Hi, I recently encounter slower plot from 20 min per plot to 160 min per plot. At first, I thought it might have to do with my nvme, however, I try put my m2 to my other plotting machine but it works fine. I tried clear cmos,ram testing, and reinstalled ubuntu. Still, nothing seems to work. I also checked my nvme using smartctl but there doesn’t seem to be any error. And this happend after I try using ramdisk as a secondary temp and my pc crash and force to reboot. Anyone got any advice here?
What is your nvme model? I’ve been using a QLC consumer drive that degraded to a crawl after putting some 100TBW on it - doesn’t seem like anything helps it.
EDIT: actually if the nvme works on different machine, the problem must be somewhere else. Are you sure ram disk didn’t overflow into swap? Doing so would burn your system drive in no time
I’m using Samsung 980 Pro 1 TB. Yeah can you explain on the overflow part I’m not quite sure how it works but I make temp of 120GB out of my ram for my 2nd temp ,and for my Ubuntu NVME drive I didn’t use it to plot and after I test using itl, it still running at slow as well .What confuse me is that after the crash both CPU and ram usage barely use like 20 to 30 percent out of total so right now I’m doing memtester for 2 loop and seems like it pass every aspect of test so it drives me crazy now.Lol!
From what you described, the ram disk is not properly using RAM, but instead swaps into system drive. I’m not an expert on linux, but I think you can check memory usage with this command:
htop
If you see swap being used, it means that memory contents is being written to disk, causing massive i/o and killing it. If your system drive is overloaded, it will destroy your operating system and all kinds of problems may occur.
For madmax, you need to have RAM disk no more than 110GB (per instructions). I personally use 107GB and rarely get errors
1 Like