Ignatios Souvatzis wrote:
>
> On Tue, Mar 28, 2000 at 09:48:00PM +0200, Ignatios Souvatzis wrote:
> > A change that might be related fixes a cache syncronization problem on
> > 68040.
> > This might slow down certain operations, exec would be one of the victims.
>
> Just looked more carefully:
>
> - the configure test you report does lots of exec(). A horrible lot of.
> - the exec test tests exec speed, of course.
> - the sh test might well depend on exec() speed a lot
> - the dc test might also depend on exec() speed a lot, if it calls dc ot
> just execute one command, then executes it again in a loop.
>
> So a cache problem, or just former cache-not-properly-handled problem, will
> result in something like the observed performance hits.
>
Both systems use 68040 cpu.
Anyway most of the rest of the tests have a drop of 3% to 5%. I think they
should not have that much performance penalty.
I attach the results files obtained for amiga with kernels compiled by me
using the same options
Regards,
Inaki.Attachment:
byte.result-amiga-1.4.1
Description: Binary data
Attachment:
byte.result-amiga-1.4.2
Description: Binary data