THE CHIA FARM
- Thread starter gb00s
- Start date
Notice: Page may contain affiliate links for which we may earn a small commission through services like Amazon Affiliates or Skimlinks.
So the HP is under control. There is a ILO4 firmware hack that presents the fan and sensor config to the ILO shell so now the fans run at 50% rather than 100% and the CPUs just get up to around 70c. Much more acceptible noise and 35min plots.
Spacepool leaderboard is interesting reading. Top farmer has around 2.5PB and has replotted all of that in 7 days (when 1.2 came out). Almost 360 plots an hour. Based on a 30 min plot time that would be around 29 machines plotting.
At 45TB replotted I have now jumped from 1.5k in the leaderboard to 750 or so. I need to beat 140TB to get in the top 100 though
. Target is likely to be moving up though as more people replot.
version 1.2.2 is out but not so many documented fixes.
Spacepool leaderboard is interesting reading. Top farmer has around 2.5PB and has replotted all of that in 7 days (when 1.2 came out). Almost 360 plots an hour. Based on a 30 min plot time that would be around 29 machines plotting.
At 45TB replotted I have now jumped from 1.5k in the leaderboard to 750 or so. I need to beat 140TB to get in the top 100 though
. Target is likely to be moving up though as more people replot.version 1.2.2 is out but not so many documented fixes.
I started farming second week of May.
End of May, number of plots 1,800 End of June , number of plots 4,900
Time to win: shortest period is 6 days, longest period is 15 days.
Won 10 XCH,
last win was on July-4.
9 days sincelast win , waiting for the next win this week ( finger cross )
Still trying to decide continue with solo farming using OG plots, or re-plot to join a pool.
4,900 plots / 200 plots per day = 25 days.
4,900 plots / 400 plots per day = 13 days, it will become a job to produce 400 plots per day.
What to do????
End of May, number of plots 1,800 End of June , number of plots 4,900
Time to win: shortest period is 6 days, longest period is 15 days.
Won 10 XCH,
last win was on July-4.
9 days sincelast win , waiting for the next win this week ( finger cross )
Still trying to decide continue with solo farming using OG plots, or re-plot to join a pool.
4,900 plots / 200 plots per day = 25 days.
4,900 plots / 400 plots per day = 13 days, it will become a job to produce 400 plots per day.
What to do????
The main advantage (small as it is) of pool vs solo at sizes that get regular payout (within acceptable timeframes) is that you theoretically get more today for your percentage of the total then you get tomorrow (due to total space increase),
so theoretically you could earn a bit more by pooling than by running solo (assuming average payout is the same, 0% fee, and continuing pool increase).
Whether thats worth replotting... not sure.
Not for me at this point with declining Chia price and energy prices over here...
Still plan on setting up an optimized plotter with MadMax so i can plot while the sun powers the plotter and then slowly replot without extra energy cost...
Have not started though, its not really much fun at this point
so theoretically you could earn a bit more by pooling than by running solo (assuming average payout is the same, 0% fee, and continuing pool increase).
Whether thats worth replotting... not sure.
Not for me at this point with declining Chia price and energy prices over here...
Still plan on setting up an optimized plotter with MadMax so i can plot while the sun powers the plotter and then slowly replot without extra energy cost...
Have not started though, its not really much fun at this point
Mad Max has a small update which allows you to add a thread multiplier (-k) for the second phase. It seems that the second phase runs faster with more threads whilst the first may not. The Github issue and details can be found here.
Whilst I have updated, I have not played around with it yet.
fellow sth users, i need your help with madmax
my plot time for 1 plot is around 3 hours and 45 min. how can i get faster plotting time's ??
my hardware running windows 10 version 21h1 and using madmax version 0.1.1 and version 0.1.2a
- dual epyc 7551 (totall of 128 threads)
- 512 gb Ram ( 16 * of 32 Gb, 2666 mhz)
- hp h240 raid card pcie v3 x8 (raid0 with 2 * 900Gb 10k Sas-6G 2.5 inch hdd) >> this my temp 1 map for plotting
- 440 GB RAM disk (made with imdisk) >> this my temp 2 map for plotting
what i have tried:
- different threads (-r 4, 16, 24, 32, 64, 100)
- different bucket size (-u 32, 64, 128, 256, 512)
it doesn't matter what i for -r and for -u as values try, i never get plotting time lower than 3 hours and 30 min.
am i doing something wrong? or is plotting time of 3,5 hours is normal ??
in totall i have 4 hdd's of 900gb 10k sas-6g.
should i make raid0 with all of 4 hdd's to gain maximum speed for temp 1 map ?
any suggestions ??
my plot time for 1 plot is around 3 hours and 45 min. how can i get faster plotting time's ??
my hardware running windows 10 version 21h1 and using madmax version 0.1.1 and version 0.1.2a
- dual epyc 7551 (totall of 128 threads)
- 512 gb Ram ( 16 * of 32 Gb, 2666 mhz)
- hp h240 raid card pcie v3 x8 (raid0 with 2 * 900Gb 10k Sas-6G 2.5 inch hdd) >> this my temp 1 map for plotting
- 440 GB RAM disk (made with imdisk) >> this my temp 2 map for plotting
what i have tried:
- different threads (-r 4, 16, 24, 32, 64, 100)
- different bucket size (-u 32, 64, 128, 256, 512)
it doesn't matter what i for -r and for -u as values try, i never get plotting time lower than 3 hours and 30 min.
am i doing something wrong? or is plotting time of 3,5 hours is normal ??
in totall i have 4 hdd's of 900gb 10k sas-6g.
should i make raid0 with all of 4 hdd's to gain maximum speed for temp 1 map ?
any suggestions ??
