On Mon, 3 Nov 2008, Cyril Plisko wrote:
>
> Indeed, kmem_zalloc() has an overhead of added bzero(). However, on
> modern CPU bzero() should be pretty fast (I'd say if it isn't, it is,
> probably, a bug).

This is often a wrong assumption.  The bzero() can result in more CPU 
stalls and cache thrashing.  This is particularly the case when the 
cleared memory is not used right away so the zeroed cache lines fall 
out of the cache and then need to be restored from main store again. 
Software can not normally make the hardware faster, but it can run 
faster by being less abusive to the hardware.

Bob
======================================
Bob Friesenhahn
[EMAIL PROTECTED], http://www.simplesystems.org/users/bfriesen/
GraphicsMagick Maintainer,    http://www.GraphicsMagick.org/

_______________________________________________
storage-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/storage-discuss

Reply via email to