mersenneforum.org

mersenneforum.org (https://www.mersenneforum.org/index.php)
-   Information & Answers (https://www.mersenneforum.org/forumdisplay.php?f=38)
-   -   Can you Make a manual assignment to extend a TF from to 74-to 75? (https://www.mersenneforum.org/showthread.php?t=26946)

birtwistlecaleb 2021-06-25 02:19

Can you Make a manual assignment to extend a TF from to 74-to 75?
 
Like changing just this:
Factor=DONOTSHAREYOURASSIGNMENTID,213560531,73,74
To this:
Factor=DONOTSHAREYOURASSIGNMENTID,213560531,73,74
Factor=DONOTSHAREYOURASSIGNMENTID,213560531,74,75
Or this:
Factor=DONOTSHAREYOURASSIGNMENTID,213560531,73,75
In the worktodo.txt or manual assignments

Uncwilly 2021-06-25 02:34

You can.... but are you using a CPU to do this?
That long string is your Assignment ID, you should not share it. It has been removed by a moderator.

If you are doing this on a CPU, there are much better ways to use your CPU. Leave the factoring to GPUs.

Yes you can do that, but once your machine turns in the result for the first bit, your assignment will be complete. It is possible, but highly unlikely that someone else would get assigned the next bit level.
Either of the ways you have put it will work.

birtwistlecaleb 2021-06-25 03:02

Ok, i updated the assignments i don't have started yet to go to 76 instead of 74, will this work?
Example:
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560537,73,74
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560563,73,74
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560593,73,74
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560651,73,74
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560693,73,74
To:
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560537,73,76
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560563,73,76
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560593,73,76
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560651,73,76
Factor=DQW4W9WGXCQDQW4W9WGXCQDQW4W,213560693,73,76
(And yes, i did repeat the rickroll link to censor the id.)

Uncwilly 2021-06-25 03:30

That will work. Again are you doing this on a CPU? And do you realize that going from 74 to 75 is twice the work as from 73 to 74. And going from 75 to 76 is twice that. So that is 7 times the original work amount.

birtwistlecaleb 2021-06-25 03:40

For some reason, it is not showing up on the status too. Is this bad?

Uncwilly 2021-06-25 04:23

Which status are you talking about? If you are talking about Prime95, did you stop it, then do the edit, then save, and then restart Prime95? Or better stop and exit, then edit etc.

If you are talking about on your assignments page on mersenne.org, it won't change the assignment with a simple edit. And since it is already assigned, you can't do another trick (self assigning the work.)

birtwistlecaleb 2021-06-25 04:35

Thanks! Will it still count as "No factors below 2[SUP]74[/SUP]"?

Uncwilly 2021-06-25 05:07

You should get credit for work done and the db will reflect it.

birtwistlecaleb 2021-06-25 05:10

So it will show "No factors below 2[SUP]75[/SUP]"?

LaurV 2021-06-25 06:28

You can edit your worktodo the way you like it, as long as you don't step on other people toes (i.e. poaching exponents which are already assigned), and as long as you stop P95 before - otherwise P95 is caching the text, and it will revert back to the old version at first checkpoint, your edit being lost. Stopping P95 means clicking "Test/Stop" or "Test/Exit" in the menu. Clicking the red X button will not stop P95, just minimize/hide it, but it still works in background. If you modify config files, like prime.txt or local.txt, then "Test/Exit" is a must. For modifying "worktodo.txt", or "worktodo.add", "Test/Stop" is enough. Don't forget "Test/Continue" after editing is finished.

Re unasigned work: Unassigned work means you do some work of which the server is not aware. If the server is not aware of the fact that you do the work, it may assign the respective work to someone else, and you both end up wasting time and resources to do the same work two times. Tip here: don't work unassigned if your specific work takes long time (like, hours, days, etc) - other people have no idea what you are doing, and somebody may do the same work, assigned or not, without you knowing it (because you don't check the assignments every minute), and you both will end up wasting resources.

Some people do unassigned work, they report the results, get credit, and everything is fine. For TF work that takes minutes, like low bitlevel, high exponent, it is common that people will take a range of exponents, add them manually in worktodo file, fire the GPU, and do the work, unassigned. Because this is sometimes more convenient, but they will do their homework, like looking to the ranges and see which has no progress in the last days/weeks, etc, where other people don't work. Or they will ask on the forum(s) about. And some TF assignments only take minutes on a GPU.

Please DON'T do TF on a CPU. As other people told you, it will be a waste, it takes ages; a GPU is hundreds of times faster at this kind of job (unless you have a "xeon phi" or something).

Also, another tip: if you DO have an assignment for an exponent to some bitlevel b, and you want to extend the assignment to bitlevel b+1 or b+2, etc (by the way, both cases in your initial post are OK, and are syntax equivalent) then DO NOT report the intermediary results until you finish the work with the respective exponent. For example, you reserve x to 72 bits (or reserve it for any other type of work, by the way, to avoid other people working with it), then you want to extend the work to 73, then do NOT report the TF to 72 when is done. Wait until 73 is done, and report both together. That is because, if you report the 72, the server knows that you did the work it was assigned to you, and it will remove your assignment, because it doesn't know your intention to go to 73. The server doesn't read your mind. So, you may have the surprise that, during you work that x to 73, somebody else gets it assigned and does the same work, wasting his resources. Or, if that somebody is me, I am faster than you, I will report the 73 before you, and then you will get an error "result not needed" when you report your 73, therefore wasting YOUR resources. :razz:

If you use Misfit, there is an option (which is default) to report the TF assignments only after all bitlevels are done. For example, taking exponent x from 72 to 75 will only report a single time, at the end. Of course, you get credit for all bitlevels, and it will show "no factors below 75", to answer your question. And that is regardless of the fact that you report each bitlevel, or you report only once at the end (i.e. you will get the same amount of total credit).

Again, be aware that bitlevel b+1 takes double time compared to bitlevel b. So, a "72 to 76", will take 1+2+4+8=[U][B]15 times[/B][/U] longer than a "72 to 73". That is why we switch to P-1 after a while, because TF becomes exponentially harder, and in a certain point is more efficient to find factors by P-1.

birtwistlecaleb 2021-06-25 07:33

Thanks for letting me know that TF is now off limits. Is P-1 bad too?


All times are UTC. The time now is 07:26.

Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2021, Jelsoft Enterprises Ltd.