On Fri, 25 Aug 2006 00:01:06 -0700 (PDT) David Miller <[EMAIL PROTECTED]> wrote:
> From: Andrew Morton <[EMAIL PROTECTED]> > Date: Thu, 24 Aug 2006 23:20:24 -0700 > > > On Fri, 25 Aug 2006 09:48:15 +0400 > > Evgeniy Polyakov <[EMAIL PROTECTED]> wrote: > > > > > kmalloc is really slow actually - it always shows somewhere on top > > > in profiles and brings noticeble overhead > > > > It shouldn't. Please describe the workload and send the profiles. > > Not that I can account for the problem in this specific case, in my > experience cutting down kmalloc() calls matters a _lot_ performance > wise. > > For example, this is why we allocate TCP sockets as one huge blob > instead of 3 seperate allocations (generic socket, IP socket, TCP > socket). > > In fact, one of the remaining performance issues in IPSEC rule > creation is that we allocate seperately hunks of memory for the rule's > encryption state, the optional hash algorithm state, etc. Part of that will be cache sharing between the three structs though. - To unsubscribe from this list: send the line "unsubscribe netdev" in the body of a message to [EMAIL PROTECTED] More majordomo info at http://vger.kernel.org/majordomo-info.html