Anton Aylward wrote:
On 03/04/2020 10:18, Carlos E. R. wrote:
El 2020-04-03 a las 14:24 +0100, Richmond escribió:
Excuse me for going on about swapping. I am using the same thread so anyone who has killed it won't be bothered. I was running firefox and had plenty of tabs open, but reasoned that if these are swapped out (that is to say the memory used by them is swapped out) then it won't matter. But the system ground to a halt. What puzzles me is how long this goes on for. I am not sure how long it would take to swap the whole 1.5 G of RAM out but it seems to be stuck for much longer than that. Because "swap fragmentation" causes excesive disk head movements. That is why this is about 'paging' and not 'swapping'.
The old UNIX V7 model or roll-in/roll-out was 'swapping' one program for another and it worked by a 'DMA' (often on the PDP-11 hardware-assisted on a separate bus) of contiguous memory to contiguous disk sectors, hence one seek at the start of the transfer then just smooth head movement.
Please stop referring to what is 'paging', often unoptimized one-page-at-a-time, demanding, as Carlos says, excessive head movement, as 'swapping'.
Like the old adage says
Virtual Memory means Virtual Performance.
I can see that you are making an important distinction, but I should point out that "swap" is an English word, and that "swapping" does not imply any particular object which is being swapped for any other particular object, so it could mean swapping pages. Additional confusion arises (for me at least) because the parameter is called /proc/sys/vm/swappiness. Does it mean paginess? -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org