[HH] solar powered supercomputer
Kurt Keville
kkeville at MIT.EDU
Thu Jun 21 16:19:35 EDT 2012
-snip-
>
> What was the name of the trash can you used? It obviously made a
> convenient sized enclosure, but aside from that, why was it chosen?
> Cheap? You happened to have one?
>
> Was there a compactor mechanism inside that had to be gutted?
We originally tried to acquire a BigBelly (http://bigbellysolar.com/) but could not afford it... the RubberMaid 9P90 looked the most like a BigBelly so we went with that...
>
> Did you build custom bracketing for the switches, USB hubs, and Pandaboards?
Yes, modeled on a 15 node design by a Raspberry Pi developer we met...
http://blog.chris.tylers.info/index.php?/categories/23-PandaBoard-Planet
USB hubs were Koutech 13-port hubs and we used USB to power the Pandas.
>
> Any comment on the many-core ARM parts that have been mentioned on this
> list previously? Do they lack the floating point hardware necessary to
> be interesting for this use?
>
We are waiting with anticipation on the availability of the ARMv8 architecture (64 bit) and ARM Cortex-A15, which may be out in Qualcomm-based cell phones already.
>
> Aside from the Pandaboard and Beagle board, are any of these things
> real? If so, they weren't mentioned.
>
We strayed from the agenda a bit for sure... appropriate links are...
http://designsomething.org/
and
http://raymii.org/cms/p_Small_Linux_PCs_overview
> I had the impression there was going to be more hardware demoed than
> just what Federico brought. (I guess Michael had 2 or 3 items.) Was it
> impractical to bring in the supercomputer? Was a Pandaboard shown, other
> than in the video?
There was to be more... I forgot the Trimslice at home, the Cotton Candy didn't come in and we decided not to show the Makey Makey...
http://www.makeymakey.com/
>
> The video talked some about the supercomputer hardware, but not in much
> depth. A walk-through of the architecture, interconnects, and software
> would have been nice. Was the presentation light on these details
> because you covered this ground in prior years? Was this talk mostly
> just a "here's where we're at now" update on the previous talks, rather
> than meant to stand on its own?
Yes, an interim report... we are't where we want to be yet but this was a good waypoint.
>
> I did like that Brian explained some of the motivation and practical
> benefit to a project like this - being able to house a supercomputer in
> a regular office environment without needing special cooling, power, or
> it generating excessive noise.
>
> Thanks Kurt and Brian for sharing your project with the group.
Yup... Brian and I will be populating a website with more info soon. Sorry, we were just flat out for a couple of days there!
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.blu.org/pipermail/hardwarehacking/attachments/20120621/e107ea4d/attachment.html>
More information about the Hardwarehacking
mailing list