Ham Radio

[Kyle Yankanich]

Hey folks,
 I was discussing with a few people in the IRC channel (ginman, peejay, doctor) about getting out ham radio licenses. I have my Tech ticket from when I was younger, but I've forgotten nearly everything about it, and would like to re-study, and consider upgrading. I was just curious if anyone else here was interested?

[Jim Fisher]

10-4

> --
> To post to this group, send email to hive76-d...@googlegroups.com
> To unsubscribe send email to hive76-discuss...@googlegroups.com
> For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en


--
jim fisher
Jedijf

irc freenode #ubuntu-us-pa
www.myfisher.org

"Do, or do not. There is no 'try.'"
-- Jedi Master Yoda

[andrew sooy]

i would be interested in getting mine 

[Jordan Miller]

looks tasty.

jordan

[PJ Santoro]

Also interested :)

sent by telephone

[Anne Farbman]

This is a software question. I'm starting to get into computational genomics, which as you can guess from the first word is a data management problem start to finish. There's generally a good deal of parallelism- usually the first few steps of any problem are solved identically over an entire 3Gbp (for humans) genome and no partial solution depends on any other partial solution until you've pared down your data to a much smaller set of things to care about. Obviously this is a good candidate for distributed processing, but I'm realizing I don't really know enough and was hoping someone would explain a few things to me:

I know from projects like SETI@home and FoldIt that its possible to get people to run screensavers on home computers with a particular purpose. My question is, is there a toolkit to build a really really tightly walled garden (say, one that can only access a single database on a server) that can then execute scripts/java? In other words, is there an app that people can run that lets you send them different tasks to process? Could you run the "AnnesHomework" app and one day its looking for promoter-sequence homologies, and then the next day its assembling a phylogenetic tree, etc, without having to re-download it every time the job changes?

So, does something like this already exist? I mean, that isn't a botnet. This is kind of a well-harnessed botnet, I guess. Ooh, maybe some of you crypto people could come up with some other uses for it.

A

[Jordan Miller]

AFAIK you're better off doing GPU computation on a single computer. the overhead of distributing a simple genome search job like this is, I think, much more complex than the job itself.

are you looking for something at the scale of SETI@home or something like just running in parallel across a local 6-computer cluster?

jordan

[ComSec]

This seems to have gotten attached to an existing thread rather than creating its own thread but I will entertain an answer. It seems like you are looking for some kind of distributed computing or grid computing framework.

http://en.wikipedia.org/wiki/List_of_distributed_computing_projects

NGrid may be a suitable open source candidate.

[Dan Shookowsky]

Map-Reduce and a Beowulf cluster of Raspberry PIs?

--

[Louis Gerbarg]

The people who run distributed screensavers usually have an interest
in the specific thing they are running, so running arbitrary code is
less interesting to them, especially since running a computationally
intense screensaver actually costs a non-trival amount of money (a
machine that pulls 500 watts out of the wall is going to cost end up
costing the person running the screen saver $15 a month assuming
normal utilization times and power costs).

The only significantly deployed system I know of that does this sort
of thing is Apple's internal fuzzing system. They have a screen saver
that can run arbitrary programs and fuzz them, and it can actually
launch VMs and fuzz their kernel. Of course in that case they achieve
a direct improvement in the quality/security of their product by using
all that electricity. Given how specialized it is they have no real
interest in releasing it.

Unless you need truly massive resources over an extended period I
would suspect using spot priced ec2 instances is usually a better
choice.

Louis

[Sean McBeth]

I've been considering a project like this, written in JavaScript, specifically to allow people to contribute cycles without having to install anything. Basically, one could just load up a page in any web browser and be contributing to the project right away. 

But as to your question, yes, there are frameworks for this. What you're talking about is more specifically called Grid Computing. Check out this list of APIs and middleware:  http://en.wikipedia.org/wiki/Grid_computing#Standards_and_APIs. I haven't used any of them, so I can't speak to any of their particular suitability.

As to Jordan's point on General Purpose GPU programming, you'd be looking at using something like CUDA. This will be significantly easier than getting a grid setup, and a grid will only start to reap benefits if you hit a large installed user base. I don't know what your budget is like or what the scope of your project is, but if you have about $700 to spend and you don't expect anything more than, say, 500 users joining your grid, I'd stick to GPGPU. 

Dude, a Beowulf cluster of GPGPU setups would be nutso kablamo.

On Wed, Sep 5, 2012 at 12:30 PM, Anne Farbman <anne.f...@yahoo.com> wrote:

--

[Daniel Toliaferro]

Oh Jordan, you never disappoint.

I'd be interested in watching you guys operate a radio at Hive.

- Dan.

[Anne Farbman]

That's interesting. I hadn't thought about looking into the commercial cloud. Honestly, I don't pay for the electricity in the lab and if it takes a while to comb through a data set, it doesn't particularly trouble me too much. There are other computers nearby, too, so even if it means I have to leave my workstation for a week, it isn't a crisis. Looking forward, though, data bottlenecks are a huge problem, especially if you're doing genome-wide stuff, and the discussion last week about thermal cyclers got me thinking about how hackerspaces could help garage biotech become a reality. Tying together a few hundred smart phones into a virtual cloud supercomputer would be a cool place to start.

