On Sun, Dec 02, 2001 at 11:46:06AM +0000, Martin J. Laubach wrote:
> | | but for libkern case, could single port have different CPU models
> | | that have different cacheline size?
> |  
> |  Yes.
> 
>   Just as an aside, none of the other bzero implementations I
> looked at (linux, darwin, hurd) did anything other than just hard-
> code "32" in there.

BTW, there has been a thread about this topic (the number of bytes
in a cache line on various PowerPCs) in comp.arch recently.

Here's one relevant article; "view complete thread" for the rest.

http://groups.google.com/groups?
selm=bruce-887C3D.22094422112001%40news.paradise.net.nz

(join those two lines.)

-andy