Every plot I have made was done from the Chia GUI, lolz! 
I am considering an attempt at creating a K35 plot.
But my two temp drives are both 2 TB NVMe drives. So based on the above chart, I do not have enough temp space to accomplish my goal.
However, I see two possible solutions, and I am looking for anyone that might know the answer to #1:
#1:
If I direct -t to temp-drive-1 and I direct -2 to temp-drive-2, will that divide up the temp usage, such that neither of my two 2 TB NVMe drives will run out of space?
Or is there some step in one of the phases of the processing that requires more than 2 TB of temp space on one of the two drives?
#2:
I could RAID the my two 2 TB NVMe drives into a RAID 0. That would effectively show the OS a single 4 TB temp drive.
I would rather try option #1 if the -t and -2 options will get the job done. But I will go for option #2 if anyone knows that option #1 will not work.
I have 64 GB of RAM, by the way. I believe that is enough for a K35 plot.
U can plott K35 with 2x2TB, if on is -t and on is -2 temp folder. U can even plott K35 with 1x2TB and 1x1TB 
From my experience using 2 temp folders is faster than using Raid0 with 2 NVMe.
2 Likes
Were you using a hardware RAID or a software RAID?
A hardware RAID controller has its own processor, and does not tax your CPU.
A software RAID controller is maintained by your OS, and will use CPU cycles.
I never used a software RAID controller, so I do not know how much of a toll it puts on your CPU.
If you were using a software RAID controller, it could explain why your two temp folders performed better than your RAID 0 arrary.
yes i tested it with software raid (linux), thats true 
but i think it will not make a big difference with only 2 drives.
u should not use more than 10-11 threads with only 64GB RAM
I am currently plotting two K34 plots simultaneously, via madmax.
For each plotting job, I have assigned 14 threads, because my CPU has 32 threads (I figured that I should leave a few threads available for the OS).
So since my use of 28 threads (14 x 2) works, then why would you feel that I should use no more than 11 threads for a K35 plotting job? Why not throw 28 threads at a K35 plotting job?
If this plotts are staked so that they never at the same time in Phase 3 then this work for K34. If u also plane simultaneously K35 plotting than 7x2 will work, but with solo plotting K35@14threads will not work.
And u cant use more than 2xK35 simultaneously because of the low disk space of the temp folders.
I have been dual plotting K34 plots for a few months. So every combination of progress (phase position) between the two plotting jobs has been reached â including both working on phase 3, simultaneously.
With my CPU, two K34 plots with 14 threads, each, works every time, regardless of the phase of either plotting job.
Perhaps one day. But not in the near future.
My goal is to complete a single K35 plot. I will not complicate my attempt by kicking off more than one job.
So you gave me the green light on the -t and -2 options, which you deem to be the solution to me not running out of temp disk space.
But what about the 11 threads that you wrote that I should not exceed. Are you sure about that? I am asking because I have been using 28 threads for my concurrent K34 plotting jobs.
yes 2TB for each temp folder works
For me more than 23 threads and 64GB for K34 solo plotting leds to memory swapping at phase 3 table 6 & 7.
Can madmax even make K35 plots? I thought it only supported up to K34
HAHA rigth only the normal slow chia plotter can K35 
and there u can use 32 threads without problems
1 Like
My version of madmax was bundled with my Chia version 1.2.11.
As of 1.2.11, K34 is the maximum size.
Perhaps Chia version 1.3.3 comes with an updated madmax that can go higher? (probably not)
For my K35 plotting job, I would have to use chia.exe (not the madmax executable).
It will run for an eternity. I believe that is due to phase 1 being single-threaded, and will take countless hours to churn through phase 1. But I know of no other way to create a K35 plot.
sorry was my mistake i thought u will use madmax
so ur only problem with the slow chia plotter will be the free temp space but 2x2TB will work fine.
1 Like
How things change and advance⊠Funny to read this old post about K33s one-by-one in the Chia GUI. Right now, have 3 new drives going & have plotted 143 K33s currently, âtwo byâ after going for 6 days straight in MM wo/touching anything except trimming occasionally. one 1tb 980 Pro shows red for controller failure and 2.453 PB written, and two other 1tb ssds show 1.629 PB and 0% & another .779 PB and ok, yet they all just work pretty much like new. SSD vendors have just been trying to scare us with TBW rating, it seems.
With the new BladBit Disk coming w/higher k capable., âkâ plot speeds should be ratcheting up yet again (for ssd users). Too bad I donât have any more drives to use it with 
Perhaps itâs time to re-plot all the old k32âŠ
1 Like
I have been doing that for 2 or 3 months.
I have four 2 TB Samsung 980 Pro NVMe drives.
Combined, they plotted close to 11,000 K32 plots.
Now I have them re-plotting K34 plots (and K33 plots to fill in space where another K34 will not fit).
I am approximately 35% - 40% done with re-plotting, and my NVMe drives are running with no signs of trouble.
I have no idea how many TBW each drive has endured. But they must be passed the manufacturerâs rating?
I doubt that my drives are a special case. I suspect that nearly all SSDs will live long past their manufacturerâs estimated TBW value. In fact, I have yet to hear about anyone wearing out an SSD.
I have heard of SSDs failing. But I have never heard that the failure was due to TBW.
Fun work that!
Samsung magician is excellent tool for their SSDs, it will tell you TBW, plus other good things, diagnostics, performance, set overprovisioning, etc.
right now no. it will help to fill in the last gb you have left in the drive. i have won in the last 2 months 11 blocks and none of them on k33 or k34. keep plotting for now k32.
1 Like
Iâve been looking through various threads. This is still relevant. This is something that nobody talks about. People need to be aware. Why? Itâs because this comes directly from the horses mouth. Chia themselves. Many people talk about plugging a hard drive to the max. Consider what Chia says about this:
- Leave free space on your drive
If you end up having your HDD or SSD have any issues that require disk repair, you will need to have free space on the disk that is larger than the largest file. For k32 plots, you will need to leave > 101 GB. This way, if your plots ever have errors (as reported by plot check tool), you will at least be able to try to repair them with CHKDSK /r. However, CHKDSK cannot repair files that are larger than the remaining free space on a drive, and will have an error
That makes NO practical sense - NONE. I donât care who says so. Itâs backward thinking at its best.
On each disk one should leave at least one plotâs worth of space. Correct? Depending on Kxx size, this amount becomes bigger and bigger as Kxx increases. For the sake of argument, perhaps youâve got 50 disks full of k33 plots. Therefore you should NOT PLOT at least 50 K33 plots across those disk >>>> >>> on the remote off chance ONE DRIVE might someday become corrupt?
Pure insanity!!
IF and thatâs a big if, a drive become corrupt, simply delete and/or move ONE PLOT on that drive! Do whatever, and put ONE PLOT back!!
3 Likes
Endertown says fill them up boys. ( I had ask him the same question a while ago)