The computer shows as 'Anonymous' because I have selected not to show my computers. If this computer needs some TLC, so do the other 12 I'm running since they all seem to have the same problem. By the way, I monitor power and temps on the computers...no anomalies. I should just write this off as 8% of every work unit I get is expected to be a waste of time?

---

I just noticed another batch.

19ap11ad.19368.8629.140733193388035.12.96_6
19ap11ad.18067.3312.140733193388039.12.154_8
19ap11ad.22063.6175.140733193388041.12.8_5
19ap11ad.24546.4948.140733193388042.12.138_3
19ap11ad.25864.9856.140733193388043.12.33_7
19ap11ad.23016.13537.140733193388045.12.221_9
19ap11ad.23606.49752.140733193388046.12.197_4


Cheers.

And here's another. 01jl12ad.5031.395923.140733193388035.12.195 Aborted when I noticed it.

---

BTW, Beta has been serving up a lot of bad ones dated 26jn11 lately. The few I looked at today were all 26jn11af, but I can't say if all of them are af, or if they're only a portion of af.

---

David
Sitting on my butt while others boldly go,
Waiting for a message from a small furry creature from Alpha Centauri.

BTW, Beta has been serving up a lot of bad ones dated 26jn11 lately. The few I looked at today were all 26jn11af, but I can't say if all of them are af, or if they're only a portion of af.

All SaH v7 tasks split at Beta from 8 Jan 2015, 6:34:47 UTC to 23 Jan 2015, 6:02:30 UTC were from the 26jn11af tape file. 4 channels were split, from the beginning of beam 4 polarity 0 to the end of beam 5 polarity 1.

I also noticed tasks of the kind with bad autocorr parameters similar to this thread, as well as some with good autocorr parameters but triplet thresholds so low that they caused immediate overflow. It will be interesting to watch tasks split from the 27my12ac tape file (splitting started 23 Jan 2015, 6:26:53 UTC).

The splitters have a caching mechanism for analysis_cfg, and there's a keyuniq value which identifies which cached cfg was used. The value is formed from 1024 times an ID value which should apparently come from the app record, plus the AR times 100 reduced to an integer so the AR range from 0 through 10.0 gives 1001 possibilities (any AR above 10 uses the slot for 10). IOW the problems with analysis_cfg sections may be AR related in 0.01 degree increments, plus it looked like the splitters at Beta were not consistently getting the right ID value. The keyuniq values varied a lot.

                                                                  Joe