Re: excessive swap-in time

phil-news-nospam@xxxxxxxx wrote:
If we could:

1. Correctly identify pages that won't be needed for a long time in the
future (those that have not been used for a long time in the past are
an estimate of probability for this)

2. Ensure that excessive outswapping caused by heavy I/O, especially heavy
writing, does not happen

3. Ensure a process swapping back in after dormancy does not cause some
of itself to swap out (assuming the working set can fit)

Then we would do well with less RAM and more swap. And I for one would
like to see the kernel reach this utopia.


You should do some tests with FreeBSD. In the past, when i had a machine
with very limited memory, it was doing quite well in this exercise.


--

Michel TALON

.

Relevant Pages

  • Re: [PATCH -mm 00/24] VM pageout scalability improvements (V12)
    ... In previous tests on my 16GB system, a 16GB fillmem (goes into swap) ... saves enough CPU time to make up for potentially worse detection of ... the working set. ... Professional qsbench users will be pleased;) It could have ...
    (Linux-Kernel)
  • Re: Is working set limited to 8GB?
    ... > drfremove@nber.org (Daniel Feenberg) writes: ... (If it's running a CPU flat out, ... Sorry, no swap ... We don't think it is merely a coincidence that the working set is 8.0 ...
    (comp.sys.sun.admin)
  • Re: newbie: redhat 9.0 very slow
    ... | recommended to be double the RAM.have u selected 256MB or more of swap ... The "twice RAM" is simply a rule of thumb. ... At any one time you have a "working set" - the pages of programs that are ... time paging (exchanging the less wanted data in memory for immediately ...
    (RedHat)
  • RE: why swap at all?
    ... > set has exceeded physical memory, and in that case, swap is just doing what ... the working set is usually beneficial. ... However where there is physical RAM ... send the line "unsubscribe linux-kernel" in ...
    (Linux-Kernel)