On Saturday, July 19, 2003, at 06:54  AM, Naoto Shimazaki wrote:

> I'm not sure the implementation that masking interrupt on demand to
> prevent to touch hardware if possible really improves the final
> performance.  But I feel it might do.

It was a huge performance improvement on i386 when it was implemented 
years ago.  Also consider that poking the hardware is going to be an 
uncached access, and thus could defeat any memory access optimization 
that the memory controller might be doing.

         -- Jason R. Thorpe <thorpej@wasabisystems.com>