This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: prefetch optimizations


On Thu, Jul 10, 2003 at 11:08:38AM -0700, Mike Stump wrote:
> On Thursday, July 10, 2003, at 04:37 AM, Niavis Panagiotis wrote:
> > I'm not sure if this is the right mailing list to post this.
> 
> No, its not.
> 
> > I'm using the function block_prefetch(follows) to prefetch a 4Kb array 
> > in the cpu
> > cache.
> 
> You might want to check out __builtin_prefetch in the gcc manual.  
> However, notice that cache lines are usually 32 bytes, not 64, on some 
> platforms, the lines increase to 128 bytes.
> 
> > Can someone explain me or point me to some documents which explain this
> > behaviour?
> 
> You will want to read documentation on your processor's architecture 
> and specific implementation.  You will want to read the documentation 
> on your memory controller.  You'll probably want to read up on locking 
> regions in memory on your OS.
> 
> Also, you can go view memory copy routines from all over the place, 
> like the Linux kernel.
> 
> Also, you can ask google for pointers...
> 

Also see http://gcc.gnu.org/projects/prefetch.html, which has general
information about data prefetch support for several architectures that
GCC supports and references for more information.

Janis


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]