I know its not massive computing power, but it is some. I have a spare 4GB Raspberry Pi sitting around doing nothing. I also have a 1GB Model 3+ powered on 24/7 but doing almost nothing. I saw a tweet that said that Rosetta@Home was now supported on these devices, but I can find nothing that tells me how to set them up (currently running raspbian 32 bit - but that could change).

Can someone point me at some simple instructions of where to get the software, and how to set it up on the two Pis.

Thanks

In one of the earlier threads in this forum.
Rosetta running on ARM platforms

---

Grant
Darwin NT

In one of the earlier threads in this forum.
Rosetta running on ARM platforms

Perfect thanks - I now have my spare 4G Pi running a number of tasks. Sadly the 2GB Pi 4 just hasn't quite got enough memory and the 3+ at 1GB definitely doesn't

Then please sign up for World Community Grid on the smaller device.
They've got COVID-19 tasks coming soon for Windows, Mac and Linux and need their queues drained of dry of everything else ASAP.
Also their system requirements are usually much smaller.

Then please sign up for World Community Grid on the smaller device.
They've got COVID-19 tasks coming soon for Windows, Mac and Linux and need their queues drained of dry of everything else ASAP.
Also their system requirements are usually much smaller.

I tried the same approach as with the other Raspberry PI to get alt_platform set up, but World Community Grid reports it doesn't support computers of type arm-unknown-linux-gnueabihf

Is there some other approach I should take?

I tried the same approach as with the other Raspberry PI to get alt_platform set up, but World Community Grid reports it doesn't support computers of type arm-unknown-linux-gnueabihf

Is there some other approach I should take?

That sounds like it didn’t take the alt_platform tag. What BOINC version and please list your cc_config.xml

---

MarksRpiCluster

That sounds like it didn’t take the alt_platform tag. What BOINC version and please list your cc_config.xml

I don't know why - i've wiped that machine for now, but it had the config set up exactly as as specified in Marks Rpi Cluster article and on my other pi.

It also did contact rosetta@home fine - just told there wasn't enough memory.

<!--
This is a minimal configuration file cc_config.xml of the BOINC core client.
For a complete list of all available options and logging flags and their
meaning see: https://boinc.berkeley.edu/wiki/client_configuration
-->
<cc_config>
  <log_flags>
    <task>1</task>
    <file_xfer>1</file_xfer>
    <sched_ops>1</sched_ops>
  </log_flags>
  <options>
    <alt_platform>aarch64-unknown-linux-gnu</alt_platform>
  </options>
</cc_config>

Its running boinc-client from the debian stable repository 7.14.2

Well that looks correct to me. Maybe the WCG server doesn't support it if they're running an older version. I'd report it over at WCG and see what they say.

The BOINC client configuration page doesn't make any mention of what version it applied from, All it says is:
<alt_platform>platform_name</alt_platform>
Specify an alternate platform name, to be included in scheduler requests. You can specify more than one.

---

MarksRpiCluster

I thought I'd post here what settings I am using on the couple of Pi4 model B 4GB that I have and what seems to work for me.

I am running 3 tasks at a time. I have seen memory use up to 918MB at times per task so I think that is about the limit of how many can run at once. I use an app_config file to limit the number running at the same time. If you have a 2GB Pi4 you should be able to run one at a time.

I had to allow 6GB of disk space in BOINC/project preferences to get all three running. It would appear to need around 2GB of disk space per task. Make sure you have a big enough SD card. I have also set the checkpoint interval to 600 seconds to reduce wear on the SD card.

This is for the 4.12 app, things might change with a new app version,

I have memory settings set to 60% when in use and 90% when not in use. All my Pi's are running headless (no screen or keyboard) so as far as BOINC is concerned they are never "in use".

I have a bunch of Pi3's but they are unsuitable for this project due to their limited memory. I have them running Einstein work.

---

MarksRpiCluster

Hi!
I would like to help with my Raspberry Pi 4 but after adding the project I get the a message saying that the project doesn't support arm-unknown-linux-gnueabihf.

Is there something that I can do?

Thank you!

In one of the earlier threads in this forum.
Rosetta running on ARM platforms

Perfect thanks - I now have my spare 4G Pi running a number of tasks. Sadly the 2GB Pi 4 just hasn't quite got enough memory and the 3+ at 1GB definitely doesn't

Good news - I just set my 2GB Pi4 up and left it, and just looked at it now and its running a job. There are several not started for lack of memory, but I guess it will attack another one of them when the current one finishes.