(By the way, one of the labs on campus here just upgraded and got rid of a butt-ton of equipment. The standard practice is to put all your obsolete stuff in banker boxes in the hallway, and after a week of letting the scavengers have at it, taking it to the surplus auction. No thermal cyclers but there was a centrifuge.)

I looked at the wiki page (thanks) and it seems like the BOINC platform at least works for developers, if not as an arbitrary-code machine. There do seem to be a lot more crypto applications than bioinformatics, but that may just reflect the interests of the sort of person who can make something like that go.

Anyway, thanks for the responses.

Best,

A

---

From: Louis Gerbarg <lger...@gmail.com>
To: hive76-d...@googlegroups.com
Sent: Wednesday, September 5, 2012 12:51 PM
Subject: Re: [Hive 76 Discussion] Distributed processing question

> To unsubscribe send email to hive76-discussion+unsub...@googlegroups.com

> For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en

--
To post to this group, send email to hive76-d...@googlegroups.com

To unsubscribe send email to hive76-discussion+unsub...@googlegroups.com

[Sean McBeth]

A decent idea, though working this to scale would probably be cost prohibitive. You're looking at $35 for a 700mhz processor. One can probably find someone giving away a 700mhz computer for free in about a week. People have tons of used machinery that they can't even get rid of. Old, Pentium 3 and 4 machines. Smartphones that have been tossed aside in favor of newer models. PDAs that got obsoleted by smartphones. If you're budget is severely limited, it is a viable alternative to GPGPU.

However, there is the Parallax Propeller chip, which is $7/ea for an 8 core, 80mhz-per-core system, that is specifically built for parallel processing. You'd have to program them in assembly, and you'd lose a core as a "manager", but you could hit some pretty awesomely parallel levels with it. No, that's insane, that is way too much work. Don't pay attention to me :)

[andrew sooy]

Are we aloud to put an antenna on the roof?

[Sean McBeth]

No, but we have access to windows on either side of the building and we're one of the tallest buildings in that part of town.

[PJ Santoro]

Unfortunately, that's unlikely :/

sent by telephone

[Kyle Yankanich]

Windows on both sides? How did we pull that off?

Also, totally possible to run a radio off one side of the building, but something like a repeater would obviously operate much better with a large, unobstructed range. 2m radios are the most repeated, and support digital modes easily, antennas can be around 3 feet, or shorter if need be.

- Dan.

> To unsubscribe send email to hive76-discussion+unsub...@googlegroups.com

> For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en


--
jim fisher
Jedijf

irc freenode  #ubuntu-us-pa
www.myfisher.org

"Do, or do not. There is no 'try.'"
  --  Jedi Master Yoda

--
To post to this group, send email to hive76-d...@googlegroups.com

To unsubscribe send email to hive76-discussion+unsub...@googlegroups.com

For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en


-- 
To post to this group, send email to hive76-d...@googlegroups.com
To unsubscribe send email to hive76-discuss...@googlegroups.com
For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en

--
To post to this group, send email to hive76-d...@googlegroups.com
To unsubscribe send email to hive76-discuss...@googlegroups.com
For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en

--
To post to this group, send email to hive76-d...@googlegroups.com
To unsubscribe send email to hive76-discuss...@googlegroups.com
For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en

--
To post to this group, send email to hive76-d...@googlegroups.com
To unsubscribe send email to hive76-discuss...@googlegroups.com
For more awesome goto http://groups.google.com/group/hive76-discussion?hl=en

[Com Sec]

I have to ask....why not the roof?

[Sean McBeth]

"Are we aloud (sic) to put an antenna on the roof?"

"No..."

That's why not the roof. We aren't allowed up there.

[Kyle Yankanich]

We're the top floor. Can we put a small extension out of our window to above roof level? Or am I reaching?

[Jonathan Simpson]

Reaching out the window

On Sep 5, 2012 1:51 PM, "Kyle Yankanich" <kyleya...@gmail.com> wrote:

We're the top floor. Can we put a small extension out of our window to above roof level? Or am I reaching?

--

[andrew sooy]

what if we said we where putting a garden on the roof, we could hide a antenna in it.

[andrew sooy]

is roof just off limits all together?

[Sean McBeth]

What makes you think we'd be allowed on the roof for something large and disruptive like a garden over a small, unobtrusive antenna?

[PJ Santoro]

Yeah, there is a gate with a lock on it going up to the roof. It's off limits.

sent by telephone

[Jordan Miller]

ham-tenna. duh.

http://www.ham-tenna.com/

jordan

[Dan Shookowsky]

Looks like someone created a garden variety cluster of Pi's

http://hardware.slashdot.org/story/12/09/12/145256/university-team-builds-lego-and-raspberry-pi-cluster 

Although as you mentioned, it isn't exactly a super computer, but it's also only drawing 13 amps.