[MUD-Dev] Architecture (Cell Rebalancing)

Rossmann Peter peter.rossmann at siemens.com
Thu Jul 3 11:54:25 CEST 2003


-EDIT-
i just want to correct my mistake before somebody jumps on me :):

On Wed, 02 Jul 2003 20:15:52 -0700
"Peter "Pietro" Rossmann" <peter.rossmann at telia.com> [ME!] wrote:

> of course it is. even when inra-server latency is measured in tens
> of microseconds (be if numa/other san) or hundreds (be it ethernet),

I wanted to write Myrinet, not Numa. I haven't seen any numa
(non-uniform memory access) benchmarks.

Just for record:

  With MPI on Myrinet, the latencies are about 11 microseconds,
  corba/myrinet 20, corba/100Eth 150microsec, gigabit
  ethernet(VMI/TCP) about 70microsecs.  Infiniband is probably best,
  about 5microsecs.

Pietro
_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev



More information about the mud-dev-archive mailing list