On Feb 23, 2005, at 6:24 PM, Stephan Mertens wrote:
> When I repeat the same measurement on the same system
> but with LAM 7.0.6 the fairness seems to have improved considerably.
> Question for the LAM development team: Do you have made changes along
> the way from 6.5.1 to 7.0.6 to improve fairness in message handling?
> What algorithm do you actually use to prevent pending messages from
> starvation?
To be honest, none that I can think of. Every process tries to shove
as much data down the wire as it can. We really rely on the TCP engine
to do most of the flow control. It might be that your system simply
has a better TCP/IP stack than the last time you ran the tests (the
Linux kernel has undergone improvements since the release of 6.5.1).
Brian
--
Brian Barrett
LAM/MPI developer and all around nice guy
Have a LAM/MPI day: http://www.lam-mpi.org/
|
