is M21934219 composite ?

[S485122]

The exponent 21934219 has now been tested 651 times, 650 tests have matching residues, one test had a bad residue. Curiously there is no periodicity in the returning of results, they are turned in every 4 days on average, but with a standard deviation of nearly 10.

Exponent 20940839 has an all numeric residue and the tests have another interesting characteristic.

Jacob

[mdettweiler]

Originally Posted by S485122

The exponent 21934219 has now been tested 651 times, 650 tests have matching residues, one test had a bad residue. Curiously there is no periodicity in the returning of results, they are turned in every 4 days on average, but with a standard deviation of nearly 10.

Exponent 20940839 has an all numeric residue and the tests have another interesting characteristic.

Jacob

For 21934219, my guess is that this "Dandot Moore" guy has something like a stray v4 computer with a read-only disk that's causing it to keep testing the same exponent over and over again. And if the guy isn't actively watching that computer any more (and may not even be participating in GIMPS any more), he might not have signed up for v5 and therefore at that point the results started coming in under ANONYMOUS.

[jrk]

Originally Posted by mdettweiler

For 21934219, my guess is that this "Dandot Moore" guy has something like a stray v4 computer with a read-only disk that's causing it to keep testing the same exponent over and over again. And if the guy isn't actively watching that computer any more (and may not even be participating in GIMPS any more), he might not have signed up for v5 and therefore at that point the results started coming in under ANONYMOUS.

Stranger things have happened. I once had my email inbox fill up because someone (or their email client/server) kept spewing the same message at me every minute. He kept insisting that he wasn't sending the repeat messages. I don't know what finally happened, but they stopped coming in after about a week.

[retina]

Originally Posted by S485122

Exponent 20940839 ... tests have another interesting characteristic.

Which is ... what?

[moebius]

What a waste of computing resources, or even a world record attempt in LL testing of the same Exponent.

[cheesehead]

Originally Posted by S485122

The exponent 21934219 has now been tested 651 times, 650 tests have matching residues, one test had a bad residue

Someone, please factor this one and put it out of its misery!

[retina]

Originally Posted by cheesehead

Someone, please factor this one and put it out of its misery!

Okay, I'll do it. Should I use TF, ECM, QS, NFS or P-1?

Perhaps Shor's algorithm would be better for this!

[cheesehead]

Originally Posted by retina

Should I use TF, ECM, QS, NFS or P-1?

Perhaps Shor's algorithm would be better for this!

One on each core of a hexacore.

[Brian-E]

Originally Posted by mdettweiler

For 21934219, my guess is that this "Dandot Moore" guy has something like a stray v4 computer with a read-only disk that's causing it to keep testing the same exponent over and over again.

If the exponent is really being tested many times, there must be many machines involved because sometimes many (20 or more) test results have come in on the same day. If the same machine is sending those results then must be a server giving multiple clients the same test.

Another possibility is that the LL-tests are not really being done each time, just re-reporting of the result multiple times, but in that case something must have gone wrong with the assignment/reporting system because that should not result in Primenet recording it as different tests, should it?

[retina]

Originally Posted by cheesehead

One on each core of a hexacore.

Oh, that would be the new hex-core from <someone> with a QC as the sixth core. Should be no problem then. I'll start it now and watch it until it completes.

[mdettweiler]

Originally Posted by Brian-E

If the exponent is really being tested many times, there must be many machines involved because sometimes many (20 or more) test results have come in on the same day. If the same machine is sending those results then must be a server giving multiple clients the same test.

Another possibility is that the LL-tests are not really being done each time, just re-reporting of the result multiple times, but in that case something must have gone wrong with the assignment/reporting system because that should not result in Primenet recording it as different tests, should it?

Yeah, that sounds even more like a system image gone awry: perhaps the person is a sysadmin and thought "hey, I can borg all of my company's machines by putting Prime95 on their master system image"--but if he just copied one Prime95 setup while it was still working on a test (in this case, 21934219), each time the image was applied the first thing the newly-spawned Prime95 would do is finish that test. Which then leaves two possibilities as to why it continues to be repeated over and over:

-Either the machines have read-only disks, and therefore each of them keep doing the same test over and over again;
-Or this imagine is in such active use that the image is being applied to a new machine every few days, leading to the influx of tests we've observed.