I am currently using a:
2 vCPU (1 shared; e2-medium, which costs enough to where it can run for an entire year and only then use up the $300 credit)
1 vCPU (1 shared; f1-micro is free for 744 hours/month, which is a month)
and Finally, a Windows Server image on AWS. Honestly, it gets a lot of time in but the Server image is on a shared CPU as well, making it VERY slow (17 hours per task? but ok). It's free for a year though, so I got it set up doing work for me :)

Am I missing something obvious? Following Falconet's instructions I'm getting the following error message at step 2: Unable to locate package boinc-client

And we're up and running. Thank you Falconet and cwierzbicki!


EDIT: @Falconet--For the preemptive instance, does boinc resume from the last checkpoint when you restart or does it lose all progress on tasks that are running when it is preempted?

EDIT 2: I've now had the VM preempted multiple times. It does appear to return to the most recent checkpoint when restarted.

gibbcorp, have you checked your device profile?

How does the "processor usage" section look like? Maybe something there is disabling computing until it becomes idle?

Nice gibbcorp. Glad to hear it's working now.

Supdood, yeah it does save everything, it's basically a normal shutdown.
I have days where there isn't a single shutdown other than the one at the 24 hour mark.
Today, for example, I woke up and 1 of the two VMs wasn't running, having shut down about 6 hours earlier at around 2 in the morning.

Either way, they do great work. The oldest VM I have right now was created on May 6 and has 1 year and 26 days of runtime with 3018 results returned as of the last update. Pretty good. Edit: Some very quick math says it's contributed about 87% of what it could have were it not for its pre-emptive status. (391 days contributed out of 448 total over a period of 28 days with 16 cores)