmes and then we have our D directive and this is our plots folder and our plots folder is possibly one of the USB drives or it's possibly the SSD and then we'll have a secondary mover script that's what we're going to end up setting up here today so the next command that we look at here is pseudo apt update Y and that's going to pull down the latest repositories and tell you what updates you can apply then we're going to look at upgrade why and that's going to actually apply those upgrades next we have pseudo app install so this is going to install software and each one of these space separated items is a different piece of software so you can see we've got Nvidia driver 535 that's the driver that we need to install we have glances this will give us a good look at the nvme activity we have EnV toop NV top is going to allow us to see what the uh Nvidia card is doing as far as its performance its Ram utilization and its CPU utilization htop is something that's nice to have around and we need to install git The Hyphen y at the end here is signifying that we already pre-answer yes so you can see that here here and here so we're just saying yes to all of these commands so it will automatically go next we're going to get clone the madmax GitHub repository and this will pull down the software for creating plotting and testing plotting for us then we're going to reboot the device this is critical because you won't see the drivers and the Nvidia graphics card until you do that so you do have to reboot after that these are pretty much just onetime operations everything underneath here could be viewed as something you could run over and over if you wanted to do a bunch of testing on different cards if you wanted to do a bunch of testing on different MVM to find the best performing one that you would have not a bad idea to start there so we're going to turn the swap off so pseudo swap off a that will disengage the swap functions so the ram will be the ram and not the you know disc or hard dis or SSD or whatever you've installed in in our case an SSD next we're going to pseudo WIP fs and then we're going to force all of the dev nvme zero and one and we only have one nvme and it does live here I'm going to show you how to find the path in case you have multiple nvmes be careful if you have two to not install on top of your first nvme that may have your operating system installed on it next we're going to do pseudo mkfs butter fs and we're going to do it on that same device why butter FS butter Fs in recent kernel releases has come quite far as far as performance for nvme and Flash and with the directives that we're going to be mounting it with it's actually a really good option and so this is a flash aware file system by Design and I like that versus some of the older ones that are not next we're going to create directories pseudo make dur hyphen V mount plot temp this is the temp directory that we had mentioned here and here then we're going to also create for SL plots and that's going to be here now if you just create for SL plots and you haven't mounted a director a device to that folder then it's going to live on your SSD or your mvme where your boot Drive lives in our case that 500 gigabyte SSD keep that in mind it's a limited amount of space on there pseudo Mount is going to be our Command that we're going to use to mount our nvme to the plot temp folder and we're going to give it the discard async no data cow so butter FS is called a copy on right file system and we are going to disable copy on right on this and no data sum so we're not going to be doing any check summing either of the data not needed for things like Chia plots and we are going to specify the device that we want to mount which is the Dev nvme0 in1 which we have formatted up here to the the folder that we have created which is Mount plot temp which we created up here so this user and other users read and write access to the plots you may not want to have the O command in there but if you're mounting from multiple different usernames on multiple different systems you probably would and then we're going to do the same for the plot temp directory automatically next we're going to pseudo make file pseudo butter FS file system reset size and we're going to shrink 10 gigs off of the nvme this will allow us to have some spare blocks at the end of it so that if it needs to do any sort of garbage collection it's got extra blocks that it can maneuver things to that will help us out greatly here next we're going to do pseudo gigahorse and this is the command structure that you can have to fire off a plot automatically with our settings and if you note here we're missing our C and our F so I'm just going to go ahead and and grab the CNF from up here that we just grabbed we have the same rest of the information here so this should contextually all come together if it doesn't rewatch this a couple of times maybe run through it line by line and watch what's happening and then after that we're going to create a check of the actual plot file and we're going to use something called chiaa the proof of space and we're going to run that in the farm mode and that is with a t threads of eight actually this computer only's got six so it'll just do