madMAx43v3r / chia-plotter

Apache License 2.0
2.27k stars 664 forks source link

Ubuntu 20.04 Crashes on ~phase2 #593

Open aknowel opened 3 years ago

aknowel commented 3 years ago

Hi, I was recently plotting by official chia plotter and everything worked fine. Wanted to upgrade to MadMax and while plotting after some time (it depends on number of cores I use) whole system crashes (it freezes, nothing moves on screen, video play stops and mouse not moving, can't do anything except reset button). When I use 1 core it freezes on ~phase 4, while using all 64 cores it passes pahse1 and freezes on phase2. I use Ubuntu 20.04 AMD Threadripper 32 cores 64 threads, plotter is new bought about 2 months ago.

Could you give me any advice what can I do?

cyperbg commented 3 years ago

What is your memory voltage? Try to increase it.

naruto4999 commented 3 years ago

What is your memory voltage? Try to increase it.

Hi, So I have kind of similar situation too, I am using ubuntu 21.04 and my whole system freezes after table 5 or 6 in phase one and then I have to force power it off. For temp dir I am using 3* 1tb gen4 nvme. I even tried plotting with using 50gib of ramdisk as bcache and something odd hapenned, after table 2 or 3, I got a warning saying "system overloaded in last 5 min" and the process just stopped, I could see it in glances but it consumed 0% cpu after table 2 but the system didn't freeze but it didn't pass table 2 it was just stuck there. Both times I used 16 threads and even 10 threads too but no difference. I am using 5900x and 64gb of 3600mhz ram. Should I try to increase the voltage also?

cyperbg commented 3 years ago

Not sure, but maybe it was running out of memory?

langubtc commented 3 years ago

+1 Ubuntu 20.04 system dead

asve99 commented 3 years ago

+1 for me too. I am running Ubuntu 21.04, AMD 5900x, 32GB RAM using 2 x 1TB nvme in RAID0 as temp1 and 2 x 1TB nvme in RAID0 as temp2. Occasionally plotting completes, but more often than not the system completely freezes in table 3 and requires a hard reset.

naruto4999 commented 3 years ago

Not sure, but maybe it was running out of memory?

I don't think it's running out of memory since 9 or 10gib was still free. Also the official plotter works fine will 11 plots in parallel it's only madmax causing problrm

aknowel commented 3 years ago

It is not lack of memory problem, because I have 64GB of RAM and I was monitoring it whole time MadMax was running and never went more than 10GB of allocated memory, I don't have experience in boosting and maybe increasing memory voltage would help but I am very curious what is going on with this freeze, is MadMax changing some deep config of operating system, changing memory voltage or something like that? There are no problems on Ubuntu Server probably but I need to test that on my machine

asve99 commented 3 years ago

+1 for me too. I am running Ubuntu 21.04, AMD 5900x, 32GB RAM using 2 x 1TB nvme in RAID0 as temp1 and 2 x 1TB nvme in RAID0 as temp2. Occasionally plotting completes, but more often than not the system completely freezes in table 3 and requires a hard reset.

Fyi, I reinstalled Ubuntu today using 20.04 Server on exactly the same hardware and have made several plots without the same freezing/crash (previously it would freeze every 2nd or 3rd plot). So at least in my case it appears to be something to do with 21.04 or the fact that it was the Desktop version.

naruto4999 commented 3 years ago

+1 for me too. I am running Ubuntu 21.04, AMD 5900x, 32GB RAM using 2 x 1TB nvme in RAID0 as temp1 and 2 x 1TB nvme in RAID0 as temp2. Occasionally plotting completes, but more often than not the system completely freezes in table 3 and requires a hard reset.

Fyi, I reinstalled Ubuntu today using 20.04 Server on exactly the same hardware and have made several plots without the same freezing/crash (previously it would freeze every 2nd or 3rd plot). So at least in my case it appears to be something to do with 21.04 or the fact that it was the Desktop version.

I'm using Ubuntu 21.04 Server version