I was just thinking do people use raspberry pi cluster computer to crunch?

---

https://boincstats.com/signature/0/user/415747/project/sig.png

It is just an ARM CPU. So if it were running Android it probably could run BOINC & SETI@home.
However, I'm not sure if anyone has it running under the default OS.
http://setiathome.berkeley.edu/forum_thread.php?id=67105
http://setiathome.berkeley.edu/forum_thread.php?id=70697
http://setiathome.berkeley.edu/forum_thread.php?id=73959

As far as clustering goes BOINC used shared memory. Which means an instance would have to be run on each separate node of the cluster.

---

SETI@home classic workunits: 93,865 CPU time: 863,447 hours
Join the [url=http://tinyurl.com/8y46zvu]BP6/VP6 User Group[

The Einstein@Home project has native Support for Raspberry Pi

It also has a very well informed thread on Parallella, Raspberry Pi, FPGA & All That Stuff

You'll find the basic stuff, like getting BOINC running on the device, covered there. Once you have that dealt with, porting the SETI app becomes much easier.

But as Hal says, each node will have to run its own instance of BOINC - which is how supercomputer clusters do it, too.

For $ 35 I am sorely tempted. What's holding me back is the micro in those guys is a Broodcom BCM2835 which is dual core. If it were quad core I'd pounce.

I have tried to crunch SETI on a RaspberryPI some time ago. I must say that I have not followed potential recent developments, but at the time performance was dismaying.
I reached a RAC of... 37.

No zeroes neglected. 37E0

Therefore I stopped.

That was under Debian Wheezy distribution.

Android is unlikely to reach The PI.

But recent events may change (or may have changed already) the picture.
Broadcom has openly released the libraries for the CPU/GPU driving the PI.

People are working on porting some parts of the OS to speed-up some functions in general. Some people are already speculating about the possibility of using the GPU section to crunch faster as we are already with the Nvidia/AMD/Intel GPUs.

Time will tell. I am obviously very much interested.
If something already happened I wil be happy to join the RasPI SETI crunching badwagon again (I have one of the first 10.000 RaspPIs sold in the first production batch, therefore it is also a question of keeping high this flag).

SETI apart, I find this device amazing (a lot of power and potentialities in little space and almost no power use) and being a tremendous incentive for application developing and an incentive for even more powerful and cheaper similar devices to appear.

Sleepy

I reached a RAC of... 37.

This is the problem isn't it.

Even if you lined up 100 of them you might match a 2nd hand pentium dual with a $50 CUDA card. You have spent ~$100, it will use less power, and be much easier to set up. Don't forget the 4 network switches you need to string them all together, and all that cable.

Now people are saying, yeah but the next ones will have more cores, and a usable GPU etc. But the technology that allows that will also move to full spec CPUs and GPUs as well. Lower power, more cores etc.

Now it would be impressive to see someone build it, but as far as "practical" goes?

That's not the same as saying you shouldn't experiment with them, but really there are better uses than trying to use them for raw processing power, they just don't have much...

Ian

I reached a RAC of... 37.

This is the problem isn't it.

Even if you lined up 100 of them you might match a 2nd hand pentium dual with a $50 CUDA card. You have spent ~$100, it will use less power, and be much easier to set up. Don't forget the 4 network switches you need to string them all together, and all that cable.

Now people are saying, yeah but the next ones will have more cores, and a usable GPU etc. But the technology that allows that will also move to full spec CPUs and GPUs as well. Lower power, more cores etc.

Now it would be impressive to see someone build it, but as far as "practical" goes?

That's not the same as saying you shouldn't experiment with them, but really there are better uses than trying to use them for raw processing power, they just don't have much...

Ian

I am running an Apple G5 power PC that I purchased in 2006 and about the best I can do with it is a RAC of around 370 running both cores. I am strongly tempted to upgrade to something newer but some of the applications I use will go away. SETI may force the issue if they drop Power PC support.

It is incredible how much processor power has gone up over a few years and how much we now expect.

---