| Author |
Message |
|---|
|
|
|
I have my first invalid WU in a long long time: llr_trp_409753_2.4M_125758507
I can't remember the last one I had. What concerns me now is all of the WUs done on this host that only had a quorum of 1.
Peter
____________
35 x 2^3587843+1 is prime! |
|
|
|
|
What concerns me now is all of the WUs done on this host that only had a quorum of 1
e.g. pps_llr_extended_124780975
Who's to say this one isn't incorrect also? |
|
|
|
|
|
I thought all of the numbers were simultaneously tested and double-checked, by two different computers.
What's going on here? What's the point of double-checking any numbers if they're not *all* going to be DCed?!
GIMPS double-checks all of the Mersenne numbers (eventually). |
|
|
|
|
|
PG uses (in the sub-projects PPS, SGS and TRP) something called adaptive replication...
That means if a host is deemed reliable some (not all) of its WUs won't be dc'ed. Primes, however, are always dc'ed.
See the link for further info:
http://boinc.berkeley.edu/trac/wiki/AdaptiveReplication |
|
|
John Honorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
                 
|
What's going on here? What's the point of double-checking any numbers if they're not *all* going to be DCed?!
Chances of finding a new prime is far greater than the chances of finding a missed prime. AR allows progress to move forward faster using today's resources on the leading edge and tomorrow's resources on the DC. Everything will be DC'd (eventually). :)
____________
|
|
|
|
|
|
Firstly I want to say that this is not a massive issue for me.
Basically I'm saying that 1 out of hundreds of WUs on that host has failed.
I know that at present it's a dc effort here at TRP.
But if that host's results are not as reliable as they first appeared to be then surely there is a chance (only a small chance) that a WU on another sub-project it reported as composite could be incorrect.
I know primes are checked and checked and checked (by wingman, primegrid and, if appropriate, Chris Caldwell), they aren't the issue, it's the missed ones... |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
Basically I'm saying that 1 out of hundreds of WUs on that host has failed.
I know that at present it's a dc effort here at TRP.
But if that host's results are not as reliable as they first appeared to be then surely there is a chance (only a small chance) that a WU on another sub-project it reported as composite could be incorrect.
I know primes are checked and checked and checked (by wingman, primegrid and, if appropriate, Chris Caldwell), they aren't the issue, it's the missed ones...
Don't forget about the almost unimaginable infinitesimally small chance of two matching invalid residues. ;)
TRP and PPS are distinctly separate searches. In TRP, a conjectured search, it is important that primes be found for the remaining 56 k's. This is necessary in order to prove the conjecture.
However, a missed prime is not the end of the world. That k may produce another prime at a higher n. In that case, it's a matter of wasted resources needed to find the higher prime. Regardless, all candidates in the TRP project will eventually be double checked so this really is a moot point.
As for missed primes in other projects, SoB is DC'd along with 321, Cullen, PSP, and Woodall. TPS, SGS, and PPS are not DC'd because a missed prime is not a significant mishap. However, they do have AR selected to help reduce the possibility of a missed prime.
TPS and SGS have both found record twins although it is possible they are not the lowest k twins for those n's. The goals of these projects are quite simple, to find record twins and Sophie Germains...not necessarily to find the lowest record.
As for PPS, a missed prime is a missed prime...no harm no foul. If a prime makes it through AR without being discovered, then that's one lucky prime. Of course, the goal is to do a thorough testing of the search space but there's no significant disadvantage of missing a prime. Also, should a DC effort start in the future for this search space, PrimeGrid will have at least one residue for each candidate tested. This will help decrease the DC effort.
Finally, a "trusted" host can go bad at any time. It could produce 10,000 valid results and then have a power surge one day that makes one result go bad. Then produce 10,000 more valid tests. There's always that chance.
A more likely outcome is a "trusted" host produce 10,000 valid results and then through hardware issues (i.e. heat buildup due to improper cooling/dust, etc.) starts to produce bad results. The good news with this is that AR will pick up on this and change the status from "trusted" to "not trusted". It takes a lot of valids to earn the "trusted" status but only one invalid to lose it.
[edit] It's worth repeating:
AR allows progress to move forward faster using today's resources on the leading edge and tomorrow's resources on the DC.
____________
|
|
|
 Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13513
ID: 53948
Credit: 237,712,514
RAC: 0
                           
|
As for missed primes in other projects, SoB is DC'd along with 321, Cullen, PSP, and Woodall.
And GFN. :)
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
And what about prpnet? Does it DC work or does not? |
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
|
And what about prpnet? Does it DC work or does not?
PRPNet can be set to DC, but that option is not selected at this time...too busy searching for primes. ;) Again, residues are always kept for any DC effort that should start in the future.
____________
|
|
|
|
|
|
After getting my last available silver I have just restarted TRP SV and I am finding a normally totally reliable computer is giving 2 good WU out of 60.
