Hi,

given the number of page faults which occur is there a formula to
calculate the effective demand paged memory access time?

Thx

Alex

Sure, set up a program with a data array bigger than main memory and
then access each page only once in a big circle. This is almost what
we do to measure the access times of the various layers of the cache
hierarchy. All you need to do is make it big enough so that demand
paging has to be going on.

Secondarily, you might need to make the access patern something that
the OS cannot predict to prevent the OS from prefetching pages into
your demand set. Thus, if you put a pointer to a semi-random virtual
page in every page, you could simply dereference the pointers as the
whole loop. Just takes a little setup.

<smawiaskud@farifluset.mailexpire.com> wrote in message 
news:cd2e68c0-705f-48b0-86fd-92c07516f2a0@t54g2000hsg.googlegroups.com...
> Hi,
>
> given the number of page faults which occur is there a formula to
> calculate the effective demand paged memory access time?

No

Peter