How to reserve very low TF work

[snme2pm1]

I have been spending part time to lift TF level on low exponent numbers.
Yes I know there is a great argument as to value, but plainly some folk are still making huge exertions trying to splinter off at least any factor, whether it's the lowest or not.
What I have been unable to discern; How is that "Sid & Andy" and no other account that I have observed has been able to reserve TF jobs in the extreme low region lately?
I have several times (from a moribund room heater with nothing better to do) found myself unexpectedly colliding with TJAOI lodgements, so for lack of apparent reservation system, opting to second guess such attention and shift region in an attempt to avoid collision with submissions attributed to that account.
Is there a reservation mechanism of some kind that I've overlooked?

[VictordeHolland]

It's always tricky to do work without a reservation, you have to accept that risk.
I've had contact over PM with Jayder and lycorn as to who is TF-ing what.

I'm currently TFing the 3M range to 65bits.
Jayder was doing 4M to 64bits and lycorn 64bits to 65bits, but that is almost finished.
monst and Nesowa were also doing TF in the <10M region, not sure what they are doing now.

I'm also running P-1 (B1=10e6, B2=200e6) on 1.5-2M expo's.

[snme2pm1]

Yes Vic, I've seen you in the area, but so far I think that we have not bumped.
Lately I've been working 1500k, and less seriously some a little 800k and some 90k.
I hopefully didn't much bump monst et. al. when working 5M a few weeks ago, but didn't find a response to queries about that potential hazard.
But what is this "Sid & Andy" business, How does anybody reverse some low exponents for TF like that?
http://www.mersenne.org/assignments/...et+Assignments

[tha]

Originally Posted by snme2pm1

How does anybody reverse some low exponents for TF like that?

During the v4 server days the Lone Mersenne Hunters subforum was created and used for announcing who was doing what. With the introduction of the v5 server more type of workloads became available, but everyone can still use the subforum for this purpose.

[chalsall]

Originally Posted by snme2pm1

How does anybody reverse some low exponents for TF like that?

Comfortable with Perl? This brought back fond memories.

[TheMawn]

Keep in mind TJAOI is only reporting factors by the looks of it, so they won't "clear" any bit levels so if you want the explicit "No Factors from X to Y" result you'll have to duplicate some of their work.

Also there isn't really any way to tell that TJAOI is actually hitting every single factor.

[LaurV]

Originally Posted by TheMawn

Keep in mind TJAOI is only reporting factors by the looks of it, so they won't "clear" any bit levels so if you want the explicit "No Factors from X to Y" result you'll have to duplicate some of their work.

Also there isn't really any way to tell that TJAOI is actually hitting every single factor.

They did hit every single factor below ~54 bits (this we can confirm), doing "by k" search (we also did that in the past, together with James, see his graphs). They started "by p" checking for higher than 56 bits (57??), bus still doing mixed work in between. You could see from their reports. At the end, they will hit every single factor below whatever their limit would be when they will decide to stop. Which works for us. Our appreciation for their work raised considerable in the last time (remember I was very skeptical in the beginning, but I didn't know how persistent they were going to be! And when they started turning in "missed" factors, actually saving lots of LL work, well... hats off!)

[snme2pm1]

Originally Posted by TheMawn

Keep in mind TJAOI is only reporting factors by the looks of it, so they won't "clear" any bit levels so if you want the explicit "No Factors from X to Y" result you'll have to duplicate some of their work.

Apparently TJAOI is doing more than one kind of thing.
A contrasting example voiding your statement is: http://www.mersenne.org/report_expon...=899893&full=1
A no factor result over a range was asserted, and PrimeNet recorded such claim.

Example http://www.mersenne.org/report_expon...=244639&full=1
shows ECM result claims as well as an apparent search by k result.

[snme2pm1]

Originally Posted by chalsall

Comfortable with Perl? This brought back fond memories.

Yes, Perl can be handy for lightweight jobs.
However I don't spend time in the Linux world, which seems to be where mprime operates.
The hint from briefly considering those ideas seems to be that even though the PrimeNet web browser user friendly manual reservation system excludes some ranges, a more direct appeal to the underlying Web API might be more fruitful.
If I can muster the energy to do that!

[LaurV]

Perl runs well in windoze too, I am not a linux guy but yet I am good in playing perl under windoze..
Many things I couldn't do without it...

[lycorn]

Originally Posted by snme2pm1

How does anybody reverse some low exponents for TF like that?

I don´t think there is much value in actually reserving exponents in such ranges, where very few people are working. There is no great chance of duplicating work. If were you, I would just go and pick a couple of exponents, do my thing, and check them in upon finishing the task. That´s in fact what I do, whenever I feel like testing low expos. I never reserve too many at a time, to further reduce the chances of collisions.
Keep in mind that in order to use a GPU to TF exponents < 1M you will need a special purpose version of mfaktc, developed by TheJudger some time ago, that is much less effective then the current version. It´s still faster than using a CPU, though.

HTH. Happy crunching!