[openib-general] Configuration of sdp

[Karun Sharma]

Hi Sean:
 
If you are using OFED 1.1 release, you will find a "README.txt" file (/usr/local/ofed/ directory). In this file, configuration steps of SDP are mentioned. 
1. modprobe ib_sdp
2. export LD_PRELOAD=/usr/local/ofed/lib/libsdp.so
3. export LIBSDP_CONFIG_FILE=/usr/local/ofed/etc/libsdp.conf <application name>. 
 
You may need to modify libsdp.conf before step 3.
 
Thanks
Karun
 

________________________________

From: openib-general-bounces at openib.org on behalf of Sean Hubbell
Sent: Tue 11/28/2006 10:34 PM
To: openib-general at openib.org
Subject: [openib-general] Configuration of sdp



Hello,

  I have a question... we have tested our network on our cluster using
iperf v2.0.2.

Our first test was using Gig-E network and the results were 941 Mbps.
(i.e. iperf -c <ipaddress> and iperf -s)

Our second test was using Gig-E network and the results were 1.05 Mbps
using UDP.
(i.e. iperf -c <ipaddress> -u and iperf -s -u)

Our third test was using IBv4 network and the results was a max of 3.8
Mbps using TCP.
(i.e. iperf -c <ipaddress> -P[1-16] and iperf -s)

Loaded ib_sdp module and then tested:
Our fourth test was using IBv4 network and the results was a max of 4.16
Mbps using UDP.
(i.e. iperf -c <ipaddress> -b1000M -P[1-16] and iperf -s -u)

UNLOADED ib_sdp module and then tested:
Our fourth test was using IBv4 network and the results was a max of 4.16
Mbps using UDP.
(i.e. iperf -c <ipaddress> -b1000M -P[1-16] and iperf -s -u)

My question is what impact would it have on our network if configured
sdp and is there an example of how to configure sdp somewhere on the wiki?

Thanks,

Sean


_______________________________________________
openib-general mailing list
openib-general at openib.org
http://openib.org/mailman/listinfo/openib-general

To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openfabrics.org/pipermail/general/attachments/20061128/2c8ef001/attachment.html>