Manually reserving work?
 

Hi all,

Is it possible to manually reserved work through Primenet?

I'm doing some low exponent TF work that is Primenet will not assign, so wanted to find out if there's at least a way to submit my worktodo to the server so those bit levels are reserved--any way to do this?

Thanks!

What sort of numbers and bit levels?

If you stick the tasks into your worktodo.txt, then do a manual communication with Primenet and check the box that indicates "Send expected completion dates to server.", Primenet will give you an assignment key [if it will acknowledge it at all].
This is the standard way to "self-assign" work.

TF on exponents below about 10 or 20 million is a waste of time -- do ECM instead. Primenet discourages this TF work by not registering such assignments.

You are free to TF anyway and the server should accept any results you submit.

[QUOTE=Prime95;524743]TF on exponents below about 10 or 20 million is a waste of time -- do ECM instead. Primenet discourages this TF work by not registering such assignments.

You are free to TF anyway and the server should accept any results you submit.[/QUOTE]

I'm doing the TF work in the ~10M range on GPUs mostly with mfakto. Is there a GPU based ECM program?

I'm very unfamiliar with doing ECM, and it seems to require more thought/manual input--is this true?

Others can tell you your best AMD GPU options. TF above 50M? P-1 in the 10M - 30M area? I don't think ECM is a good GPU choice. Others can comment more accurately.

[QUOTE=mnd9;524747]I'm doing the TF work in the ~10M range on GPUs mostly with mfakto. Is there a GPU based ECM program? [/QUOTE]None I've been able to find above a few kbits, so not suitable for what you intend, no. Prime95/mprime will do it, on cpus. Uncwilly's post presumes using them. ECM in general is not effective above 20M, so not suitable for work on the current wavefronts, ~86M first test, ~48M doublecheck.
See also the bottom of the attachment at [url]https://www.mersenneforum.org/showpost.php?p=488291&postcount=2[/url]

Thanks all for the input.

The reason I'm doing TF work in this low range is

1) my GPUs get almost double the throughput at this low bit level!
2) try and find a few more factors for low exponents to run PRP tests on Mersenne cofactors in prime95 (I'm doing a lot of this these days).

You can do TF as the PRP cofactor project is worthwhile and fun. Remember these guidelines as you select your range:

1) TJAOI has found all factors below 2^66.
2) The more ECM that has already been done on an exponent, the less likely you are to find a factor.
3) As exponents increase, ECM gets harder while TF gets easier. I'm not sure where the crossover point is.

[QUOTE=Prime95;524756]3) As exponents increase, ECM gets harder while TF gets easier. I'm not sure where the crossover point is.[/QUOTE]
Mersenne.ca can give you some idea of where that might be:
[url]https://www.mersenne.ca/exponent/11111119[/url]

[QUOTE=Prime95;524756]You can do TF as the PRP cofactor project is worthwhile and fun. Remember these guidelines as you select your range:

1) TJAOI has found all factors below 2^66.
2) The more ECM that has already been done on an exponent, the less likely you are to find a factor.
3) As exponents increase, ECM gets harder while TF gets easier. I'm not sure where the crossover point is.[/QUOTE]

Just to clarify regarding point #1, TJAOI has found ALL factors below 66 (not just first factors)? I see a lot of exponents with small known factors that don’t have much other TF data officially recorded, but its safe to assume TFing up to 66 has been done for all exponents (even those with a factor already factored at a lower bit level?) and is a waste of time...?

Take this exponent for example:
[url]https://www.mersenne.org/report_exponent/?exp_lo=2100011&full=1[/url]

Which has a small known factor but no TF data... would any TFing below 66 still be pointless?

[QUOTE=mnd9;525749]Take this exponent for example:
[url]https://www.mersenne.org/report_exponent/?exp_lo=2100011&full=1[/url]

Which has a small known factor but no TF data... would any TFing below 66 still be pointless?[/QUOTE]

Yes. Especially since ECM is much better for such small exponents.