[Fwd: Re: [Beowulf] Cell in HPC]

[Greg Lindahl]

On Wed, May 31, 2006 at 05:38:55PM -0400, Mark Hahn wrote:

> why are these cases not nicely covered by the placement 
> instructions implemented in mmx and followons?  you can control 
> how a load or store behaves wrt different levels of cache cache.  

How do I effectively control placement when I don't know how big my
cache is?

How can I cache block loop nests when I don't know how big my cache
is?

IBM has experience with shared caches on Power4/5 -- and they seem to
do a good job of fairly sharing. The rest of us, well, we live in
FUD.

-- greg