Good news - I just set my 2GB Pi4 up and left it, and just looked at it now and its running a job. There are several not started for lack of memory, but I guess it will attack another one of them when the current one finishes.

In your Account, Computinmg preferences,

Memory
When computer is in use, use at most     95 %
When computer is not in use, use at most 95 %

will give BOINC more RAM and should allow 2 Tasks to run at a time.

In

Other
           Store at least 1    days of work
Store up to an additional 0.02 days of work

would be good options to avoid ending up with more work than you can process before the deadlines while the Estimated completion times settle down.

---

Grant
Darwin NT

Good news - I just set my 2GB Pi4 up and left it, and just looked at it now and its running a job. There are several not started for lack of memory, but I guess it will attack another one of them when the current one finishes.

In your Account, Computinmg preferences,

Memory
When computer is in use, use at most     95 %
When computer is not in use, use at most 95 %

will give BOINC more RAM and should allow 2 Tasks to run at a time.

In

Other
           Store at least 1    days of work
Store up to an additional 0.02 days of work

would be good options to avoid ending up with more work than you can process before the deadlines while the Estimated completion times settle down.

Actually I was already more aggressive that that

Memory
When computer is in use, use at most     100 %
When computer is not in use, use at most 100 %

Other
           Store at least 0.1    days of work
Store up to an additional 0.5 days of work

Its a headless server running a single nodejs application under PM2 and an ssh deamon so I can reach it. Its doing nothing most of the time.

Actually I was already more aggressive that that

Memory
When computer is in use, use at most     100 %
When computer is not in use, use at most 100 %

Other
           Store at least 0.1    days of work
Store up to an additional 0.5 days of work

Its a headless server running a single nodejs application under PM2 and an ssh deamon so I can reach it. Its doing nothing most of the time.

Things get ugly when a machine runs out of memory. I'd limit to 95%. There are moves to try and improve things in the Linux kernel when it runs out of memory but they aren't there yet.

---

BOINC blog

Unfortunately Rosetta doesn't speaking about ARM64 issues like on RPi4 which is fighting overheating problems from the beginning without any load on CPU so it has a big chance Rosetta app will kill that boards shortly. Be careful if you would take care about your RPi4.

Other
           Store at least 0.1    days of work
Store up to an additional 0.5 days of work

Its a headless server running a single nodejs application under PM2 and an ssh deamon so I can reach it. Its doing nothing most of the time.

Generally you're better off with those values reversed (if you actually want 0.5 days of work; otherwise set it to 0.1 and 0.02 to get 0.1 days worth (although you'd have to reduce the Target CPU runtime as well).
That way you tend to end up closer to your "Store at least xx days of work" setting.

---

Grant
Darwin NT

Unfortunately Rosetta doesn't speaking about ARM64 issues like on RPi4 which is fighting overheating problems from the beginning without any load on CPU so it has a big chance Rosetta app will kill that boards shortly. Be careful if you would take care about your RPi4.

RPi4's aren't the only ARM64 devices on the market. Many mobile phones as well as other single-board computers run ARM64, including powerful compute SoM's like Nvidia Jetsons, which do come with large heatsinks. The issue of overheating is specifically a RPi4 problem, not an ARM64 problem.

I'd hope RPi4 owners would be aware of CPU overheating issues, since they were prevalent long before this, but the advice holds true for any compute device that doesn't have the proper thermal controls to dissipate heat.

As a final point, it's stated here under Liability: [https://boinc.bakerlab.org/info.php]

Rosetta@home and University of Washington assume no liability for damage to your computer, loss of data, or any other event or condition that may occur as a result of participating in Rosetta@home.

Most if not all BOINC projects (along with virtually all software nowadays) carry the same liability for much of the same reasons: avoiding issues from users that run their programs and don't ensure proper thermal controls that damage their hardware.

It's not the project if your hardware overheats. If RPi4 has overheating problems yeah well, time to get a better heatsink!

---

If RPi4 has overheating problems yeah well, time to get a better heatsink!

Pimoroni is the solution

---

Yes they making good heatsinks. BTW I tried ARM64 on ArchlinuxARM but not get WU-s on my RPi3B and Odroid-C2. I wouldn't try on RPI4 because of stock heatsink...