however many threads you have so uh just you can leave that eight or whatever D we're going to leave that at 100 that actually might be smaller for something like this computer and the actual file that we just created so it will create a file in this directory this is The Hyphen D the final destination is here and so you see the file that it's going to check is here and it will have some randomly generated name but it will have a plot extension so it's going to do all of this stuff really quick when we copy paste and run this so just slow things down a little bit maybe if you want to but probably just go back and do this a couple of times now after you reboot your system you would need to remount your uh in vme every single time I like to do it this way on systems for testing however when it comes time to actually put something into production you probably either want to create a mount script service or you want to look at putting it in your fstab this is going to be good enough for us to get up and running and test things out here today to the actual IP address 192.168.1.2 one1 and it's going to ask us if we would like to connect that fingerprint type yes and it's going to ask us for the password that we created when we installed the system so now that we're inside here let's run the first few commands and that should get us installed and everything set up running that we're going to have consistently there on this system make that a bit bigger for you and do a reset so it comes from the top down paste it in all at once and it's going to ask for our password as we are not executing as the natively root User it's going to automatically update all this stuff for us next and at the end it will reboot the machine as well you do have to hit enter at one point there after it is installing some of the software and it will take a while to install the Nvidia driver software as well just like it would on a Windows machine and it's now get cloning the gigahorse software into your user folder in this case that's DSP for me and it'll just dump you back out to your own Powershell command prompt here on your local machine as it's doing its rebooting so so the next thing that we're going to do is ssh in back again to the machine so the nice thing about using the terminal is that you can be lazy just press the up arrow and it'll bring up your last command that you ran which should be your SSH into that machine and so if you see the setup I've got here I ran a command called lsblk and this is listing the Block Level devices which are my SSD which is 500 gbes and you can see that it has SDA 1 and SDA 2 with uh one Meg here and uh the remaining space here this is the Mount point that you see here if you are unfamiliar with Linux this is kind of a primer on how you can read disc space next we see this next dis device here and this is nvme0 in1 and that is our 298 gibes of space the 320 is gigabytes of space and again the difference between those two is roughly about 10% 9% something like that and if you're in Chia you're going to learn that one really quickly so that is a for slev SL nvme0 in1 so in the commands that we're about to run here you're going to see that it is going to be addressed as for slev nvme0 And1 yet it's not here the next thing that we're going to do is also do a DF hyen H before we start this this will give you a layout of the actual allocation of space in the user space realm and you can see the for SL Mount that we have here for for the SDA to with 465.16 and now if we just press up up up up up a few more times there we go dfh now you can see we have the mounted Mount plot temp directory here and we have that as Dev nvme0 in1 and you can see that we've got 3.6 Megs used and 287 because we reduced the file size again if you recall 10 GB to leave some free blocks at the end of the SSD now our plot command which we have here there is one slight alteration that you need to make to it so that it will actually work on a sub 8 gigabyte card and that is the streams setting it to two the default is three so setting that to two will prevent it from crashing out here so let's go ahead and create the plot and test the plot you can see we've got our compression set at 17 as well on this and we are creating one plot it will kick off now and over here we've got our puny little P2000 and its memory is running at just 3.5 GHz and its GPU is running at 1.7 gz it is negotiating 3 at 16x that is important to make sure that you see at 16x whether that's Gen 4 or gen 3 you definitely want to see the at 16x so that you know that your PC uh bus is being fully utilized if it for some reason shows adx things will slow down quite drastically next after that we've got our receives and transmits and you can see the Gibby bites and mby bites per second that are happening there you can see the temperature is only about 43 44° because of the fan that we've got running at 50% on this card however we also have our external fan pointed at it keeping everything inside there nice and chilly and on a card like a Tesla P4 which we'll be testing out here soon that is actually going to play a bigger role than you might think you can see the p is the amount of work that is being done by the GPU on the leftand side currently