Trippple Checks

[frmky]

35500249 needs a TC.

[Mark Rose]

Originally Posted by frmky

35500249 needs a TC.

Mine.

[LaurV]

36793843 was a regular (and legal) assignment (not from Madpoo's lists, but taken directly from PrimeNet). It mismatched and I TC-ed before reporting it, just to avoid increasing my number of bad results. The TC matched the DC, so my result id 100% correct (same hardware but different shift, and I reported both). However, if someone wants to play with a QC, to remove the "self DC doubt", be my guest.

[Mark Rose]

Originally Posted by LaurV

36793843 was a regular (and legal) assignment (not from Madpoo's lists, but taken directly from PrimeNet). It mismatched and I TC-ed before reporting it, just to avoid increasing my number of bad results. The TC matched the DC, so my result id 100% correct (same hardware but different shift, and I reported both). However, if someone wants to play with a QC, to remove the "self DC doubt", be my guest.

Sure, I'll do it.

[Madpoo]

Originally Posted by Mark Rose

Sure, I'll do it.

Ah, I'm glad I read this. I was just about to run it myself when I saw it pop up in my "self verified exponent" query.

Thanks Mark.

[Mark Rose]

Originally Posted by Madpoo

Ah, I'm glad I read this. I was just about to run it myself when I saw it pop up in my "self verified exponent" query.

Thanks Mark.

I should have it done within a week.

[Dubslow]

Kindly requesting that petrw1 speed up his TC of http://www.mersenne.org/report_expon...5019883&full=1

Or maybe we can get a report on the other machine's results. I have full confidence in my own machine.

Edit: Going through all my unverified LLs, I have one suspect result that remains unverified, though a second run was done independently. I don't remember what caused the errors, but it seems to have been a one time thing. Would someone like to TC http://www.mersenne.org/report_expon...5299699&full=1 for no particularly good reason?

[Mark Rose]

Originally Posted by Dubslow

Kindly requesting that petrw1 speed up his TC of http://www.mersenne.org/report_expon...5019883&full=1

Or maybe we can get a report on the other machine's results. I have full confidence in my own machine.

Edit: Going through all my unverified LLs, I have one suspect result that remains unverified, though a second run was done independently. I don't remember what caused the errors, but it seems to have been a one time thing. Would someone like to TC http://www.mersenne.org/report_expon...5299699&full=1 for no particularly good reason?

Will do. Should be done by the end of the year.

[Madpoo]

Originally Posted by Dubslow

Kindly requesting that petrw1 speed up his TC of http://www.mersenne.org/report_expon...5019883&full=1

Or maybe we can get a report on the other machine's results. I have full confidence in my own machine.

Edit: Going through all my unverified LLs, I have one suspect result that remains unverified, though a second run was done independently. I don't remember what caused the errors, but it seems to have been a one time thing. Would someone like to TC http://www.mersenne.org/report_expon...5299699&full=1 for no particularly good reason?

In the case of M35019883:
Your system (cpu id by year and by app version): 23 good, zero bad
Other system: zero good, 1 bad

In the case of M45299699:
Your system: 1 good, 1 bad (and your result was marked suspect)
Other system: 13 good, zero bad

Based on that I'd guess you had the correct result on M35019883 and the wrong one on M45299699.

[Dubslow]

Originally Posted by Madpoo

In the case of M45299699:
Your system: 1 good, 1 bad (and your result was marked suspect)
Other system: 13 good, zero bad

Is that only in the last year? That computer has done at least 4 tests in the last year that all matched. That computer has had at most 2 guid's, and the same one for at least 4 years, and several more than 2 results. By my count, it has ~75 LL results. http://www.mersenne.org/report_ll/?e...exfactor=1&B1= (note that one of the verified results did have a non-zero error code).

Meanwhile my laptop has a dozen or so good results and one bad, my GPU has another ~70-80 results with 2 known bad, and my final machine has ~200 (or maybe more) verified results with none bad (and a small handful with non-zero error codes, all sharing the same form as the other error code above that was verified, while the one I've asked about here has an error in a different category as well).

Originally Posted by Madpoo

Based on that I'd guess you had the correct result on M35019883 and the wrong one on M45299699.

I concur.

[Madpoo]

Originally Posted by Dubslow

Is that only in the last year? That computer has done at least 4 tests in the last year that all matched.

I break down the stats by:
user id + cpu id + year + app version

It helps to narrow down the stats to smaller time frames to help isolate periods of time when a machine may have been flaky versus running okay. Some systems are *always* flaky... no matter the year or what version of the app they were using. Some of them though seem to start out pretty well but then go downhill, and some are the opposite (maybe they overclock too much at first and then settle down to more stable values).

That computer has had at most 2 guid's, and the same one for at least 4 years, and several more than 2 results.

Yup, those would be broken down in subsets of data, not one lump classification.