2008/10/31 Giorgos Keramidas <keramida@freebsd.org>:
> Nice :)
>
> I didn't notice this in -hackers, but it may be worth letting the buffer
> size grow above 1 MB when we have more than, say, 512 MB of memory. By
> running dd(1) with buffer sizes of 1, 2, 4, 6, 8, 10, 12, 14, and 16 MB
> there are larger buffer sizes that reduce involuntary context switches
> even more.
>
> My laptop (with more than 2 GB or memory), with powerd disabled, running
> in single user mode, and copying a file of 200 MB using the script:
>...
> got the following results for sizes < 10 MB:
>...
As I read it, for x memory you have y context switches:
64 kB - 11194 (old, pre-patch behaviour)
1 MB - 1493
2 MB - 1381
4 MB - 1342
6 MB - 1292
8 MB - 1275
16 MB - 1246
The difference between the best and the worst is 247 context switches
(total, not per second).
> It looks like it's probably worth pushing the limit up to 8 MB when we
> have lots of memory :)
Yes, it's an improvement, but who will ever notice? :)
I have no strong reason against increasing the buffer for cp to even 8
MB but I really think it's not worth it at this point. In the patch,
I've actually allowed the buffer size to grow to 2 MB if MAXPHYS ever
gets bumped up (and would gladly increase *this* limit up). I think
that, at this point, increasing MAXPHYS will give better performance
than increasing this particular buffer.