[Beowulf] Terrible scaling with a Cisco Catalyst 2948G-GE-TX Switch

[Mark Hahn]

> is a cheap DELL switch (PowerConnect 2624, ~$300). I run Molecular Dynamics

relabeled SMC, iirc.  reasonable commodity (in a good way) hardware.

> I recently got another 32 pc's to build a new cluster, but this time I had to
> buy a more expensive switch (Cisco Catalyst 2948G-GE-TX, ~$3,000). To

definitely an older-generation switch - can't even do full line rate 
on all ports, since the backplane has just 12 Gbps bandwidth.
http://www.cisco.com/en/US/products/hw/switches/ps606/ps5502/
(which is feeling-lucky if you google the switch's name).

> my surprise, the scaling is just TERRIBLE, and after a lot of tests I finally
> found that the problem is the switch (I'm not sure what or why exactly, though).

bandwidth.  the switch appears to be a hopped-up 100bT switch,
kind of sad actually.

> exactly the same program I get very different scaling (please see Fig. 3 at
> http://chem.acad.wabash.edu/~trippm/Clusters/performance.php).

when your speedup flattens and falls, you know you hit a bottleneck...

> Any other thoughts will be appreciated too.

ebay it; maybe someone will buy it for an office (where the design is 
actually OK, since ports won't saturate for long.)  Cisco doesn't have 
any special magic in networking, certainly not in anything as commoditized 
as ethernet switching.