My main computer seems not to be reporting its work rate correctly.
I had been running work stored on an external drive to a laptop which I then migrated to my main computer after its last two jobs had been completed. I moved both the work to do, pxxxxx_xxxxxxxx.xxx and pGxxxxxx files and started to run the files on Oct. 9. Test "n" and progress on the pending test management page correspond. However, since my last two jobs had been submitted on Oct. 9, the work rate seems to be not reporting correctly on the user stats page.
What could be the cause of this problem? Was it an untested computer I was initially running it on and thus requires verification before work done is reported?
I guess first check the obvious... Look in prime.txt and make sure DaysBetweenCheckins is set to 1 and there is no line ManualComm=0 in prime.txt.
You can try this... (Although, I have done what you describe several times and have not noticed any problems with reporting, and I don't have this set in prime.txt...)
Originally Posted by undoc.txt
I have checked prime.txt and it is set to the correct value. I have added the fixed hardware statement with no changes resulting (there was no such statement originally). This is quite an odd problem. I have gone so far as to overwrite prime.txt with the original file and I'm at the point where I may just delete prime.log and local.txt. I'm falling asleep trying to fix this. Will try again tomorrow with a fresh start. Better to be lucid when taking chances. Thanks for the suggestion. Any other recommendations?
Night.
I had gone through some of the files in my Prime95 folder and noticed some inconsistencies.
In my results.txt file:
[Sat Oct 17 03:05:42 2009]
Iteration 4956612 / 16862242
Iteration 5101398 / 16862316
This corresponds with my Prime95 output window approximately 25 hours later:
[Worker #1 Oct 18 04:21] Iteration: 6140000 / 16862242 [36.41%]. Per iteration time: 0.064 sec.
[Worker #2 Oct 18 04:24] Iteration: 6360000 / 16862316 [37.71%]. Per iteration time: 0.053 sec.
So far both match for n and work done.
However, my current pending tests, according to my Pending Tests Management page on the SeventeenorBust site for my main computer are:
1004949 21181•2^16862228+1 65.95.140.3 Mon Sep 21 19:20:27 2009 Sun Oct 18 06:00:57 2009 6003493 35 % 747495
1004950 22699•2^16862302+1 65.95.140.3 Mon Sep 21 19:20:27 2009 Sun Oct 18 06:00:58 2009 6182782 36 % 769820
The progress corresponds with, as a percentage, my pending tests, however, for the wrong "n". In my Prime95 directory, I have:
p21181_16862228
p21181_16862228.bu
p21181_16862228.bu2
p22699_16862302
p22699_16862302.bu
p22699_16862302.bu2
pG017598
pG017651
Something is wrong here.
I hope I've detailed the symptoms without causing too much confusion. I will reboot into Windows and see if the above is true.
Regards,
Dan
Pursuant to my previous message whose output was using mprime on Ubuntu 9.10 Beta, using Windows XP Professional x64 edition differs only in the output of the Prime95 window:
[Main thread Oct 18 05:08] Mersenne number primality test program version 25.11
[Main thread Oct 18 05:08] Starting workers.
[Worker #1 Oct 18 05:08] Worker starting
[Worker #1 Oct 18 05:08] Setting affinity to run worker on logical CPU #0
[Worker #2 Oct 18 05:08] Worker starting
[Worker #2 Oct 18 05:08] Setting affinity to run worker on logical CPU #1
[Worker #2 Oct 18 05:08] Setting affinity to run helper thread 1 on any logical CPU.
[Worker #1 Oct 18 05:08] Setting affinity to run helper thread 1 on any logical CPU.
[Worker #1 Oct 18 05:08] Resuming PRP test of 21181*2^16862228+1 using all-complex FFT length 1536K, 2 threads
[Worker #2 Oct 18 05:08] Resuming PRP test of 22699*2^16862302+1 using all-complex FFT length 1536K, 2 threads
[Worker #1 Oct 18 05:08] Iteration: 6180660 / 16862242 [36.65%].
