cmcalibrate 0.21 win32/64 lin32/64

[yoyo]

cmcalibrate 0.21 increases now the maxsize for HMM banded DP. The last long running cmcalibrates aborted because the estimated size was larger than 2G. The maxsize is now 4GByte.

yoyo

[FalconFly]

Does that mean extremely long-running cmcalibrate 0.20 WorkUnits (like #157707) should be aborted ?
(right now it's still running on two Systems but will likely take at least one more month to complete despite being far beyond its deadline (!) )

[yoyo]

Does that mean extremely long-running cmcalibrate 0.20 WorkUnits (like #157707) should be aborted ?
(right now it's still running on two Systems but will likely take at least one more month to complete despite being far beyond its deadline (!) )

This one will probably fail with cmcalibrate 0.20. You should consider to abort it.
yoyo

[ocgbargas]

Hi good morning. First apologies for not knowing English. This is translated with google translator.I have a unit like this in my i7 920. Specifically is cmd_awin64 (rfam-9.1full). This unit takes 42 hours and places that still has 31.This is normal?? As waste is often a failure of process time.A greeting

[yoyo]

Hi good morning. First apologies for not knowing English. This is translated with google translator.I have a unit like this in my i7 920. Specifically is cmd_awin64 (rfam-9.1full). This unit takes 42 hours and places that still has 31.This is normal?? As waste is often a failure of process time.A greeting

Can you post the ID of this job or your host ID?
For some it is normal, so you should keep it running.
yoyo

[ocgbargas]

Hi good morning. First apologies for not knowing English. This is translated with google translator.I have a unit like this in my i7 920. Specifically is cmd_awin64 (rfam-9.1full). This unit takes 42 hours and places that still has 31.This is normal?? As waste is often a failure of process time.A greeting

Can you post the ID of this job or your host ID?
For some it is normal, so you should keep it running.
yoyo

The unit is this http://www.rnaworld.de/rnaworld/result. ... id=1288088 with this name cmc_aWin64 [Rfam-9.1-full] _RF00551_bicoid_3_Rfam-9.1_1267959122_840_1

[yoyo]

Yes let it run.
How much progress has it and how many hours is it running now?
yoyo

[ocgbargas]

Yes let it run.
How much progress has it and how many hours is it running now?
yoyo

Processed 51 hours. You have 3 hours. 95%.
You can pay at least 10,000 points but my anger will be very very very great.

This duration is normal?
I have another outstanding start saying that it took 25 hours.
If my team i7 920, 6 gb ram it takes, I do not imagine an older PC

A greeting.

[ocgbargas]

Every time I play something on time increases. Just do it with this project. Toys 1 hours last night and the remaining time rose to 90 hours. Now in the morning gets you is left go 63 and 34 hours elapsed.
This is not normal

[Michael H.W. Weber]

Do not worry - it is normal. The problem is that the progress bar dynamically adapts which is an integrated feature of BOINC. Moreover, we assign the WU duration on the basis of a set of mini simulation runs we perform for each (!) CMC WU (and CMS WU as well) on the server machine. These mini simulations just represent a tiny snapshot of what happens during full simulation that you perform on your machine. The code has stochastic elements and interconnected loops that do not allow for an exact run time prediction. It can only be an approximation. For small WUs this approximation is very good. The large the WU is (rising complexity) the less accurate is this prediction method. Experimentally, we determined that the predicted run time of very large CMC WUs might be only half of the "real life" computation duration. Please keep that in mind. Now, once such a misinterpretation took place, BOINC notes that and will increase the progress bar indicator dynamically for the next CMC WU that pours in. It assumes that the delay was caused e.g. by external influence such as parallel gaming. Then, this BOINC-derived progress bar adjustment might again fail if the next WU is e.g. a small one for which the original prediction made by our server usually is quite accurate (an dwas now mis-calibrated by BOINC itself). Do you get the complexity of the problem? The bottom line is: Just sit it out. If there is no hardware problem, lack of memory or so, the WU should complete.

Michael.

[ocgbargas]

Processed 75 hours. You have 95 hours. 26%

[ocgbargas]

Processed 85 hours. You have 116 hours. 23%. it never ends