Is it me our is there a problem?
I will be looking at the failed units in a couple of hours when I get time at work.
This is very disappointing to me.
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 |
|
|
|
|
|
My host:174914 is now useless on long WUs.
I guess it's time to strip out the GPU and build a new box... |
|
|
|
|
After getting my last available silver I have just restarted TRP SV and I am finding a normally totally reliable computer is giving 2 good WU out of 60.
Is it me our is there a problem?
I will be looking at the failed units in a couple of hours when I get time at work.
This is very disappointing to me.
The error message is :--
Stderr output
<core_client_version>6.10.60</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
BOINC sr2sieve wrapper 1.12
Using Geoffrey Reynolds' sr2sieve 1.8.10
Unrecognized XML in parse_init_data_file: hostid
Skipping: 171718
Skipping: /hostid
Unrecognized XML in parse_init_data_file: starting_elapsed_time
Skipping: 0.000000
Skipping: /starting_elapsed_time
Unrecognized XML in parse_init_data_file: computation_deadline
Skipping: 1337544571.136000
Skipping: /computation_deadline
Major OS version: 6; Minor OS version: 0
pid: 116, thr: 120
Factors file not found
called boinc_finish
</stderr_txt>
]]>
Can anyone help?
____________
Member team AUSTRALIA
My lucky number is 9291*2^1085585+1 |
|
|
|
|
After getting my last available silver I have just restarted TRP SV and I am finding a normally totally reliable computer is giving 2 good WU out of 60.
Is it me our is there a problem?
I will be looking at the failed units in a couple of hours when I get time at work.
This is very disappointing to me.
The error message is :--
Stderr output
<core_client_version>6.10.60</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
BOINC sr2sieve wrapper 1.12
Using Geoffrey Reynolds' sr2sieve 1.8.10
Unrecognized XML in parse_init_data_file: hostid
Skipping: 171718
Skipping: /hostid
Unrecognized XML in parse_init_data_file: starting_elapsed_time
Skipping: 0.000000
Skipping: /starting_elapsed_time
Unrecognized XML in parse_init_data_file: computation_deadline
Skipping: 1337544571.136000
Skipping: /computation_deadline
Major OS version: 6; Minor OS version: 0
pid: 116, thr: 120
Factors file not found
called boinc_finish
</stderr_txt>
]]>
Can anyone help?
Hi DaveB,
http://www.primegrid.com/forum_thread.php?id=3182&nowrap=true#34611
posted this 29 Mar 2011
will resume TRP Sieve soon as well and respond the result's
parabol
____________
I'm a prime millionaire !
9*2^3497442+1 |
|
|
|
|
|
Hi all,
this is the Stderr output of my last sieve file:
<core_client_version>7.0.25</core_client_version>
<![CDATA[
<stderr_txt>
BOINC sr2sieve wrapper 1.12
Using Geoffrey Reynolds' sr2sieve 1.8.10
*** Note: the wrapper is 32bit; the application that does the actual work is 64bit ***
Unrecognized XML in parse_init_data_file: userid
Skipping: 57406
Skipping: /userid
Unrecognized XML in parse_init_data_file: teamid
Skipping: 2280
Skipping: /teamid
Unrecognized XML in parse_init_data_file: hostid
Skipping: 256643
Skipping: /hostid
Unrecognized XML in parse_init_data_file: result_name
Skipping: TRP_sieve_1817225_1
Skipping: /result_name
Unrecognized XML in parse_init_data_file: starting_elapsed_time
Skipping: 0.000000
Skipping: /starting_elapsed_time
Unrecognized XML in parse_init_data_file: using_sandbox
Skipping: 0
Skipping: /using_sandbox
Unrecognized XML in parse_init_data_file: gpu_type
Skipping: /gpu_type
Unrecognized XML in parse_init_data_file: gpu_device_num
Skipping: -1
Skipping: /gpu_device_num
Unrecognized XML in parse_init_data_file: gpu_opencl_dev_index
Skipping: -1
Skipping: /gpu_opencl_dev_index
Unrecognized XML in parse_init_data_file: ncpus
Skipping: 1.000000
Skipping: /ncpus
Unrecognized XML in parse_init_data_file: computation_deadline
Skipping: 1337693270.000000
Skipping: /computation_deadline
Major OS version: 6; Minor OS version: 1
pid: 152, thr: 156
.numbers
Factors file not found
called boinc_finish
</stderr_txt>
]]>
However I do get credit on it.
Could one explain what happens here ?
Tks..parabol
____________
I'm a prime millionaire !
9*2^3497442+1 |
|
|
|
|
Factors file not found
Are you wondering about this? If no factors are found there is no file. This means the number gets passed to the LLR for a further look. |
|
|