sometimes going up there into the almost 60 watts range and on the right hand side 75 watts is the max because this does not have a power plug on it if you do not have a power plug plug but you have a 75 wat pcie slot it's many many pcie slots out there but you should check on your motherboard to make sure that is so then you should be able to run this and you can see the ma the MIM that we're actually using here is just 3.89 4 gbes currently so really cool that this can work on a 5 gigabyte total card very very cool big hats off to Max for making this work on such lowend Hardware over here you can see our reads and writes happening at just tremendous speeds when you're thinking about the reading and writing both going on at the same time like this and we're seeing multiple gigabytes in each Direction sustained and that sustained is what you get with pslc that you are not going to get with consumer drives and that right there is why this process goes faster than it would on something like an EVO 970 plus and you can see the amount of ram that we're actually using is just south of 5 gigabytes right now I think it will get a little bit higher than that but if you have 8 gigabytes you should be okay I didn't test it with 8 gabt because I wanted both channels to be populated and I just don't have DDR uh sticks that are 4 gigabytes so this is the best that I was able to do but I would imagine as long as you have two sticks populated in one channel each you're going to see some pretty similar times to this as long as your DDR is reasonably speedy this is some I think 3200 speed ddr4 that I have plugged into this system right now and we can see our phase one time is about 300 seconds our Phase 2 time is about 30 seconds and we are almost done with phase three phase four goes really quick and that is the Final Phase so we are very close to the end of our plot creation process and 27 seconds for phase three there so this is this is a good time this is actually a pretty good time when we look at it it's going to be right around 600 seconds almost 10 minutes to create a plot on an low-end system that's fantastic and I'm going to say it right now in case for some reason you don't watch these other exciting results come in because there's more to learn here but this is just any any machine Any machine is going to be able to create plots that are pretty darn fast even if you got probably eight gigs of RAM 16 gigs of RAM 32 gigs of RAM and a fast nvme the fast nvme is pretty critical here for getting that to actually work for you you can see that we are writing right around 450 to 5 550 uh megab per second here so this plot will be copied over to the SSD pretty quick and as soon as that gets copied over then it will perform the check on it and if you get this error with the chiaw that means that you don't have something installed here the lib open CL and this is what you need to install here this open CL ICD library and now we'll go ahead and check our Chia plot and this is going to give you an estimate of your max Farm size and if you are looking at doing some estimating on multiple different plots at different compression sizes could also set up a secondary card and that secondary card could also be yellow end card and you could use that for farming if you look at how much memory is being used as we're checking here you can see that 1.48 gtes is being used so that does give you some idea that possibly maybe something even like a 2 GTE uh GP would be able to hold its own possibly and do some solving so you could have maybe something like a 1030 uh doing some solving on something like a c17 at the same time plotting with something like a P2000 or even a 1060 and so now let's take a look at this and see what it's telling us so at a difficulty this is the partial difficulty which you could set on your pool of 100 and setting a difficulty is more important with these super high compression levels you would be able to it looks like Farm somewhere around 8381 Raw Pibb bytes of space at the 512 filter and that's good for probably up until sometime around June of 2024 when we will transition to the 256 filter and that would be 419 and this is an absolute potato of a card let me just say again this is a P2000 this is not a recent card and this is not a uh Powerhouse of a card at all so for the farming you can see performance benefits by going with bigger and bigger cards if you check some of the prior videos and you should check out all of these prior videos if you're getting into Chia farming this is the channel for you then you should see the difference between a 3090 and a p200b well let's just say night and day difference all right let's put in another card and take a look at what we're seeing as far as the plotting capacity capabilities of a different much higher end card we're going to go all the way to the 3070 on the potato machine to to do that I'm going to go ahead and pseudo shut down now and now you can see that we've got our 3070 uh much much better card here and we should see some differences especially on the farming side but take a second here drop a comment and let me know what you think is going to be the outcome of the 3070 plot time put 3070 plot time on that and yeah I can only say