This week the 3rd Pimoroni fan shim seized up. That is 3 out of 3 failed after a few months of running. I've done a temporary arrangement of suspending a Noctua 40MM fan over the top of the SoC and plugged it into the GPIO pins for power in order to get the Pi4 going.
BOINC project news
Last week the Einstein@home ran out of BRP4 work so the Pi3's all had
a holiday. The project have since provided new work. They normally have
plenty of work for the slower devices like the Pi.
Asteroids@home released new apps. This included a new 102.13 app for the Pi. I don't recommend you use it. The 4 work units I downloaded ran for over 35 hours. I've since removed the project from the Pi that I tried it on.
Rosetta@home also released a 1.20 app for all platforms. A long asked for improvement with this app is it unpacks its databases into the project directory rather than slot directory. This allows them to be shared with other tasks, saves a lot of writes to the SD card (the databases are up to 1.1GB) and reduces the number of copies to one. Previously it used to unpack them into the slot directory at the start of each task and then delete them when the task finished.
1 comment:
The IBM-sponsored "World Community Grid" on BOINC now has an "OpenPandemics" project... and to my great surprise, it does support ARM 32-bit for Linux. It also has very low hardware requiremens and a 7-day deadline. I have a RPi 1 with 512 MB RAM running 24/7 as a mini-homeserver. It manages to finish an OpenPandemics work unit in 5-6 days.
Curiously the CPU load is only at 2/3, and I don't see any I/O wait either. I'm not sure where it spends its time - maybe waiting for main memory? - but as long as it works...
Obviously the contribution is tiny, but given the RPi "runs anyways" and full load increases the temperature only by 1°C (according to "/opt/vc/bin/vcgencmd measure_temp"), I'm happy that also this machine can join in the fight.
Post a Comment