I’m just wondering, with all the talk about TBW for the NVME drives, how many people here have actually killed one of them because they exceeded the TBW number? I know that number is for warranty purposes. So how far have you gotten beyond the specs? I think it would be interesting. Maybe we can begin to get a list of real world endurance of specific NVME models. I am running Samsung 980 Pro’s and 970 Evo Plus’. But I am nowhere near the TBW number. But I’ll keep an eye on them and post the numbers when it comes time.
No I have not killed one, but can give you the answers you seek.
I am also using a 2TB Samsung 970 EVO Plus NVMe. It is warranted for 5 years or 1200TBW.
The Samsung Magician management software allows you exactly follow your TB writes and gives you fair abilities to monitor and repair problems with the drive over it’s lifetime. I hope to get a few more than 1200TBW out of the drive.
I have found that plotting one K32 uses about 1.2TBW on the NVMe. Simple math give me hope for over 1000 K32s plotted before drive failure.
Including tax, I paid $441.00 CAD for the drive.
My estimated projection: Each K32 costs under 50 cents in terms of NVMe replacement.
BTW so far, I highly recommend this drive for it’s cool running (usually 54C when running full blast. No extra heat sink used or recommended) and it’s management software.
You can decrease this cost if you buy the extra insurance at microcenter (about 10% cost of the item) - they are no questions asked on your first replacement under that…
I agree the 970 Evo Plus is a great drive. But temp really depends on your setup. I had heat issues with mine in the beginning. Had to add a heat sink with fan. Now it runs nice and cool. But I’m sure it is just the design of my case and the lack of air flow in that area.
I have no intention of warranting the drive. When it has done it’s hard work I will replace it.
Only if it failed at 600TBW or the like would I consider asking for warantee and even then, I would be up front about what I was using the drive for. I would hope that Samsung would honor their warantee even so if the drive had failed that early and had not been overheated.
I turned my system fans on full time in the bios.
If you have low ambient temperature in the case the EVO cards should be able to cool themselves quite nicely without a heat sink.
Given Samsung’s description of the cooling and even the cooling stickers I am a little afraid to use a heat sink myself … especially as mine runs at 54C now.
Reducing the temp is the most important issue so if your heatsink is working for you, of course, you should continue with it.
Also, be sure you have the washers under the NVMe so it sits parallel to the board with open space beneath. If it is slanting down towards the board this can present cooling issues. This is another reason I worry about the heat sink. With a heat sink you end up with almost zero space between the bottom of the sink and the board. Heat sinks are definitely required for most plotting NVMe but the EVO series, not so much.
I have seen PCIe to NVMe external extender/adapters. Keeping your NVMe outside your machine with a fan blowing on it would seem to be another excellent option.
Oh, I would never use the manufacturer’s warranty for it - that would be against the spirit of the warranty.
But, when the paid extra warranty was offered I initially turned it down explaining what I was using it for and Microcenter said they would honor it anyway, that it was no questions asked for the replacement. So, I took them up on it. Really, I figure that if that same warranty covers someone throwing a game controller at their TV and breaking it (it does), and I am being encouraged to make use of it, then I am not going to worry about it.
Well, lol, thats just sorta paying Microcenter to go, “against the spirit of the warranty” on your behalf.
Either way, the manufacturer is ultimately faced with the wholesale cost of replacement.
I’m not dinging ya fer trying to get your money’s worth, lol!
The extended warranty is third party insurance - the manufacturer is not involved or responsible for anything in it.
Also, for me to be more on topic, I have been using CrystalDisk Info to monitor the health of my drives, since it works with all manufacturers. It is showing that my NVMEs have about 80% left currently, but I am not sure if that is an objective measurement of the state of the memory on the drive or just comparing the write history to the manufacturers stated lifetime.
My evo plus (1TB):
SMART/Health Information (NVMe Log 0x02)
Critical Warning: 0x00
Temperature: 63 Celsius
Available Spare: 100%
Available Spare Threshold: 10%
Percentage Used: 40%
Data Units Read: 586.223.122 [300 TB]
Data Units Written: 532.540.286 [272 TB]
Host Read Commands: 866.355.141
Host Write Commands: 859.833.141
The 40% used almost corresponds with the 272TB written (vs 600TBW).
I think the Available spare is a better indicator.
The good news is that they set the warrantee threshold low so as not to have to honor it. In other words, they expect the drive to live and write more that its warranted 600TBW.
I have the 2TB EVO (warranted for 1200TBW) and hope to get 1500TBW out of it.
I have not had one fail yet. I have been using the PNY XLR8 CS3030 NVMEs. Not the most performant drive, but at the time that I purchased them the TBW rating was great. In fact, they appear to be basically the same core design as the Inland Premium using the P12 controller and Toshiba NAND chips. However, on May 17, PNY updated the specs and significantly lowered the TBW rating (without changing the part number). It seems that they changed the design too. Of the three drives that I purchased (together), one appears to be the new crappy design. They all have about the same amount of use, but one is reporting much less remaining life. If it does fail significantly before the others, I will probably try to get a warranty replacement.
And this is why I started this thread. As people get more and more use into their drives I hope they post here saying something like “Samsung Evo Plus currently at 1657 TBW and still running strong” or “WD SN750 at 576 TBW and died”. I think actual endurance numbers will help.
Will keep you updated and let you know when I fail, lol!
My drive is working well(ish) it performs well till the first phase of first set of plots finish in precise stagger time, the second set that starts has had its phase 1time double for some reason… I know it used to be faster…perhaps thermal throttling at that point… I am hesitant to check the health till I have finished plotting through all the drives…
If it is thermal throttling then it better to slow than kill, lol. Are you using a heat sink? Is the ambient temp around the NVMe? low?
What drive are you using? Does it have management software? How many TBW is it warranted for and how many has it written?
Lastly, why are you hesitant to check your drive health in between plots?
Are you regularly trimming the drive? Not doing so can reduce performance. I have a job set up that trims about once an hour.
I did not know about this…I will look into it…
I can’t do much about the thermals, I am on a laptop…best I can do is put it in an air conditioned room which I have…
You might want to get high end laptop cooling pad if you have not already.
Can you describe how I can identify the new crappy design? I’ve a unused spare PNY CS3030 2TB laying on my desk and if this is the crappy one, i will return it immediately.