get ready this is surprising and cool so one of the nice things about the script and the way that I've got this kind of set up is this second half here you can just grab this whole thing at once toss the whole thing in there now we're going to get rid of the S directive here again like I mentioned that's for sub 8 gigabyte cards and so that does have a performance impact that is negative so we're not going to hinder it by doing that we'll just go ahead and set this back to that and we're going to keep it at the c17 everything else is the same we've reformatted re wiped and re getting ready to go on our farming here and plotting process so let's kick it off and if you know this is a 3 at 16 negotiation again because the b450 does not have pcie4 slots we're going to test this this same 3070 on another computer that does have pcie4 slots and so there we are with 561 seconds 9365 minutes and yeah the performance difference not not tremendous and I think that is what we're noticing that is important here is the fact that you really can get far without having a tremendous amount of plotting capabilities I mean this is an absolute potato system I would imagine there are quite a few Dell Inspiron 25 to $50 boxes that are local to you that you would be able to find and they would be able to do quite good for you again this is a setup and testing script that you need to run especially with the new difficulty settings that we've got with gigahorse if you haven't taking a chance to check out some of the live streams also they are long we cover a lot and they have some great stuff in them I would really recommend after checking out that Windows video which by this time you pretty much have to have to check out some of those as well and you can see the difference in farming for sure shines and you can see that 5.2 is a lot of pib bytes on a 370 2.6 pib bytes on the 256 filter so easily single card 3070 able to farm quite a lot at a c17 level and just wow the plotting capabilities that we saw between these two is not a huge difference at all and so we'll run a command really quick here called inoc and you can see that 32 is displayed that we run the htop command you can see 32 thread 16 core 5950xt toss up Envy top here really quick so you can see that we've got the 370 in here as well and we'll toss over glances on the side all right so let's grab our script here and get our plotting started and you can see the negotiation is happening at 4 at 16 X speed there for the pcie generation but the GPU is also not being utilized very much at all and we're currently at 21 183 so very high wattage being used in the table 4ish area and that's one of the higher wattage utilization times table four table 5 so definitely the workload increases quite substantially then in the Chia plotting process 223 we just hit there so yeah lot of extra Watts being used on this system right now but will our end time be dramatically faster 66 seconds 10.27 minutes wow so I think we can draw a conclusion here and that conclusion is really low in machine out there going with good in BME for your plotter rather than investing in a whole new machine just for a plotter I mean 10 minutes is not a bad time there's some other benefits to having a 10-minute plot time also that we'll get into uh they're not necessarily something you might think of right off the bat is being a benefit but if you have a plot time of about 10 minutes you're going to easily be able to clear off your temp space and move that to USB very very easily and we're going to do that next also going to show you how to calculate how many plots per second this is that you would be able to create in a full day that'll help you estimate when you would need to come back and change out your USB drive and put in a different one so at 616 seconds here can toss up the old calculator and we'll do 86400 ided 616 here on this 140 plots per day that you would be able to create now if you think of this in the terms of what that would be like for old school Chia plots well that would actually be really good to be honest with you that would be really really good on a single machine to do that that so let's take that 140 and times that by the size of the plot and so we will CH do 140 * 63 Gib bytes and that is roughly 8.82 terabytes a day on a potato you can plot 8.82 terabytes a day let that sink in one more time 8.82 terabytes a day on a potato literally I think you could build this for $200 a potato plotter that would plot 8 terabytes a day and you know it's it's a it's a very interesting proposition for people that are out there looking at USB drives and thinking how can I you know put some plots on them I've got some CPU I've got some GPU it's it's not good is what most people think however that bias is just not well founded to be honest with you because it's going to be okay is what we're seeing with 8.82 coming off of any one of these machines that we just plotted on because they all were right around 600 seconds 10 minutes per plot some amazing stuff right there and you can see that farming size we can hit 4.92 2.46 and so yeah that that tells us the story right there yes a 3070 can Farm much more space however plotting wise if you got only a potato Don't Go Breaking the