Can first-time LLs be converted by user to PRP?
 

I have several first-time LL assignments queued up on my old Haswell quad, which I would like to move from the CPU-side of things to the new Radeon 7 GPU I recently installed in the same box - no point having the CPU burning similar watts as the GPU, for less than 1/10th the throughput. But GpuOwl only supports PRP tests, not LL. Since these are first-time-test assignments, can I simply munge the worktodo entries to convert to PRP, and submit those results to the server?

If so, what does the trailing '1' (= p-1 done) in the Test= entry map to in terms of PRP? George described that trailing-digit entry in first-time PRPs (2nd from last in PRP-DC, which append 2 more entries, base and residue type of the 1st test) as "the number of PRP tests that will be saved if P-1 is done and finds a factor". So since p-1 has already been done on the expos in question, that means a trailing '0' in the manually-created PRP lines, yes?

[QUOTE=ewmayer;537006]I have several first-time LL assignments queued up on my old Haswell quad, which I would like to move from the CPU-side of things to the new Radeon 7 GPU I recently installed in the same box - no point having the CPU burning similar watts as the GPU, for less than 1/10th the throughput. But GpuOwl only supports PRP tests, not LL. Since these are first-time-test assignments, can I simply munge the worktodo entries to convert to PRP, and submit those results to the server?

If so, what does the trailing '1' (= p-1 done) in the Test= entry map to in terms of PRP? George described that trailing-digit entry in first-time PRPs (2nd from last in PRP-DC, which append 2 more entries, base and residue type of the 1st test) as "the number of PRP tests that will be saved if P-1 is done and finds a factor". So since p-1 has already been done on the expos in question, that means a trailing '0' in the manually-created PRP lines, yes?[/QUOTE]

You can unreserve your LL numbers in your PrimeNet account and then request new PRPs.

[QUOTE=ewmayer;537006]If so, what does the trailing '1' (= p-1 done) in the Test= entry map to in terms of PRP? George described that trailing-digit entry in first-time PRPs (2nd from last in PRP-DC, which append 2 more entries, base and residue type of the 1st test) as "the number of PRP tests that will be saved if P-1 is done and finds a factor". So since p-1 has already been done on the expos in question, that means a trailing '0' in the manually-created PRP lines, yes?[/QUOTE]

See this post for the explanation of the worktodo entry for PRP
[url]https://www.mersenneforum.org/showpost.php?p=535506&postcount=6[/url]

The all-GIMPS-applications worktodo entry format reference post is at [URL]https://www.mersenneforum.org/showpost.php?p=522098&postcount=22[/URL]

Note that not only are there PRP residue types 1 through 5, there's also a type 0 relating to gpuowl PRP-1, where PRP and P-1 are done simultaneously in certain versions. This was cleverly done by using a base much larger than 3 for the PRP iterations.

[QUOTE=kriesel;537010]The all-GIMPS-applications worktodo entry format reference post is at [URL]https://www.mersenneforum.org/showpost.php?p=522098&postcount=22[/URL]

Note that not only are there PRP residue types 1 through 5, there's also a type 0 relating to gpuowl PRP-1, where PRP and P-1 are done simultaneously in certain versions. This was cleverly done by using a base much larger than 3 for the PRP iterations.[/QUOTE]

Yes, I get all that - after all I've implemented the PRP assignment parsing in my own code (which ignores the p-1 field, hence my query re. that one) - but would like a definitive answer as to whether one can switch the test type oneself. Per Paul, sounds like 'no'. That's fine, I can just queue 'em up on my little Intel NUC, which has proved highly reliable at LL testing (more so than the Haswell) and will be needing more work soon.

[QUOTE=ewmayer;537012]Yes, I get all that - after all I've implemented the PRP assignment parsing in my own code (which ignores the p-1 field, hence my query re. that one) - but would like a definitive answer as to whether one can switch the test type oneself. Per Paul, sounds like 'no'. That's fine, I can just queue 'em up on my little Intel NUC, which has proved highly reliable at LL testing (more so than the Haswell) and will be needing more work soon.[/QUOTE]

I think there might be a conflict if double checking is going on. The server will get its knickers in a twist!

[QUOTE=paulunderwood;537013]I think there might be a conflict if double checking is going on. The server will get its knickers in a twist![/QUOTE]

That's why I specifically said "first-time tests" in my OP - obviously you wouldn't want to do such a thing for DCs.

[QUOTE=ewmayer;537014]That's why I specifically said "first-time tests" in my OP - obviously you wouldn't want to do such a thing for DCs.[/QUOTE]

Yes. But if a first time is given out as LL and you change it to PRP and it is given out to another as LL-DC before you return your PRP result then it would be an awful wast of resources. But what do I know? -- maybe DCs work differently, being issued only after the first time has been received by the server.

You can just create your own PRP lines for the Radeon, and then turn in the manual PRP results which should be accepted by the server.
Then your LL assignments will eventually expire or you can probably manually cancel them by then.

A bit off-topic: Something I have been thinking about. There is PRP and then there is LL. If I understand it correctly, PRP is probabilistic and LL is deterministic. It seems to take the same amount of time to run a PRP as it does a LL. I do not understand so much emphasis being placed on PRP when the same test will need to be repeated by a LL test. Anyone?

[QUOTE=storm5510;537022]A bit off-topic: Something I have been thinking about. There is PRP and then there is LL.[/QUOTE]PRP can't [B]prove[/B] a prime. But, it can disprove many, many non-primes. And with the error checking that can be done, DC's will find less incorrect results. So, overall, less work for the project and less likely that a prime will slip through the cracks.