Re: cache hit ratio on fill interval

From: Alex Rousskov (rousskov@measurement-factory.com)
Date: Fri Jun 01 2001 - 07:46:15 MDT


On Fri, 1 Jun 2001, Erik J Burckart wrote:

> I was interested in raising the cache hit ratio on the fill cycle.
> I know that this will just take longer...but I think it would be
> more representative overall of a real load.

It would be indeed. We use fill phases to fill the cache (fast), not
to measure its performance, but that should not stop you from doing a
more realistic fill.

> My two thoughts were to have the cache hit rate increase over the
> course of the fill cycle or to just have the cache hit rate at
> 25-50% the whole time. I imagine the first would take code
> changes...but is this changeable in a workload?

Both methods can be inplemented using PGL with no source code changes.
For examples, see recur_factor_* manipulations in workload files. For
documentation, try
  http://www.web-polygraph.org/cgi-bin/htsearch?words=recur_factor_

Alex.

P.S. You must subscribe to the list to post further messages.



This archive was generated by hypermail 2b29 : Tue Jul 10 2001 - 12:00:19 MDT