On Sun, 2 Sep 2001, Eiji Kawai wrote:
> With PolyMix-4 workload, the participants must provide an
> auto-generated report to polyteam by Sep. 11. I guess I also have
> to provide the report with WebAxe-4, however, the platDur
> parameter in webaxe4-guts.pg is set at 1 hour and the generated
> report complains "Experiment duration (169.91 minutes) is less
> than 240.00 minutes and may be too short." In addition, the
> documentation about WebAxe-4 on the WWW says the top* phase lasts
> for 4.00hour.
I suggest that you run the test with platDur set to 4 hours. We need
more information to determine whether one hour WebAxe plateau is
sufficiently long to collect performance measurements. We know that 1
hour is not enough for PolyMix, but WebAxe is easier on the cache and
may reach steady state faster.
> Do I have to change the value of platDur parameter?
Please do.
> Is the about 3-hour test OK for the pre-cache-off report?
Yes. However, I would ask you to submit a longer test so that we can
determine if shorter tests are OK. If all we get is "short" test
reports, we may not have enough info to make a decision.
> Thank you and I hope to meet you at cache-off.
Same here!
Alex.
This archive was generated by hypermail 2b29 : Mon Feb 06 2006 - 12:00:19 MST