Subject: Re: copy performance
To: <>
From: David Laight <>
List: port-arm
Date: 03/21/2002 12:23:45
On Thu, Mar 21, 2002 at 12:15:04PM +0000, Richard Earnshaw wrote:
> > I don't see a massive problem in having cpu dependent code in the
> > kernel.  It is a bigger problem for user space code.
> Nor do I, provided that the cost of indirecting through a function pointer 
> (rather than just using BL) doesn't become a problem.

I was actually thinking of the problem of deciding what the fastest
code was - where you want to load a new routine every minute
after changing a single instruction, then rerun a quick benchmark.

I was also going to run some tests with an I-cache invalidate
prior to each copy call.  Probably simulates real life better
- esp. for copyin/out.


David Laight: