I've been getting strange work units from Beta on both of my crunchers. The units crunch for five seconds; then, stop stating "Postponed...memory error." I end up having to abort them; because they won't crunch.

They've ranged from CUDA23 to CUDA42 WUs; and just won't crunch. Anyone else having these issues??? (Problem started yesterday, 1-7-2015.)


TL

---

TimeLord04
Have TARDIS, will travel...
Come along K-9!
Join Calm Chaos

See the WUs with very unusual Autocorr parameters thread.

Claggy

I've just noticed that I have a small group of WU that are reporting the same problem.
They are in a group from 10fe09ab4848 & 10fe09ab1847, which doesn't appear to be the same group as those being reported on main.
They run for a few seconds then stall with the status message "Postponed: Cuda runtime, memory related failure, threadsafe temporary Exit". All are assigned to run as cuda42 tasks.
They run for typically 3 seconds, stop and sit around for a bit, then restart from 0 seconds elapsed, ~00:18:50 remaining.

I've just noticed that I have a small group of WU that are reporting the same problem.
They are in a group from 10fe09ab4848 & 10fe09ab1847, which doesn't appear to be the same group as those being reported on main.
They run for a few seconds then stall with the status message "Postponed: Cuda runtime, memory related failure, threadsafe temporary Exit". All are assigned to run as cuda42 tasks.
They run for typically 3 seconds, stop and sit around for a bit, then restart from 0 seconds elapsed, ~00:18:50 remaining.

Check their stderr.txt's, you'll find them in the slots assigned to those Wu's, it'll probably say:

Error on launch (ac_reducePartial<<<grid3, block3,blksize*sizeof(float3)>>>( (float *)dev_AutoCorrIn, dev_ac_partials )), file c:/[Projects]/__Sources/sah_v7_opt/Xbranch/client/cuda/cudaAcc_autocorr.cu, line 200: invalid configuration argument

Claggy