From: Prakash Velayutham (Prakash.velayutham_at_cchmc.org)
Date: Tue Aug 11 2009 - 11:40:07 CDT
Thanks Axel. Thats what I thought.
On Aug 11, 2009, at 12:25 PM, Axel Kohlmeyer wrote:
> On Fri, Aug 7, 2009 at 4:02 PM, Prakash
> Velayutham<Prakash.velayutham_at_cchmc.org> wrote:
>> Does anyone here use 10GbE interconnect between compute nodes to
>> run NAMD
>> jobs? What kind of performance improvements can be expected
>> compared to a
>> Gig ethernet network?
> the improvements should be very little. your main problem are the
> latencies introduced
> by the TCP/IP protocol. current myrinet is also using 10GigE fabric
> but uses special
> switches and an MPI library that bypasses the TCP/IP layer.
This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:53:09 CST