Lei Yang wrote:
This guess actually is close to mine. But I'm wondering if there is a
sanity check, for example, if I make the network bandwidth to a very
big value, or if I use indefinite queue size, would the problem go?
Because if it is, then this may not be a problem of my protocol, but a
problem of Ruby configuration, because in actual hardware design, we
might have bigger bandwidth or bigger queue size. Right?
Well it certainly sounds like the increased messages are causing
additional queuing in the network. If you want the problem to go away,
the incresae the network bandwidth.
Dan, how do I monitor the queuing contention?
The delay_cycles that you are seeing exists in MSI_MOSI_CMP_directory to
do exactly this-- monitor queuing delays. See of the .sm files use the
profileMsgDelay() whenever removing a message from the network. I
believe this is the only protocol that currently profiles the delay cycles.
|
![[Computer Systems Lab]](http://www.cs.wisc.edu/pics/csl_logo.gif){kind=logo}