[url=http://www.mersenne.org/M40682749]40682749[/url] is a low exponent that needs a triple check

DoubleCheck=40682749,72,1

[QUOTE=Madpoo;440527]Thanks. I have a hunch that when you turn in your result for 43408411 you'll see the weird thing I'm talking about (assuming you match my result). :smile:[/QUOTE]

[URL="http://www.mersenne.org/M43408411"]M43408411[/URL]

Hmmm, I see what you mean: [B][U]DD[/U][/B]3D1B[B][U]00[/U][/B]3CBD[B][U]FF88[/U][/B]. Four out of the eight pairs consist of two identical hexadecimal digits. The odds of that are only one percent (actually 0.990%). :grin:

[QUOTE=GP2;440645][URL="http://www.mersenne.org/M43408411"]M43408411[/URL]
The odds of that are only one percent (actually 0.990%). :grin:[/QUOTE]

Yup! Even funnier, quite a few similar "low odds" results in the last few days, in the same range. One flaky (and fast, certainly fast) machine, perhaps? Or "premature optimization" (from iteration 1, jump straight to the last :smile:)?

[QUOTE=Madpoo;440527]Thanks. I have a hunch that when you turn in your result for 43408411 you'll see the weird thing I'm talking about (assuming you match my result). :smile:[/QUOTE]

Ohhh, I see! Did the alerting system work?

[QUOTE=Mark Rose;440649]Ohhh, I see! Did the alerting system work?[/QUOTE]

It did in fact. :smile:

Unfortunately this same user has been having issues with a (small) subset of machines returning false positives (while doing double checks), so I've been doing a lot of triple-checks recently to make sure. This one just happened to need a quad check because the first result was also bad.

We're (well, George really) trying to figure out how these false positives are showing up since it's using the latest mprime 28.9 and a Xeon E5-2697 v3 (same CPU I have on one system that hasn't had any issues).

Anyway, thanks for the quad check just so we could make sure to close the books on that one.

[QUOTE=Madpoo;440669]It did in fact. :smile:

Unfortunately this same user has been having issues with a (small) subset of machines returning false positives (while doing double checks), so I've been doing a lot of triple-checks recently to make sure. This one just happened to need a quad check because the first result was also bad.

We're (well, George really) trying to figure out how these false positives are showing up since it's using the latest mprime 28.9 and a Xeon E5-2697 v3 (same CPU I have on one system that hasn't had any issues).

Anyway, thanks for the quad check just so we could make sure to close the books on that one.[/QUOTE]

I'm really surprised to hear of an issue with a Xeon - I also have some 2697v3s + ECC memory and had always worked under the sssumption that system would never make mistakes.

[QUOTE=Madpoo;440669]It did in fact. :smile:[/QUOTE]

I should really get around to building my own automated system to alert me when a potential prime is reported.

[QUOTE=airsquirrels;439925]I have 17 Verified, 2 bad, and 12 mismatches from that card. If anyone wants to TC. So far all the 7xM tests that card has done have mismatched.[/QUOTE]

Final tally: 3 verified, 9 bad (out of 12 mismatches)

[CODE]
38096281 bad
38133923 [B]good[/B]
38324009 bad
38324521 bad
39299047 [B]good[/B]
39334423 bad
73602083 [B]good[/B]
73648273 bad
37569533 bad
38096281 bad
73500677 bad
74049323 bad
[/CODE]

[QUOTE=Mark Rose;440672]I should really get around to building my own automated system to alert me when a potential prime is reported.[/QUOTE]

LOL... well, hopefully if things are working right on the server, there won't be any hints of what the exponent is until it's been officially announced.

Speaking of primes though, we have some additional false positives from that one batch of systems... We're down to just one unconfirmed one but it's in a weird category from that user where it has a zero residue but it knows it wasn't prime. Not sure how that works... the code may have identified issues during the run and even though it's zero at the end, it knows it's not prime?

In other words, it's been a mix of "is prime!" with a zero residue, and "is not prime" with zero residue. How bizarre.

And there's another one that I'd like to get a triple-check on if possible. It's from that wonky version of CUDALucas that was generating false positives... I've done my double-check so I'm sure it's not prime but a triple-check will put my mind at ease.

It's this:
[URL="http://www.mersenne.org/M77127829"]M77127829[/URL]

Ignore the active assignment... that's the user that reported the false positive and since it was manually reported, there's fortunately a system in place to keep "is prime" results from being auto processed manually. Instead we get notified to look into it and make sure it's legit.

Kind of funny, all the false positives lately...hopefully we figure out that one user's issue soon.

[QUOTE=Madpoo;440887]LOL... well, hopefully if things are working right on the server, there won't be any hints of what the exponent is until it's been officially announced.[/quote]

Even if I did find a way, I wouldn't advertise it: I wouldn't want to spoil the fun!

[QUOTE=Madpoo;440887]And there's another one that I'd like to get a triple-check on if possible. It's from that wonky version of CUDALucas that was generating false positives... I've done my double-check so I'm sure it's not prime but a triple-check will put my mind at ease.

It's this:
[URL="http://www.mersenne.org/M77127829"]M77127829[/URL][/QUOTE]

I queued it.