Poaching (please don't!)

[kriesel]

Poaching 361398269 was a complete waste of time, except for the 6234. GHD credit the poacher received.
I've had this running essentially continuously with proof generation on an old pro-workstation system with ECC ram, now >80% complete. The poacher apparently did not produce and upload a proof file, so completing my run will be necessary to provide a DC, and an inexpensive TC will come from the power 8 proof file and its cert (current ETA Dec 2022). The total cycles wasted by this poaching will likely be at least ~6234 * 255 / 256 ~ 6210 GHD. That's enough for more than 12. first time PRP with proof generation at the current wavefront, with ~39. times the chance of finding a prime.

If interested in helping test some of the remaining 7-digit or 8-digit rhyme length mersenne rhymes, please coordinate with me by PM to avoid duplication and waste.

Originally Posted by XZT

Is there a way to view more detailed assignment and update history?
Is there a way to be notified if an assignment is expired (e.g. an old assignment reports in late) or potentially poached?

Not that I know of, or I'd be using it. Reviewing our own individual https://www.mersenne.org/workload/ for missing assignments becomes less workable, the more assignments there are.

[tuckerkao]

Originally Posted by kriesel

Poaching 361398269 was a complete waste of time, except for the 6234. GHD credit the poacher received.

The poacher must have seen that M361398269 had the trial factoring worked all the way up to 2⁸³ while all other unfactored exponents were in 2⁷⁴ for subrange M361.3M.

The solution to avoid that would be to have several exponents brought up to the same TF depth or even higher while you are running the PRP test for only 1 of them. To see my case, check my own thread.

[kriesel]

384989999 also poached by ShenTianYing, and even worse, it now appears it was gpuowl with no proof file uploaded, 9 days after submission of the PRP result. (Gpuowl does not automatically upload proof files or results on its own like primenet-enabled prime95 would.)
Proof files should be generated and promptly uploaded.
Otherwise a second full run is needed, and may as well generate a proof file in case the PRP residues don't match, so the without-proof-file first run produces effectively a 100.000% complete waste of resources.
It's especially important on large exponents. This one causes a waste of >7000. GHD, enough to DC over FORTY wavefront DC exponents.