: > Any benchmark that loaded the system very heavily,
: > especially if it produced swapping.
: The differences shouldn't be that large any more. The asynchronous
: swapping code and the swap deamon in newer linux kernels help performance
: under load. Again, I'm more inclined to think that one system may be
: better on some hardware, while the other might be better at something
: else.
Well, I hate benchmarks because all they have nothing to do with
real life, but...
Linux really has POOR swapping code, and kswap improved
it insignificantly. This opinion is result of every day experience
rather than benchmarks. I have Pentium 133 64Mb AIC7881 and run program
that eats >=128Mb of virtual memory (certainly, matrices).
It sweaps through these matrices with period ~20min.
When I use FreeBSD (or the same program on Solaris2.3/SS20/64Mb),
I DO NOT feel that this program is running, moreover
the time of pass per 1Mb of swap does not decrease(!) up to 180Mb.
When I use Linux (just tested 1.3.89), it is disaster!
Hard disk (7200rpm, 4Gb Barracuda) grunts as piggy,
the time of pass per 1Mb starts to decrease at 64Mb and
at >=128Mb becomes ridiculous.
Note, that when I rewrote the program to swap "by hands",
this behaviour changed. Linux and FreeBSD became comparable.
Apparently, Linux swap code is not enough intelligent.
Anyone may check it.
Alexey Kuznetsov.