bank to get out of the potato situation just get a GPU that's a lowend decent GPU for plotting and if you're okay with farming maybe you've already got something around uh 10:30 might be able to farm at certain sea levels pretty effectively and those are bus powered also so that's another thing to consider there's a lot of possibilities here a ton of possibilities you can see that we've got our 1030 down below we've got our Tesla P4 up above we've got actually a USB card here so that's providing four extra USB ports potentially you got two USB headers underneath here that you would be able to access I believe each one of those can give you a couple more USB ports and we've got four C slots you be able to utilize as well and we've got our two USB drives down here and on the back if you can see might have to turn on the flashlight back we've got 1 two 3 4 56 total USB slots so this little Franken machine may actually be a pretty cheap way to go if you're looking at setting up a small potato farm this might actually be something that you just have kind of hanging around and so my Franken farmer is getting right around 140 Watts with two gpus while one is plotting I don't have the 1030 farming yet but the 1030 is going to be farming here shortly and I'm pretty excited to see uh what kind of wattage we can get because this is um this is optimistic that we see a low wattage like this for a plotting and farming and this is a c17 so by running this statement here we're going to put the GPU that will be solving lookups into the GT 1030 so let me go ahead pull this up here and let me grab a quick listing of plots we've got a growing number of plots so next thing that we're going to do is be moving some plots after we're done with this by the way so yeah we got three plots in there right now so let me grab one of these and we're going to do our lookup against that plot I think we only have six here and we have the the difficulty set to 100 let's give it a check can see the gt1030 firing up there on all of its very few PCI Lanes cylinders and I will be back to check on what these results look like in a little bit here we're going to have a bunch of Windows set up by the time this is all done but I'm going to open a new tab and maximize this so that we can keep our eyes on the focus here so we're going to need to install a couple of additional pieces of software here so let's go ahead and SSH into the machine again and let's do pseudo app install Nano and we're also going to do screen I've already gone ahead and installed these so you'll see that it'll show as having these installed on my system on your system you'll need to press yes to install those once you have those installed we're going to go ahead and clone the latest gigahorse release next we're going to go ahead and extract and then remove the tar file after that's done now if we LS you can see that we have a CH gigahorse farmer folder here as well the copy. sh full uh script that we have is here this is going to be something that we set up in just a second don't worry about that one right now the next thing that we're going to do is jump into the gigahorse folder and take a look at what the structure looks like for the files inside here so you see the chad. bin executable in here that can be utilized by typing. bin and it's going to give you the list of commands Chia bin and anit and we already have the keys here so we don't have to do anything it'll prompt you for whether or not you want to do that next we're going to go to Chia Keys ad it's going to ask us to enter the pneumonic that we want to use so we're going to fire back up our Chia client here and we're going to copy these Keys again don't share your keys with people I'm sharing these keys because they are essentially wasted at this point but it does help people visually see exactly what's going on as best as possible so we've got the Linux wallet demo here and we'll go to details and I'm going to reveal the seed phrase and we're going to take that and minimize this back down paste it in here and we can assign a label to this so we'll call this a tiny plot so if you have the Windows machine set up you're going to have a copy of the full database that was how you were able to get the keys so I'm going to assume that you have that we're going to make a copy of that here and put it into here and so this is a 2 tbte SSD this will be the final drive that we add we're going to walk you through the steps of formatting any of the USB drives that you're going to use for plot storage you're going to want to follow that same pathway and here's a quick look at what the path looks like for you if you're a user that do IIA folder may not be visible so View and show hidden items I've already got this copied over here so going to go ahead detach this and take it over so now on our farmer device when we type lsblk again we'll see that we have a new dis that's showing up here that is the sdd drive and you can see that we have a partition that is sdd2 here now this is the partition that we actually will take now we're coming over from uh windows with this machine so we're going to actually need to install a couple of extra pieces of software here as well to be able to mount this next we're going to