Hello everyone. I was a BOINC user several years ago, but haven't contribute for recent years, and I have to register a new account to post this...
I want to know whether BOINC will run properly on this device: http://www.nvidia.co.uk/object/jetson-tk1-embedded-dev-kit-uk.html
I found this recently, it have an ARM CPU and a CUDA enabled GPU, and it just make me think about BOINC. I think it might be a great dedicated device to run BOINC 24 hrs a day, with low power consummation, low heat, no noise, low cost, and HIGH speed/efficiency by the 192 CUDA cores, just wonder will BOINC run on it?
If it can do this, I might buy one of this sometimes in the future and leave it running just like a router, thus restart my BOINC life. Please give me some advices. Thank you everyone.

There is a BOINC for Android for ARM CPUs;
There is a BOINC for the Raspberry Pi, running on its Linux, in its repositories.

So perhaps that the Linux for that system has a BOINC.
But then you have the next problem to tackle, as BOINC is only part of the problem. Next you'll have to find a project that made a science application that can run on the CPU of the Jetson. Nothing said about the GPU.

Perhaps the best place at the moment to ask is at the Einstein forums, where they have a long thread on support for systems like that.

---

In the mean time, why not get your PCs and/or Android phone going?

---

My Detailed BOINC Stats

Thank you. I know that there's BOINC for Linux, and most project also support Linux, I checked the datasheet and found that Jetson's GPU have OpenCL support which is needed for GPU computing. I just want to confirm will BOINC projects really run on a ARM CPU + Nvidia GPU + Linux?

My computer is in use most of the time when turning on, and I don't have an Android phone.

I know that there's BOINC for Linux, and most project also support Linux

Yes, but the problem here is the CPU. According to the small lettering on the link you gave about the Jetson it supports Linux for Tegra. It doesn't support just any Linux though, as most Linux distros are for x86 and x64 CPUs only, not for ARM CPUs.
So you need that special Linux, developed by Nvidia that is able to run on that CPU, and in its repository should you look if someone ported a BOINC that can run on that system. Berkeley certainly doesn't have a BOINC that is capable of running on an ARM CPU, not unless you also have the Android Operating System and a mobile device such as a smartphone, or a tablet.

And even when you then have managed to get a BOINC running on that system, you're only half-way there as there may not be any projects that have a Linux_ARM application. Some have ARM applications, but again, those are for Android. And even then, one ARM application cannot run on all ARM CPU out there.

So you'll have to ask the specific projects if they have an application that can run on an ARM Cortex-A15 CPU, under the special Linux developed by Nvidia. Yet, again, only after you managed to get a BOINC running on that system.

As for OpenCL applications being able to run on that GPU, even that is difficult as Nvidia is no longer putting anything into developing for OpenCL, but only gearing towards their own CUDA.

---

Thank you very much, Umm, it's much more complex than I thought, maybe my plan will fail...

I can give instructions on compiling Boinc for Linux, the Software prerequisites required as I've compiled Boinc for the Raspberry Pi running Raspbian Linux, and the Parallella running Linaro Ubuntu,
what you might have difficulty is in the GPU detection,

Einstein and Albert at home both have Arm Linux applications in Neon and non-Neon versions, they should work without too much problem.

Claggy

Thank you for your help. What do you mean by "compiling"? Will I need programming skills? I have little knowledge about programming or Linux...
I first thought BOINC will run on Linux easily because many people say Linux have a very good compatibility over different platforms, but it seems BOINC will be different on different CPUs, even they are all ARM based? I have a thought to use this Jetson board is because it have a powerful GPU but ARM-based CPU, so I think it will be very efficient.

If you're looking for a low-power crunching machine and don't want to get into compiling apps and the like, you might be better off with an Intel Atom-based solution. They are x86-64 based processors and can run Windows or Linux and use the regular desktop apps already available for those platforms.

---

My Detailed BOINC Stats

Thank you, but I don't think atom is good for BOINC, because the CPU speed is very low, but I just think is it ok if I put a graphic card to it.

Thank you, but I don't think atom is good for BOINC, because the CPU speed is very low, but I just think is it ok if I put a graphic card to it.

My Atom N450 is almost as fast as my fastest ARM device, and a lot faster than my other ARM devices:

All Binary Radio Pulsar Search tasks for computer 10847 Atom N450 @1.66GHz

All Binary Radio Pulsar Search tasks for computer 9756 HTC One S ARM/Qualcomm Snapdragon S4 @1.51GHz

All Binary Radio Pulsar Search tasks for computer 8807 2012 Nexus 7 ARM/Nvidia Tegra @1.20GHz

All Binary Radio Pulsar Search tasks for computer 11206 Parallella ARM/Zynq-7020 ARM A9 @700MHz (I think)

Claggy

Good to know, Claggy.
Have you got any informations about average power consumption?

---

My reasons to participate in grid computing:
1. I'm using true renewable energy (German accreditation; no certificate purchasing)
2. Production and disposal (will) amount to about 95% of the energy "used" during the lifetime of my PC
3. Helping

Good to know, Claggy.
Have you got any informations about average power consumption?

Digging out my wattmeter, my HTC One S is pulling around 3 Watts, the 2012 Nexus 7 the same,

Claggy

Is that on full load?
I'm guessing 30 watts for the N450 (Is that an Asus EEE?).

Still no reason to buy a new device - since most energy is "used" in production.
Sam, why not just run BOINC on your current device? You can throttle/autosuspend work as you like. (If you insist on a new device I would go with an ARM quad-core with Android.^^)

---

My reasons to participate in grid computing:
1. I'm using true renewable energy (German accreditation; no certificate purchasing)
2. Production and disposal (will) amount to about 95% of the energy "used" during the lifetime of my PC
3. Helping

Thank you everyone, I just think another way, I have a Pentium 4 computer not in use, maybe I can put a Tesla card on it, that will be very efficiency. Any ideas?

Thank you everyone, I just think another way, I have a Pentium 4 computer not in use, maybe I can put a Tesla card on it, that will be very efficiency. Any ideas?

Pentuim 4's are slow, and very inefficient compared to modern desktop CPUs.

Claggy

It also depends on what kind of Pentium 4 you have, and whether or not you have a PCI-E slot on the motherboard. If a Prescott P4, it might be able to keep up. But if there is no PCI-E slot on the motherboard, there's no way to add a Tesla.

(Before anyone says PCI -> PCI-E riser card, I'll say really?)

---

Pentuim 4's are slow, and very inefficient compared to modern desktop CPUs.

Claggy

It's more inefficient to put a working device out of use.

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
EDIT: Sam, maybe you should ask your favorite project's leader what s/he wishes you to use/buy.^^

---

My reasons to participate in grid computing:
1. I'm using true renewable energy (German accreditation; no certificate purchasing)
2. Production and disposal (will) amount to about 95% of the energy "used" during the lifetime of my PC
3. Helping

I knew, I won't use the CPU to do the project, only GPU.