I’ve been saying this for a while, plotting is trivial, especially with all the plot speed code optimization in the last few months… all these plotting services are silly, the real service is in farming…
I’d bet many are just using equipment they probably should not have bought or rented to try and take advantage of the many who, understandably, want to fill up their storage yesterday … I just saw a post in buy & sell offering plots at 50 cents each. I might consider it at that rate if they shipped the finished plots to me on high TB drives with no mark-up. Probably not gonna happen, and the chance of fraud just makes the proposition more problematic.
Three K33s every 17.5 hours fer me! Chug, chug, chug. lol! I think I can, I think I can … Second plotter finally getting to work as soon as we receive a PCIe NVMe riser that I think is on a ship somewhere.
I’ve been saying that TBW is a warranty rating only for a while. Although I don’t plan on using my NVMe drives in Chia once they hit like 25% or less life remaining as reported by Crystal Disk Mark, I still plan on using them for less intensive tasks. They can still be good for something, just maybe not plotting.
For modern SSDs this is very true. For SSDs of 8 years ago, TBW was not a joke and drives seriously don’t last long after. Things have improved massively.
Just an FYI, write cycles void the warranty. Just because the manufacturer doesn’t specifically list TBW and instead says stuff like “rated at 5 years lifespan with average usage”, doesn’t mean that a warranty couldn’t become null and void before 5 years due to TBW. That is why I am never purchasing SSDs of any type that don’t specifically call out their TBW.
@juppin Thanks for posting the PNY CS3030 story and confirming many of my suspicions. I am glad to hear that the NVMe drives that I purchased all fall under the original warranty. Two of them appear to be wearing slowly, but one will probably wear out early.
I guess I’m going to get to see this first hand. I just checked my Samsung 980 Pro’s health and all of them are around 40%. So, since they need to finish filling my current space (which is almost double what is already filled) and then replot the space that already contains old style plots, I will easily run the health to zero. We’ll see how long they actually go. Will they make it through the plotting process that I require? Any bets?
I’m plotting with madmax on a Seagate FireCuda 510 2TB. I’m not sure if I ran the command correctly or not but after 240 plots, smartctl on the Ubuntu box reported that 416TB was written to the drive. I read from somewhere that roughly 1.4TB is written in order to create a plot so not sure if I did it wrong.
Basically the command looks like this:
Temp 1 drives on plotters 1 and 2 are SSD drives, not NVME’s. But that is now one drive that has ran out its expected life per specs. Below are the current details for that drive.
453,049 GB - Total Host Reads
1,151,103 GB - Total Host Writes
56 - Power on count
1045 - Power on hours
I am putting it on my calendar to go in and record this information weekly. I would like to know at what point it fails and I assume that once it fails, I won’t be able to pull this info. So I’ll try to keep on top of it.
I just did my weekly check on drive health and found something really odd. This time, I recorded the “Total Host Writes” for all drives. Even if they were above 0% health. What is odd is that one drive is at 2,123,897 and another is at 1,378,712. But the one at the lower number has been at 0% health for over a week (see last week’s notes). But the one that is at over 2 million just dropped below 0. So that means the same drive model had dramatically different warranty numbers coded in. One drive didn’t hit 0% health for almost double the written data. Anyone know why?
BTW, I now have 2 drives at 0% health. Still running strong. I will reach capacity later this week and will begin replotting the 3500 plots I have on Hpool.