put in CD till. slet DB this is on our machine where the database is going to live if you do an LS hypen La you can see that it's a pretty small little file here and so what we're actually going to do is remove that so pseudo RM the blockchain file and now if we do an LS you're going to see that we have this empty you'll see that we have sdd here this is the 2 terab drive that is an SSD that I just brought over and it's USB based and we've got sddd2 this is the partition that we're going to be mounting so the next thing that we'll do is create a directory that we'll be using for mounting the drive and we're going to call that SL DB move Temp and that'll be at the root directory the way that we created that next we also need to install a couple of additional pieces of software if you don't have them on your system already that's NTFS 3G and rsync I've already install them and here we're going to mount the sdd and that is going to be changed on yours if you noticed I had that as sdx on mine and in the script you'll need to change that variable to whatever it is mapped to here and whichever partition it should be the large one most likely on the drive that you're moving over your database from and now we're going to use something called rsync to move the database from the DB temp folder and it's a do sqlite and there may be a path difference so I know mine is in something called Chia space DB on that machine in a specific folder and we're going to move it to the current directory and once it's copied over go ahead and unmount the temp directory that you created so now we're going to type in lsblk and we're going to see that we've got our 2 terab disk sdd plugged in here and you're going to go ahead and do a WIP FS hyphen fa and Dev SD need to type pseudo before that and do it one more time just to make sure that you got it all the next thing that we're going to do is create the directory that we're going to mount the Drive 2 and so we're going to use pseudo makeer Mount USB 3 for that and the next thing that we're going to do is format the sdd drive with our ext4 commands that will give us the most size here if you haven't seen this one yet this is a uh glad you're welcome welcome to the channel this is going to give you about 1.5 extra gigabytes per 4.5 gig uh terabyte drive so that that's the to pay attention to get a little bit of extra space everywhere you can and the next thing that we're going to do is Mount up our SD drive so pseudo Mount Dev sdd to our new Mount point which is Mount usb3 and now if we do a dfh we can see that we've got about roughly 1.9 terabytes of space on usb3 here so that's great the next thing that we're going to do after that is give it some permissions so that it can be written to and we're going to give that USB 3 and now if we do an LS La of Mount you can see that we've got USB 1 two and three here and they all have the same permission sets applied to them now if we do an LS you can see that we've got the Chia gigahorse farmer folder here CD into the folder we can start executing things by typing sl. bin and it'll bring up a list of commands that you would expect the next thing that we're going to do is we're going to add our usb3 directory with plots space ad hyphen D Mount USB 3 this will tell Chia that a drive that is holding plots lives here as well next we're going to move on to running the gigahorse farmer start farmer and since this is a fairly upto-date database it shouldn't take long for us to get the blockchain fully in sync and once we get fully in sync we will be farming and it does take a little bit of time for it to get up to the point where it starts to say syncing and after it gets there you should be able to do a Chia show s and get a status of where you're actually at syncing the blockchain you can see here we are pretty close according to the date so we are just a couple of hours out of sync on that it'll be up to date in no time and when we do chia Farm summary here you can see that we have 69 plots and that is a 4.24 five Tipp BTE on dis size that we've currently plotted and that leads us to 6833 Tipp bytes of effective space that are being harvested right now and you can see over here everything is just sailing along in the green pretty well if we go to EnV top really quick we'll see that we have a little bit of ization every now and then for the look up of a challenge on our pool and so let's check out while we're at this how you can see more detailed information about the pool you're on so once you log into your client here and you can do this from the CLI also but most likely you're going to have a Linux client or a full node desktop that you have for this setup you can see over here we've got these three dots view pool login link and we can copy this and open that up in a browser window and we have the difficulty set at one here so we should get be getting some challenges pretty quickly and as soon as we get the first challenge we're going to go ahead and update that difficulty to be substantially higher so that we don't punish these uh poor uh gpus because this is this is not a very big beefy GPU that we've got here with 69 plots the challenges that come through one in 512 currently uh passing that plot filter so we will be