Fingers crossed that what I'm gonna upload will be accepted.
What has been marked as "abandoned" usually do not get accepted even if you have a valid result
So I havetwothree machines sitting idle now with full bunkers. Seems a waste, not sure what to do.
All three all running WCG CPU only. Two of the boxes can run GPU tasks, but one of them continuously gets computation errors on Einstein when running WCG, so I paused Einstein on it for a while. One box is still crunching away at some Einstein units, I think, but it's definitely done with what WCG it has. 9 more hours, I'll probably just sit tight because I'm working and can't do anything that will take more than a few minutes.CPU-only machines, or GPU too?
If CPU only, you could go through the motions of setting up multiple data directories as per the guide at OCN which I linked to in post #74, and then download more Cosmology@Home. (OpenZika is not feasible at this point in time, because that requires circa 3 hours until they let you download into a new client without limit.)
If GPU, you could arrange to block uploads to C@H and WCG, e.g. with the host file in post #93, and then start bunkering Einstein@Home. (Is a 5-day bunker for E@H feasible?)
Edit:
Oops, my 6950X with mostly C@H is now idle too. I look into that 2nd option now.
#!/bin/bash
while true ; do
boinccmd --get_tasks | grep -B12 -A4 EXECUTING | grep ' name:\|checkpoint' | grep -B1 'time: [89][0-9][0-9]' | grep 'name:.*eah' | sed -e 's/^ *name: //' | xargs -n1 -I{} boinccmd --task http://einstein.phys.uwm.edu/ {} suspend
sleep 30
doneI thought every race had a defined starting time. Can you explain in more detail?Oh, one more warning about flushing: Don't flush until you see thewhites of their eyesstats on their site! They won't count work flushed before then.
Every race has a time when SETI.Germany collects starting stats from a site. Once they've done that, they'll post starting stats here and you can flush. Sometimes a site isn't accessible for awhile, like one last year. If they don't post stats immediately, hold your flush until they post stats.I thought every race had a defined starting time. Can you explain in more detail?
To allow me to get more Einstein at that point, I wrote a script, in BASH, for Linux, that will suspend Einstein work before it finishes. I can then resume everything, get more work, and maybe suspend most of them again.
I don't understand. Did the Einstein work fail to resume? Or did my script prevent it from progressing? (It should do that; you should stop the script before dumping results.)This left me with two tasks getting stuck, blocking one of the cards, and I had to abort them.
Oh, that's definitely too long. Typically I was seeing a minute or two before resumption of GPU usage, mainly on machines with slower CPUs.Thanks @crashtech, I keep that in mind for the next occasion. I aborted those two tasks at a runtime of 26 and 28 minutes though, while they should have finished at 12 minutes.
[Cross Country, Einstein@Home]
1100 tasks downloaded, then its "Not requesting tasks: too many runnable tasks" time again. (AFAIU it is a client-side limit of 1000 tasks which was overshot by the last download of >100 tasks.)
Triple GPU machine, running 6 tasks at a time, 12...15 minutes per task: That's merely a bunker for ~1.6 days. Compared to WCG, it seems trivial though to set up multiple clients for Einstein.