Source-Changes archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]
Re: CVS commit: src/sys/netipsec
> Arguing that gcc does a nice job would likely be met with skepticism,
> especially in the absense of good, hard numbers for various data sizes
> on various CPUs. FreeBSD has a fairly well optimized bcopy for
> different cpus. There is also a long history of distrust of gcc's
> ability to inline memcpy in a non-lame way. You can search the
> archvies for a number of instances where hand written memcpy or bcopy
> routines did much better than gcc.
Indeed, on a modern x86 you do not want (ever) to execute movs{b,w,l}
unless it is repeated AND the repeat count is considerable
(unless you are optimising for space).
ISTR that something like:
1: mov (%esi,%ecx,4),%eax
mov %eax,(%edi,%ecx,4)
dec %ecx
jnc 1b
is faster than 'rep movsw' for %ecx < (about) 16.
David
--
David Laight: david%l8s.co.uk@localhost
Home |
Main Index |
Thread Index |
Old Index