I have the same situation. Probably slow re-plot will do a bit more income, but I'm bored with this stuff, just left old plots and switch plotter to new ones to fill up the last 100TB from 1PB total.
Why would anyone use anything except RAM if there's 512Gb of it? Just use ramdisk for the first and second temp. You may try even faster plotter - harold-b/bladebit
this plotter is not available for windows as compiled binaries. i am not able to compile the source code for windows. above my capabilities.
and i also have tried RamDisk for temp1 and temp2 drives....it didn't go faster than the normal way
Probably, that's good point to install Debian/Ubuntu and make things much easier)
ImDisk is shit, NTFS sucks balls, you need to move over to Linux and make a tmpfs temp 2. The learning curve is steep, but oh so worth it.
Umm, you can use the RAMDisk for 4 x temp 2 directories, that will be the most efficient thing to do for scaling up.
With 512 GB's of ram, you can have one parallel plot completely in ram and have another plot with a tmp 2 RAMDisk.
On the other hand, if you invest a little more (4 X SAS 10K drives) for EACH parallel plot to be the temp 1 directory, you can get WAY more plots in a given period of time.
Yes, should be about 2 times faster.
Here's what I would do with 512 GB of RAM:
Test on Linux with 2 X SAS drives as temp1 and compare it to 4 SAS drives as temp1. If there is a big enough difference, I would go for a setup 16 SAS drives, 4 drives for each temp1, doing 4 in parallel.
If you see diminishing returns, then 8 drives would be enough, 2 drives for each temp1.
Last thing I did before ending the CHIA adventure was testing plotting totally diskless. Only using HDDs for storing the plots. It works. Just 1x single China E5 2666 v3, 512GB RAM and that's it. I wasn't able to achieve the advertised 220-240 plots per day, but 168-174 plots with a single machine as direct I/O really hurts. CHIA plots are just to big to be able to go just with the flow One-by-One. Going with a temp of 4x365GB ioDrive2's in RAID0 just got me up to 191-198 plots per day. But this was not the point and is cheating if you are testing diskless plotting.
But my CHIA adventure ends here while moving to another country and sending my servers on a 3-4 week journey. Then I will need them back for some business work. Just learned a lot more about storage solutions. So thumbs up ...
Good luck everyone ... It's your thread.
But my CHIA adventure ends here while moving to another country and sending my servers on a 3-4 week journey. Then I will need them back for some business work. Just learned a lot more about storage solutions. So thumbs up ...
Good luck everyone ... It's your thread.
Finally got some sort of Splunk view together for my 2 plotters.
The Failed copied box is to report if a Plotter hangs due to no disk space left. My Temp1 drives are around 1G NVME so can take another 9 or so plots before it all grinds to a halt so hopefully this will highlight before that occurs.
I can sort out alerting in Splunk but another job for another day I think. That and trying to sort out some colour scheme rather than the random Splunk one.
Chia is down to USD166 now and my rewards have gone from 0.016 to 0.008 per block so not sure how much longer I will go on with this.
Choice is now whether to buy a 12TB drive or two Chia coins....
The Failed copied box is to report if a Plotter hangs due to no disk space left. My Temp1 drives are around 1G NVME so can take another 9 or so plots before it all grinds to a halt so hopefully this will highlight before that occurs.
I can sort out alerting in Splunk but another job for another day I think. That and trying to sort out some colour scheme rather than the random Splunk one.
Chia is down to USD166 now and my rewards have gone from 0.016 to 0.008 per block so not sure how much longer I will go on with this.
Choice is now whether to buy a 12TB drive or two Chia coins....
I am still solo farming with OG 4,900 plots
Won 10 XCH , 4 XCH in May, June , 2 XCH in July-4
Experimenting with space pool.
Set up a new full node in a separate network with different public IP address.
126 new plots.
Last 24 hrs , 0.003312 XCH
For 4,900 plots
4900/126*0.003312=0.1288 XCH per 24 hours.
3.864 XCH per 30 days.
About 4% less XCH than solo.
Won 10 XCH , 4 XCH in May, June , 2 XCH in July-4
Experimenting with space pool.
Set up a new full node in a separate network with different public IP address.
126 new plots.
Last 24 hrs , 0.003312 XCH
For 4,900 plots
4900/126*0.003312=0.1288 XCH per 24 hours.
3.864 XCH per 30 days.
About 4% less XCH than solo.
yea with your slower tmp drives, but that much ram, bladebit would be a good option, worth switching to ubuntu and test it out if theres no windows binaries. or test on a nvme disk.
can always spin up a win10 on kvm if you need windows aswell
I did around 0.019 in the last 24h (around US$3) with 720+ plots. This gives a ROI of around 20 years. I am still replotting and have space for just over 1k total plus another 12TB on the way but I really don't see any value buying any more storage now.
With spacepool you need over 300TB to get in the top 100. I am now ranked at 1,321
with 76.5TiB estimated.
how do you do reploting ? what commands ? madmax plotter or something else ??I did around 0.019 in the last 24h (around US$3) with 720+ plots. This gives a ROI of around 20 years. I am still replotting and have space for just over 1k total plus another 12TB on the way but I really don't see any value buying any more storage now.
With spacepool you need over 300TB to get in the top 100. I am now ranked at 1,321
do you need extra hdd space for replotting ??
say you have 91 plots (10 TB hdd), do you need another 10 TB disk for replotting ?? or is it done on the same hdd without the cost of extra (empty) space ??