seeing uh it'll take a little bit of time but not too long for us to get our first partial submitted here and as we see there we just kicked up a little bit of activity on sdb that means one of our plots pass the filter and you can see it coming across here and that's going to be a challenge that it's going to be providing its answer to and we should see that reflect as a partial here shortly and if we come to our Harvester here you can see that we have three valid shares that have actually happened so far and so for most lowend machines you're going to be stopping and starting that's just kind of the way that it's going to work if you have just a single GPU and it does not have 12 GB of vram you're going to have to plot and then Farm not plot and farm at the same time so that's what we're demonstrating right here and most people out there this is going to be how that happens so next let's take a look at how to create some plots so we're going to go ahead stop things really quick and then we're going to assign a farming GPU and we're going to assign a plotting GPU so we're going to do this statement here and this is going to tell us to use the gt1030 for our farming operations when we start back up the Chia farmer and as you can see here this Dev ID here zero Maps up to this device ID up here and so this tells us that we are farming on just one GPU right now and that is the GT 1030 you can see that generation negotiation happening to at 4X as well there so once you start the farmer you don't have too much more to do so we're going to go ahead and close out of this tab now now the next thing that we're going to do gets a little bit more complicated because we're going to be plotting on one GPU the Tesla P4 because it does have 8 gab of vram and we're also going to be using the plot sync and and the plot mover copy script so really quick let me just go ahead and show you the plot mover copy script so in our base directory here you can see that I've got copy. sh so let me run cat copy. sh for you and you can see these lines here this is the code that you need to have and the DSP should be changed to whatever your username is to be able to run also make sure that you set T to 127.0.0.1 and we're going to be scanning for anything that has a Dot Plot extension so this will allow you to pick up the plots and it will wait 15 seconds and then just loop back and do it again and this will just run pretty much to infinity and wait for those plots to show up now it's going to look for those plots coming into this plots path which is on our SSD now this is going to be the D directory that we're plotting to on the same machine yes you can write direct to these discs however that incurs a massive stop penalty so to accelerate things you want to have a specific drive that is being written to and that specific drive then spooling off and that is not the destination directory that you want to have things will slow down dramatically if you don't do it pretty much just like this SL let pseudo doopy Dosh to get this running and it'll give you a failed to copy if there are not plots in the folder it'll wait 15 seconds and it'll give you another one of those so that's to be expected that's the normal behavior next let's take a look at our uh script that we're using for the sync now the plot mover sync is an amazing thing here and you can see we've got Chi gigahorse plot sync this is a command that you run once and it will not necessitate having a loop script for it because internally it's just going to see things that are coming in and it will automatically put those on the next available drive that is a good drive for those so let's go ahead and put the first one here as Mount usb3 and make sure you end those with a forward slash so we have three Mount directories here we have Mount USB 3 USB 2 and USB 1 so it's going to automatically distribute to one of those and the really cool thing is that if you have multiple plots which our machine's really not going to be able to do that much it can alternate them to different drives and balance things out so we'll let this run it'll give you a read out there on the size that you have available on each of these also and the port that it's listening on so the combination of these two will allow us to make plots move over to the directory really really fast now this part is pretty important because we're going to start creating the plots here and they are going to live in the plot temp directory and just like we demonstrated that nvme is the crucial part of keeping this operation up and running if you are a hyphen 3 which is very very low-end mode farmer the negative one here will just keep going until it fills up all those drives and the S is set to two because this is a pretty darn low in device that we've got here as well our compression level is set at 17 but we need to go ahead and just use one GPU and we also need to specify which GPU that is that we're going to use we don't want to use the 1030 to plot because it cannot plot it is incapable of plotting and the P4 is capable of plotting however the lowest one that I saw that could create a plot is the uh P2000 however I'm very interested if a 1050ti which is a 4 GB card would be able to and it may or may not be