It's not just a different polynomial, it's a completely [URL="http://factordb.com/index.php?id=1100000000127408717"]different number[/URL] that was done [URL="https://mersenneforum.org/showthread.php?p=541651#post541651"]last year[/URL]. That job had the same name as this one, 417__011_3m1, hence the confusion. It seems the job last year was given that name in error: it should have been 145__031_3m1.

This also explains the yield discrepancy that RichD noticed in the 14e queue management thread.

417__011_3m1
 

Back in Aoril 2020 there was a number requested [url=https://www.mersenneforum.org/showpost.php?p=541651&postcount=2329]here[/url]. After going through the test-sieving and backward tracing to get the sigma relationships I must have copied the wrong prime(p93) into the post. The number and poly was correct.

Now, the real p93 needs factoring. Perhaps some of the old files from 17 months ago were still hanging around and not overwritten. That’s just my guess.

I'll move on to the next number pending resolution of this issue.

I found the cause! 417__011_3m1 was also the name of a composite factored in April 2020. The server must have reverted back to the original .poly file once it started processing(?)

Need a naming scheme which avoids this problem, or manually double check before creating a new job name.

ETA: a day late and a dollar short!

I think the naming scheme is fine. The problem started with me posting the wrong p93 so the title could be generated. I have a feeling the two numbers in question were right next to each other (at the time) in the t800 file and I grabbed the wrong one. I am the one that needs to be double-checked...

I guess the resolution would be to resubmit with a suffix "a".

[QUOTE=RichD;587704]I guess the resolution would be to resubmit with a suffix "a".[/QUOTE]

Agreed. I will resubmit the job as 417__011_3m1a.

Greg will have to delete this problem job, I can’t do it.

You all solved it! Yes, the polynomial from last year with the same name was found and used by BOINC. Our first name collision.

[QUOTE=frmky;587717]Our first name collision.[/QUOTE]

It was by accident. The prime base in the post was incorrect for the number (and poly) being requested. Technically, there should have not been a collision. :smile:

363__249_11m1 is [URL="https://pastebin.com/raw/qxztWbq2"]done[/URL].

Feeding the GPU right?! Lol