Re: [Gems-users] Checking new protocol


Date: Wed, 15 Feb 2006 09:59:57 -0600 (CST)
From: Bradford Beckmann <beckmann@xxxxxxxxxxx>
Subject: Re: [Gems-users] Checking new protocol
> Would test a 16-proccessor SMP configuration with the MOSI_SMP_bcast
> protocol (one processor per chip), for a duration of 10000. I am unsure
> what the unit of duration is--I suspect it may be memory requests from
> "processor 0", though I could be mistaken.

The unit of duration is different depending on the tester.  For the random
tester, the duration unit is a successful completion of a four stores, one
load sequence, otherwise known as a action-check pair.  The four stores
write four sequential values to a subblock, i.e. false sharing.  Then a
load is performed to ensure all four stores succeeded.  The loads and
stores are issued by a random set of processors.

Brad


[← Prev in Thread] Current Thread [Next in Thread→]