able to but do sound off in the comments below if you can test out one of those I'm just curious so we're going to use the hyphen G and one and that should put the workload for plotting onto the Tesla P4 nope it tried to do it on the Tesla p on the gt1030 okay so if it does that then just come back and change that to zero yeah this one was a little bit odd uh I forgot there is a interesting export and mapping kind of thing going on here and there we go now the Tesla P4 is doing the plotting as you can see the GPU is starting to get utilized there and we've got our device here waiting for the copy script to kick in and we've got the sync sitting there listening waiting for plots to show up and you can see the nvme device getting couple of gigs of traffic up and down here going on it so everything's working out here pretty fine you can look at the used space you have per each Mount Point here we have 2.1 2.09 and we have a lot of space that is just empty here so very little 24K used there so we've got this much left to plot and seeing that we were able to get so much work done so quickly in the past really has me feeling like the ability of our Farm to get pretty big pretty quick is really impressive because we used very little electric and also this just we don't have any gpus that need to be plugged into a power supply here also that's that's pretty cool so we'll let the first couple plots come in so you can see what this looks like as the plotting kicks in then I'm going to run you through what it looks like to have 12 hours of plotting continuously go this is uh something I recorded last night but I'm splicing it in here this is how we got this size here on these as well and I was slightly concerned that it might not be able to pick up shares uh partials with everything going on so I went back down here checked our proof times and indeed the proof time doesn't look like it's doing too bad so 4.5 4.96 so totally acceptable in my and as that plot finishes on the nvme this is the reason that having the setup the way we do is important because we now have this reading out of here and onto the SDA drive this is again the boot drive that I just am using that is a 500 GB SSD uh just a Evo 870 just something I had laying around and this will be a pretty Speedy copy right around 2 minutes or so but you can see that it's starting the copy and it's not stopping the plotting and that's important because you want to keep this process running at all times or else you dramatically add additional time and while we see 10.51 minutes as the actual plot time and indeed that is the actual plot time that we would expect when you add in the copy time to that things do slow down a bit so we'll see the second plot and third plot and success successive plots slow down just a little bit as that impacts the ability to read and write to the SSD so Table Three should be about where we see that plot finish its copy and so we see that it took 167 seconds to copy that plot over and almost instantly the script was able to pick that up just in that 15-second window and that starts the copy over here to one of the directories and we're seeing that it's picked USB 2 as the directory over here so you can see this being read off of the nvme or the SSD here and written to the platter hard drive here all while plotting is continuing that's the important part keep it going so there we have it a very good way for people with low-end machines to get up and compressed plotting with c17 and again these mvme are the reason why this time and this sustained time looks so good you could also replicate the same with several ssds or MVM in a raid zero or even platters many many platters in a raid zero array as your D drive for these however one nvme at $80 on shop. digitalspace port.com may be the best bet here for some folks and so after you're done plotting let's say you pulled out your Tesla P4 because you wanted to try to save every last Watt on your system and let's say you pulled out all three extra dim slots so you just have one dim in there at the time and you might be wondering what kind of Statistics am I going to get on that people that have 8 gigs of RAM 16 gigs of RAM have been wondering when me when me is answered with now and the timing of this to be able to plot at 13 to 10 minutes that's really good I mean I didn't think it was going to be this good with the low-end cards that we're using I mean what doesn't make sense but at the same time I'm glad it doesn't make sense shout out again to the massive engineering effort to get this up and running that Mad Max has put out there as well all right everybody I'll check you guys out next time

Channel: Digital Spaceport

Views: 15,279

Keywords: chia, chia xch, xch, chia guide, chia mining, chia farming, chia mining guide, chia farming update, chia plotter, chia plotting, chia plotting guide, chia farming 2023, chia mining 2023, xch farm, xch farming, chia network, budget chia farming, hard drive mining, hard drive mining 2023, gpu plotting, gpu plotting 2023, low end pc mining, low power mining, low power chia farming, low end pc chia plotting, chia 2.0, chia gpu plotting, how to farm chia

Id: zcbbZAQPdow

Length: 67min 12sec (4032 seconds)

Published: Fri Nov 10 2023