[Beowulf] ethernet performance testing

George Wm Turner turnerg at iu.edu
Fri Mar 17 07:53:04 PDT 2017


A good place to start is with the Energy Sciences Network (ESNet) site https://fasterdata.es.net <https://fasterdata.es.net/>   Lots of good data on  optimizing ethernet networks; though, they tend to focus on long haul issues.  Still well worth a look though

iperf3 is the latest version; ESnet has some recipes utilizing nuttcp; but, iperf3 can do those test now that older versions of iperf didn’t have the capability to do.

George Wm Turner 
High Performance & Research Computing Architect
turnerg at iu.edu <mailto:turnerg at iu.edu>


> On Mar 17, 2017, at 10:33 AM, Lux, Jim (337C) <james.p.lux at jpl.nasa.gov> wrote:
> 
> This is a bit of a blast from the past question.. Back when Beowulfs were the “new hot thing”, we all used Ethernet as the interconnect
> There was a bunch of traffic on the list about measuring Ethernet performance using various utilities, and all the ways to optimize the performance (zero copy Ethernet drivers and the like, some interfaces just being slow, etc.).
> 
> What’s the current strategy and tools for looking at the Ethernet stack and identify?  I have a colleague who is working with a software radio peripheral (USRP) that streams samples out, and he’s getting better performance with USB3 than with GigE - it may be that this is just the way it works USB is dedicated master/client, while GigE is peer to peer.   So I suggested he do some back to back testing using a pair of PCs to know what the limit is.
> 
> And he turned it around and asked “so what tools can I use?”
> 
> iperf?
> netcat?
> 
> I’ll dig through the archives too..
> 
> 
> James Lux, P.E.
> Task Manager, DHFR Space Testbed
> Jet Propulsion Laboratory
> 4800 Oak Grove Drive, MS 161-213
> Pasadena CA 91109
> +1(818)354-2075
> +1(818)395-2714 (cell)
>  
> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
> To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20170317/bfa4f3a0/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 3988 bytes
Desc: not available
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20170317/bfa4f3a0/attachment.bin>


More information about the Beowulf mailing list