Re: Kernel code and processor's Cache




The designers have already devoted as much chip area as makes economic
sense to cache.  Therefor every line of kernel cache means one less line
of user cache, and the kernel cache will be idle almost all the time.
It makes much more sense to trade off the cost of the rare occasions on
which the kernel flushes useful user data out of the cache for the
benefit of maximizing user cache all of the rest of the time.

--
John Hasler
jhas...@xxxxxxxxxxx
Dancing Horse Hill
Elmwood, WI USA

Thanks John. I am convinced by this argument as i have stated above. I
was just just thinking why not cut off a bit of L3 cache to cache
only kernel routines. what impact will this have on the system
performance ?
.



Relevant Pages

  • Re: Kernel code and processors Cache
    ... David Schwartz writes: ... What if that cache contained data that had been modified and was ... isn't cached in the dedicated kernel cache. ... have the separate kernel-only cache for security reasons. ...
    (comp.os.linux.development.system)
  • Re: Kernel code and processors Cache
    ... Therefor every line of kernel cache means one less line ... of user cache, and the kernel cache will be idle almost all the time. ...
    (comp.os.linux.development.system)
  • Re: Linux box reboots some times without reasons
    ... Great for pin-pointing when the failure occurred, ... the date command and the ps command is not something I would worry about. ... The problem of the cache not being written out is much more worrying. ... >cache, but it wouldn't force the write all the way (kernel cache, the ...
    (comp.os.linux.misc)