Posts by Ananas |
|
1)
Message boards :
Problems :
Dear Alyssa Milburn - Forum moderator - Help!
(Message 12974)
Posted 173 days ago by Ananas I am NOT an admin anywhere but I THINK he can go to any account on any board and see his "cross project id" under 'my account' on the webpage and use that to login with instead of his username and password. The CPID is public, it would sure not be such a good idea to allow this to login. What you probably mean is the account ID (and I sometimes had to figure it out like this) : Even if you're logged in here with a email+password, the project creates a cookie (if your browser setting allows it to) with the account ID, cookie name is "auth" If the cookie hasn't timed out yet, Byron might still have it in his browser. |
|
2)
Message boards :
Problems :
Damaged WU series
(Message 11278)
Posted 1063 days ago by Ananas Sorry, I just see that I posted the same WU two times, one was supposed to be this one instead : abc_sieve_wu_01833896 |
|
3)
Message boards :
Problems :
Damaged WU series
(Message 11271)
Posted 1065 days ago by Ananas abc_sieve_wu_01830571 abc_sieve_wu_01830571 [17:37:39 ABC] can't open abc_sieve_state to read header data
[17:37:39 ABC] starting from scratch ERROR: Invalid parameter detected in function (null). File: (null) Line: 0 ERROR: Expression: (null) Unhandled Exception Detected... - Unhandled Exception Record - Reason: Breakpoint Encountered (0x80000003) at address 0x0000000077EF2AB0 Engaging BOINC Windows Runtime Debugger... |
|
4)
Message boards :
Problems :
Strange in-compatability AQUA / ABC
(Message 10943)
Posted 1177 days ago by Ananas Probably a cache size problem (ABC has a shorter deadline than Aqua), did any WUs run high priority? A team mate told me, that especially the 6.x BOINC clients act very weird when your cache size is close to the deadline of one of the projects. |
|
5)
Message boards :
Problems :
Computation error after 153,085.40 seconds of work -> no credit given
(Message 10887)
Posted 1193 days ago by Ananas ... I had no errors in the last days, but the reason could be ... ... that you haven't found them ;-) http://abcathome.com/result.php?resultid=31299878 Integrator : You had one of those too : http://abcathome.com/result.php?resultid=31385086 The bad ones are still underways, they have a max. error count of 8, so it will take a while until they become extinct (I guess that will be no more than 4 weeks after their first occurance ... Error quota=8 divided by Quota=2 multiplied with deadline=7, so after the first week of March they should mostly be gone). |
|
6)
Message boards :
Problems :
Computation error after 153,085.40 seconds of work -> no credit given
(Message 10866)
Posted 1197 days ago by Ananas All those problems are still the WU configuration problem on project side : "Maximum elapsed time exceeded" means, that they have set a way too low rsc_fpops_bound value, the WUs need more power than they are allowed to have, so the BOINC client aborts them. Some result need about the fpops allowed, so some computers are able to finish them while others abort the same result. It has nothing to do with your hardware, as rsc_fpops_bound does not refer to the CPU time alone but to the CPU time * benchmark values. Old hardware is allowed more time than current boxes. There is a certain dependency to the BOINC client though, some clients return lower benchmark values and in this case this is an advantage, as they have a bit more time to finish the results before reaching the fpops limits. It can be fixed by manipulating the client_state.xml file - either increasing rsc_fpops_bound or reducing the benchmark result - but then you'd need a wingman who does the same. My recommendation (and this is what I did) : Crunch different projects until the ABC project admins stop ignoring the problem and return to ABC when it's fixed. Afterall the problem exists for at least a week now. * * edit : make it nearly 2 weeks :-( |
|
7)
Message boards :
Problems :
Maximum elapsed time exceeded
(Message 10835)
Posted 1205 days ago by Ananas Just as expected : it crashed. Setting to NNW, aborting all unstarted WUs and waiting for someone to fix it on project side. |
|
8)
Message boards :
Problems :
Maximum elapsed time exceeded
(Message 10833)
Posted 1206 days ago by Ananas This is one I currently have - it will take a week to finish on my box (Xeon L5520) - or a few hours less to crash :-( Looks very much like a too narrow fpops limit. p.s.: I could edit client_state.xml of course and increase the limit - but then ... without a wingman, the result would still be trashed :ยด( |
|
9)
Message boards :
Problems :
Maximum elapsed time exceeded
(Message 10832)
Posted 1206 days ago by Ananas Is it really a time limit or rather a fpops limit (rsc_fpops_bound)? The latter would mean, that a faster CPU has a lower time limit and will not be able to finish the result in time either. Example : <workunit> <name>abc_sieve_wu_00805134</name> <app_name>abc-sieve</app_name> <version_num>200</version_num> <rsc_fpops_est>9600000000000.000000</rsc_fpops_est> <rsc_fpops_bound>480000000000000.000000</rsc_fpops_bound> !!!! <rsc_memory_bound>300000000.000000</rsc_memory_bound> <rsc_disk_bound>100000000.000000</rsc_disk_bound> <file_ref> <file_name>abc_sieve_wu_00805134</file_name> <open_name>abc_sieve_wu</open_name> </file_ref> </workunit> the check for a violation of rsc_fpops_est uses the time and the benchmark result to limit the CPU time used. |
|
10)
Message boards :
Problems :
WU Deadlines too SHORT!
(Message 10831)
Posted 1206 days ago by Ananas I received a few WUs lately that will have runtimes between 2.5 days and nearly one week (Xeon L5520), those should really have a few more days. One is currently at 2d 6h and nearly done, the really fat one is only at 34% after 2d 2h. But then ... if the replacement delivery doesn't return earlier than mine, they will probably (hopefully) still be accepted after the deadline. |