[Worker #2 Oct 18 05:08] Iteration: 6406084 / 16862316 [37.99%].
[Worker #1 Oct 18 05:11] Iteration: 6184000 / 16862242 [36.67%]. Per iteration time: 0.052 sec.
[Worker #2 Oct 18 05:11] Iteration: 6408000 / 16862316 [38.00%]. Per iteration time: 0.095 sec.
Prime95 seems to acknowledge the existence of the work files, however proceeds to work on two other different tests.
Something seems wrong. Rather than force the expiry of these two tests, delete the work files and start over, perhaps this is a bug in the program that needs further inquiry. I've migrated work from other machines prior without error. This seems to be the first time I've seen such an occurrence. I suspect this has something to do with my current work rate not being reported as I had mentioned at the start of this thread.
Any suggestions?
Last edited by imwithid; 10-18-2009 at 05:24 AM. Reason: Grammar
Maybe if you just finish the tests you are running now prime95 will start by doing the "waiting" tests. It seems to pull in new tests at some time before it is finished with the ones being worked on at the moment.
I think I'll just wait the course and see. I'll stop one of the workers to allow the second to use both cores to accelerate its completion.
I think I'm wrong on this supposed problem. I have checked another computer I've been running and it seems consistent with this one. Nothing I can find can explain why my current work rate is not updated from this computer.
My assumption was that the n does not match the number of iterations but it seems that it is always off by 14-15. I guess I've never really noticed this before but I assume that it is required for error correction or some other purpose.
I'll continue to investigate.
You are correct in that the number of iterations does not match the 'n' value - this is normal. (I've been tripped up by this before myself)
So, just so I make sure I have this correct - your Pending Tests page shows the correct completion percentage and is being updated as normal, but your stats page is not being updated with just those two tests (i.e. other tests are being shown in your stats)
Have you tried making a brand new prime95 folder and get a brand new test and see if that shows up on your stats correctly. If so, you might just cut your losses and cancel the other tests, unless Louie or someone is able to figure it out soon on the server end. Sounds like it is probably something on that end, since your progress is being shown correctly on your pending tests page...
That is correct. One other oddity I've noticed is that my Overall Work Rate has been increasing as I have left this computer on 24/day to see what changes would occur, however, my Daily Work Rate and both Daily and Overall Work Done have not been changing in tandem.
I will try as you have stated to start a new folder and run a brand new set of tests.
This has been very bizarre. I really hope it is an isolated problem on my end.
The problem seems to have resolved itself after I received new tests. I ran the previous tests to completion in case this helps discover a potential problem.
FYI:
Adding FixedHardwareUID=1 did not fix the problem (these files were transferred from an external hard drive to a desktops hard drive.
Over writing both prime95.exe and mprime as well as customization files did not fix this either.
All suggestions and help were appreciated greatly.
Regards,
Dan.
It was the IP address all along. This is what throws the daily stats off. If work is started on a given IP address and then taken to a second computer at a different location where the work is continued to completion, the daily stats don't update properly. I have a fourth computer operating which started at my home location and is now operating elsewhere and the daily stats are running as though it's not there (and will not for probably a month or two). As soon as its current job is complete, I think I should see a minor increase in the daily work rate.
All I wanted was to know why. It was driving me mad.
Are you sure? What about all of the people with dynamic IP's (I know my PPPOE connection at home changes almost daily, but I haven't run SoB there for a while now) - wouldn't they have problems as well? (Unless the change of IP needs to be outside the ISP's subnet or something for it to cause problems) Perhaps it's a combination of a new IP with a different hardware ID?
The IP prefix for both locations is different (one is 99 and the other 64). I've transferred tests between computers at home without problems (I have a static IP and the other locations are dynamic). It is when I transfer between computers at different locations that I have had this happen. I will look further into this but it will take some time as I am not at the other locations very often.
Posting Permissions
Reply With Quote