Post by 888 on Aug 15, 2020 20:25:22 GMT
This is a bit specific to my setup, but it might be useful to someone in the future.
I noticed that my syslog file was becoming huge - over 1GB for a text file over quite a short period of time. My machine then crashed on 3 separate occasions - when it had been extremely stable before I fitted the second video card. The culprit was the GTX1060, and the way my machine was handling communication to the pcie bus and this card.
It appears that this is common with boards that have the Intel X99 Express chipset and certain pcie cards. My RTX2070 is unaffected, but the GTX1060 creates a huge amount of errors per second. The card does work, with no noticeable effect on performance - at least until the machine crashes.
The solution was to change the way this communication occurs, from Memory-Mapped PCI Configuration Space to using the older, less efficient I/O ports - there doesn't appear to be any significant performance hit so far.
This was achieved by adding the kernel option pci=nommconf to the /etc/default/grub file. This is for Linux and Windows will have its own method.
GRUB_CMDLINE_LINUX_DEFAULT="quiet splash pci=nommconf"
My machine has been stable and fast since I changed this. And the syslog file is small, as it should be.
Like I said, this is likely to be of no use to anyone, as it is specific to my setup - but it could be useful info......