My guesses are close to the mark. I am not going to state exact figures, because that would require me to wipe a drive, re-plot with k32 plots, and when done, re-plot, again, to put back my larger plots.
1 Like
You can use the page I provided you before, and then a sheet of paper and a pencil. Thatâs why we do math to avoid what you just outlined as an excuse.
Just to verify if your math is sound, you can compare one k-X to another k-Y plot.
1 Like
I have exactly the same hardware I had, from when I had k32 plots, compared to today, after re-plotting.
All along the way, as I replaced plots, my estimated time to win never changed (other than slightly, as network space changed).
Chia is reporting the same âTotal size of plotsâ as it did when I had 100% k32 plots.
Everything is the same, except my âkâ sizes, which remedied my overloading my harvester.
Explanations, for the sake of transparency, are not excuses.
Had I not explained that my figures were a guess, would you have let that go, or would you have commented that my figures were inaccurate?
I do not have exact figures, and I stated so. That is not an excuse. Everything that I wrote was true.
There is no down-side to using larger plots, other than your estimated time to win taking an insignificantly more amount of time for a win.
If your wins are estimated at 6 months, then 6 months and a day will not matter.
If your wins are estimated at 3 months, then 3 months and 12 hours will not matter.
I am guessing at those figures, but I should be close, based on my own re-plotting of over 10,000 plots.
1 Like
On a 18TB drive you can fit 165 k32 plotsâŚ
1 Like
And what machine did you use to replot 10,000 plots? And how long did it take to do this task?
1 Like
I used two 3950x computers, both outfitted with two 2-TB Samsung 980 pro NVMe drives, and both with 64 GB of RAM. All plots created with madmax.
It took months. I do not recall when I began re-plotting. It might have been November of 2021 (I have k34 plots from then), and I finished approximately two weeks ago. But I also had a few empty drives back then, too. So I am not sure if they are re-plots or not.
2 Likes
how long will k32 last?
2 Likes
I have undertaken the same exercise. Replaced pretty well all of my K32 with K34 and K33 depending on what was appropriate for a particular drive size.
I have a XEON 5280 with a 1TB SSD and a 480GB RAM drive for the temp and temp-2 folders respectively.
Using madmax from the CLI, where you can utilize the overlap, it takes me just under 4 hours to do a K34 plot.
So about 6 plots per day.
Using HD Sentinel to monitor the SSD, it drops 1% per day. So 600 K34 plots and you are into the dodgy area of the SSD.
In practice I got a few extra days before I started to get some strange plots created. Put in a new SSD and all OK again.
As expected no change to estimated time to win.
The log still shows a few entries creeping over the 5 second limit, but that doesnât seem to cause SpacePool to report a problem (Stale plots etc.)
I think my Synology NAS, which contains some of my plots causes the delay as it does a lot of other stuff at the same time. I donât think the Synology DSM is that good at multi tasking. Seem to prioritise writes over reads. Sorry I have started to digress off-piste.
3 Likes
Until someone breaks the, what was it⌠30 second plot window I believe. Apparently if you can plot fast enough (and you have cheap power), itâs essentially proof of work. Requiring bigger plots will start that quest all over again 8). 3.14 Relevant Attacks and Countermeasures | Chia Documentation (short range replotting attack halfway down).
3 Likes
I figure that by the time k32 become obsolete, Iâll have about 300 XCH.
If itâs a successful project by that stage, I probably wonât be worrying about replotting.
If it isnât a successful project by that stage, Iâve probably backed the wrong horse
4 Likes
Even if its broken it doesnât magically print money. It can be done now but the equipment costs would be higher than the XCH gained by 10x or more.
2 Likes
(and you have cheap power), I keep telling people to run the power cord next doorâŚ
2 Likes
Does anybody outside North America have cheap power anymore ?
2 Likes
I am not familiar with HD Sentinel. I do not know how it arrives at its 1% per day figure. But Iâll take a shot at the reason:
I suspect that it knows the manufacturerâs advertised TBW (terabytes written â lifetime) value, and reports on your SSDâs health by comparing your lifetime number of writes to that advertised TBW value.
My take on the TBW value is:
- The manufacturers are not sure how how many bytes their SSDs can endure.
Take Chia, for example. I have been hammering away on my SSDs for over a year, 24/7/365, and see no performance degradation.
So unless the manufacturer has some other way of determining how many writes their SSDs can endure, they played it safe with an estimate that they are sure will be high enough.
So even if their SSD can handle 10x the number of writes, the monitoring software is still rating the SSDâs health based on the advertised TBW value.
And the manufacturers are not going to pound away on their SSDs for years, to ascertain the true endurance of their SSDs, before putting them on the market. So the manufacturers have to come up with a number.
-
A percentage of folks that check on their SSDâs health will see it being reported as âCriticalâ, due to exceeding the manufacturerâs TBW value, and will replace it (buy another SSD).
-
When you exceed the TBW value, you end your warranty (now isnât that handy for the manufacturer?). Your drive could still have 90% of its endurance remaining, but you hit their magic âno more warrantyâ number.
I do not run any 3rd party software (I keep my Chia systems clear of all 3rd party software). But I suspect that my SSDs would probably have reported âCriticalâ a long time ago, and yet they show zero sign of any issues, and run no slower than when they were new.
Assuming that my assumptions are correct, then I would not worry about getting into the dodgy area of the SSD.
I have never heard anyone report that their SSD wore out. And this forum is the place where that would be reported. Yes, SSDs will fail. But those failures will likely be due to a manufacturerâs defect, or the user took a power hit.
When I got my first generation i7 CPU (when it was all the rage), I wanted 1 TB of SSD storage. That was expensive (approximately) 15 years ago. So I purchased four, 240 GB OCZ SSDs, and put them in a RAID 0 (cost less for roughly the same storage capacity).
I never had an issue with them, until ~year ago. One of them kept dropping out of the array. I never used the full capacity, so I rebuilt the array with 3 of the drives. Why did one drive fail? I do not know. But it took 15 years. I do not think that I wore it out. The other 3 drives are working fine.
2 Likes
China, India, Iceland, Costa Rica, and others.
3 Likes
I guess itâs just Europe suffering thenâŚ
2 Likes
And California+New York.
2 Likes
thanks for all of your answers. i had no idea the thread will touch so many hearts.
now, shorty and literally in two words. is it good idea or not?
superstitious block below
also what about my superstitious idea about k32 being not that good for farming as it posted in white(green?) papers, i was asking if any of you sense something, you know sense by your ass.
to be clear i donât give a thing for warnings the winning probability is all that bother me.
2 Likes
All Iâve heard is it shouldnât make a difference, I had crap luck when I had just K32, and I still have crap luck now I have a mixture.
I have 1767 K32, 627 K33 and just 4 K34.
K34 was just an experiment, but I used K32/K33 to max out my drives, I have an average of 3.328GB spare per dedicated Chia driveâs, of which there are 19.
2 Likes