Hi,
You may have noticed that we didn't produce new work units over the past few days. This was due to an infrastructure upgrade in our data center. While we made sure that our production project wasn't affected, our test project ran out of work. Oh well, it's a test project after all.
Anyway, we began producing work again, at least for the binary radio pulsar search, so you should be able to fetch new work units soon.
FYI, we're also going to update the BRP apps themselves over the next days. They include improvements that should make them more robust against external termination, in particular the GPU apps.
Thanks for your patience!
Oliver
Copyright © 2024 Einstein@Home. All rights reserved.
Comments
Work availability
)
Update: new CUDA and OpenCL applications v1.31 released (see above for details).
Cheers,
Oliver
RE: Update: new CUDA and
)
thx, three of them arrived on my PC suggesting 20 h,
but in real they will be done >1h i think
cu, hf, stephan
[16:47:35][10121][INFO ]
)
[16:47:35][10121][INFO ] Checkpoint committed!
[16:47:50][10121][ERROR] Error during CUDA device->host time series length transfer (error: 700)
[16:47:50][10121][ERROR] Demodulation failed (error: 1008)!
16:47:50 (10121): called boinc_finish
grmml, whats happend there?
This means an error occurred
)
This means an error occurred during GPU operations. You seem to be the only one with this particular error so it's not a general issue. Could very well be a driver issue...
Oliver
RE: thx, three of them
)
Yes, that's an effect of BOINC's new runtime estimation scheme we test here at Albert@Home. BOINC needs to gather statistics each time we release a new app version until the numbers converge on the actual/real/correct value...
Sorry,
Oliver
ok, thinking the wu is died
)
ok, thinking the wu is died by a lost network connection, grmml, shit happend..
it's still good man!
)
it's still good man!