mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   A problem when manually assigning P-1 factoring (https://www.mersenneforum.org/showthread.php?t=27134)

 congsz 2021-09-11 01:15

A problem when manually assigning P-1 factoring

I intend to P-1 factor M323232323. In Prime95 (v30.3, Windows) I clicked "Advanced"--"P-1" and typed 1*2[SUP]323232323[/SUP]-1, B1 = 3232323, B2 = 323232323. But after I clicked the "OK" button, the communication thread told me that this was an "unsupported assignment work type:3", why did this happen? Are these bounds B1 and B2 I chose incorrect? (I had visited the [URL="https://www.mersenne.org/report_exponent/?exp_lo=323232323&exp_hi=&full=1&ecmhist=1"]exponent status[/URL] page and saw the current bounds for M323232323 is now B1 = 2 220 000, B2 = 45 510 000 before my assignment) If so, how can I properly(and carefully, manually) select the appropriate bounds for P-1 factoring?

I'm new here and still yet to fully understand the math behind it and how these assignments work. Expecting someone to help me explain this. Thanks!

 kriesel 2021-09-11 03:35

Welcome to the forum. The bounds you entered are somewhat excessive compared to what's recommended at [URL]https://www.mersenne.ca/exponent/323232323[/URL]. I usually aim for the GPU72 row bounds on a GPU.

But I can't reproduce your issue on prime95 [URL="https://mersenneforum.org/showpost.php?p=577159&postcount=256"]v30.6b4[/URL].
It generates a worktodo entry Pminus1=1,2,323232323,-1,3232323,323232323
and if that is rotated to the front of the worktodo queue for the worker (by manual edit), begins to process it.
See the [URL="https://www.mersenneforum.org/showpost.php?p=522098&postcount=22"]worktodo entry formats[/URL] reference post for the meaning of those fields. (Part of a large and growing compilation of [URL="https://mersenneforum.org/showthread.php?t=24607"]reference info[/URL].)

 kruoli 2021-09-12 19:49

The calculation should have started either way, correct? The error in the communication thread is not critical. PrimeNet does not support P-1 with manual bounds, which is the work type started when entering P-1 via the GUI. Either way, the calculation can be executed and the result will be submitted. If no prior or lesser P-1 has been done, you will even get appropriate credits.

I guess you choose more excessive bounds knowingly. If this was not your intention, please consider using the suggested bounds. But since the exponent is of some special form (it looks "special" in base 10), I assume you wanted the higher bounds. You can safely proceed the computation in this case. It is your PC and of course you may do what you want with it.

@kriesel: You have not tested with PrimeNet enabled. This is necessary for comparison because the OP had output in the communication thread. I deduct that from you post where there should have been an AID or N/A in the worktodo.txt line you gave.

 kriesel 2021-09-13 00:12

1 Attachment(s)
[QUOTE=kruoli;587780]@kriesel: You have not tested with PrimeNet enabled. This is necessary for comparison because the OP had output in the communication thread. I deduct that from you post where there should have been an AID or N/A in the worktodo.txt line you gave.[/QUOTE]False. PrimeNet [B]was[/B] enabled for my entire session and more. The Pminus1 line I gave in the earlier post is the form the advanced dialog generated in a session WITH PRIMENET ON THE CLIENT SET OPERATIONAL. ("UsePrimenet=1" in prime.txt, etc.) PrimeNet communication is periodic, not instant or continuous. Pminus1 is a known prime95 form but is not an accepted PrimeNet API form. But the prime95 client will run with it anyway. I use that routinely to perform mersenne.ca GPU72 row bounds on various exponents. Results are accepted; assignment is not, so no status is available for that worktodo entry on the personal assignments web page.

You can test all that for yourself, by pasting the worktodo line I gave into your own worktodo, confirm there is no AID or N/A, confirm your PrimeNet configurations are complete and correct and operational, let it run a while but not so long it does a PrimeNet sync, observe progress occurs with no error yet, check the worktodo still has no AID or N/A, then use Advanced, Manual communication, send new expected dates, or wait until the client gets around to its usual sync, and check again. Check your personal assignments page too if you like.

(I didn't spend that much time because PrimeNet sync seemed irrelevant to what I then thought to be the OP's question, an issue with the client software not accepting it as input or generating an error on its own, and concern about bounds. I thought he meant immediately after clicking OK on the advanced P-1 menu that an error appeared. Note the OP did not indicate in post 1 anything about a "Primenet error 44" or time delays before an error appeared. I missed the "communication window" clue.)

Eventually, PrimeNet gets involved and rejects the Pminus1 assignment type during the sync, THEN inserts the N/A in the worktodo line. That could be seconds, or many hours, after the initial manual creation of the Pminus1 line at the client through the advanced menu, since sync happens typically daily.

Note also, this from [B]primenet.h[/B] of prime95 v30.4b8 or v30.6b4:
[CODE]/* Valid work_preference values */
...
#define PRIMENET_WP_[B]PMINUS1[/B] 3 /* P-1 of small Mersennes --- [B]not supported[/B] */
#define PRIMENET_WP_PFACTOR 4 /* P-1 of large Mersennes */
[/CODE]Pminus1 worktodo entry form is for prime95 where we can explicitly specify bounds, though.

OP: sometimes a screen capture, or Edit, Copy Window, post the relevant window text portion, will help greatly in clarifying the issue and details you're actually seeing, so responses can be more pertinent the first time.

 kruoli 2021-09-13 08:42

[QUOTE=kriesel;587790]You can test all that for yourself, by pasting the worktodo line I gave into your own worktodo, confirm there is no AID or N/A, confirm your PrimeNet configurations are complete and correct and operational, let it run a while but not so long it does a PrimeNet sync, observe progress occurs with no error yet, check the worktodo still has no AID or N/A, then use Advanced, Manual communication, send new expected dates, or wait until the client gets around to its usual sync, and check again. Check your personal assignments page too if you like.[/QUOTE]

Of course I did that. I got [C]Pminus1=N/A,1,2,323232323,-1,3232323,323232323[/C], as expected. This is consistently a few seconds [I]after startup[/I], nearly immediately. Everything else would be a bug, since Prime95 would need to try to reregister this assignment on every startup since it cannot distinguish between "not tried" and "not successful". This is on one of my machines which are doing P-1 regularly, so I guess there might be some other option that differs between our versions. I have not tried to change the worktodo.txt while Prime95 was running since I deem that bad habit. I never had the problem that I had to wait for the next communication timeframe for new worktodo lines to be registered or being in need to force manual communication. Quite the contrary, when using worktodo.add, it will register everything immediately as soon as Prime95 grabs the .add file.

 kriesel 2021-09-13 12:59

Stop all workers, edit worktodo.txt, continue workers, without exit and restart program, is supported, as can be seen by reading the prime95 source code. OP's issue had nothing to do with worktodo.add or worker stop/start. He used the advanced menu P-1 option to create a worktodo item. Which would go directly in worktodo.txt. It's quite common to need to force the sync manually to avoid the lengthy delay. See for example the directions in [URL="https://mersenneforum.org/showpost.php?p=510268&postcount=1"]Strategic DC & TC[/URL].

 All times are UTC. The time now is 08:00.