[HH] Figured I would post this question here...

Kurt L Keville kkeville at mit.edu
Fri Oct 30 09:25:35 EDT 2015


Seems like the logical place for it... and so far Tom and Bill on this list have the best suggestions... if this isn't hardware hacking I don't know what is...

Here's my problem... it might have an electronic fix but it more likely is an EE101 and NEC code thing...

I have 2 (20A) Geist PDUs...   http://geistapp.geistglobal.com/specs/1136
I have 3 (12A) servers...  http://www.supermicro.com/aplus/system/2u/2042/as-2042g-6rf.cfm

In the aggregate I am fine on current (36A < 40A) but my grains are a little course here... is there a way to load balance the 20A services? There are some rather esoteric tie-bar  and breaker ganging ideas out there but I would have to convince an electrician I would deploy this box only under certain conditions (like the upstream wiring being specced for 40A)...

We spent a lot of time upgrading the servers; I suppose I could downgrade them to the point that I could put 2 on one circuit... this is an odd problem and stems from the SCC we are entering next month... kind of a dumb rule seeing as we are supposed to be solving real-world problems...
http://sc15.supercomputing.org/conference-program/student-programs/sc15-student-cluster-competition/scc-competition-rules
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.blu.org/pipermail/hardwarehacking/attachments/20151030/fe363805/attachment.html>


More information about the Hardwarehacking mailing list