A B.Cereus case of assignment poaching... - Page 2

tcharron · 2013-03-12, 21:04

86073649

chalsall · 2013-03-12, 21:16

Quote:

Originally Posted by tcharron

86073649

OK. This is what Primenet knows: 86073649.

tcharron · 2013-03-12, 22:19

There's no record there of the work assigned to me... why is that?
Here is some of my log...

Code:

[Comm thread Mar 10 22:15:10] Got assignment F3E357C79128A156A90F2B4F60FAC175: Trial factor M86073649
.
.
.
[Comm thread Mar 12 00:19:10] Sending expected completion date for M86073649: Mar 12 2013
.
.
.
[Worker #5 Mar 12 00:40:22] M86073649 no factor from 2^69 to 2^70, We8: 3A363CCB
[Comm thread Mar 12 00:40:22] Sending result to server: UID: timc995.system/dl580_g5, M86073649 no factor from 2^69 to 2^70, We8: 3A363CCB, AID: F3E357C79128A156A90F2B4F60FAC175
[Comm thread Mar 12 00:40:22] Result was not needed.  TF on M86073649, sf: 69, ef: 70

In any event, it would have been nice if primenet notified my client that this work was not needed when my client tried to update the expected completion date. Maybe that would be a useful enhancement to the server response? Only 20 minutes in this case, but maybe more in others...

chalsall · 2013-03-12, 23:48

Quote:

Originally Posted by tcharron

In any event, it would have been nice if primenet notified my client that this work was not needed when my client tried to update the expected completion date. Maybe that would be a useful enhancement to the server response? Only 20 minutes in this case, but maybe more in others...

You clearly were legitimately assigned this candidate to TF by Primenet.

Unfortunately, some work independently. This can cause duplication of efforts -- as you have just experienced.

Edit: And yes, it would be nice if Primenet notified your client that the scheduled work was not needed. Unfortunately that functionality doesn't exist at the moment.

Edit2: There is no record of you being assigned this because your client submitted the results (even though they weren't needed) and basically said "I'm done".

ixfd64 · 2013-03-13, 00:12

George could also just configure PrimeNet to accept duplicate TF results. It already accepts duplicate LL results, so I don't see any reason to reject TF double checks.

tcharron · 2013-03-13, 00:27

Quote:

Originally Posted by ixfd64

George could also just configure PrimeNet to accept duplicate TF results. It already accepts duplicate LL results, so I don't see any reason to reject TF double checks.

Sure, but I wasn't just suggesting an ehancement to deal with accurate crediting of work...
What I mean is that if the server would notify the clients that "what you are doing is pointless" (or is only now a double check), then the client could move on to more useful work...

chalsall · 2013-03-13, 00:47

Quote:

Originally Posted by tcharron

What I mean is that if the server would notify the clients that "what you are doing is pointless" (or is only now a double check), then the client could move on to more useful work...

This has often been requested....

ixfd64 · 2013-03-13, 01:27

Better yet would be the ability to control an individual client remotely, as opposed to changing the settings on PrimeNet and waiting for the client to contact the server. But implementing that would probably take a lot of work.

LaurV · 2013-03-13, 05:09

The server could do nothing about it. It takes like 20 minutes for a good GPU to do the 72-bit level for an 85M exponent. Sometime the poacher goes in and out after you get the assignment and the exponent is actually not assigned when you start to work on it, and as you don't contact the server in meantime, any action from the server would be futile. The server can't contact you by itself, and that, if possible, would be VERY bothersome!

Some people work ranges without reserving them (which would be too complicate, this is what I did for 332M before GPU72 was able to offer them for official assignments, and before MISFIT could handle the fetching, I just worked them "unofficially", sometimes stepping on other people toes. But they knew

So, if you want to work ranges which are out of the "official", you can look at this page before, eventually click on "yesterday", "a week ago" and "a month ago" and see which ranges are mostly changing, those are "hot" ranges in which you risk to step on other people toes, or they on yours (read: futile duplication of work). And pick wisely.

Right now, 80M is a very hot range. Click on it and see subranges. Try to pick your own, if you want to do it manually, by yourself, and let the people know, so they can avoid you. There will always be some "non-conformists", which intentionally or by pure knowledge can bother you, but fortunately they are not many.

chalsall · 2013-03-13, 15:54

Quote:

Originally Posted by LaurV

The server could do nothing about it.

That's not actually true.

A couple of examples are 59197987 and 59343421. These have been out 1373 (!) and 514 days respectively. And each of these two users have six and eight such assignments (again, respectively).

People have gotten impatient, and all of these have been TFed to 73 bits, and P-1ed. Meanwhile, Primenet happily accepts "updated completion dates" for these, even though the work is no longer needed.

It would be simple for George / Scott to release these back into the pool for LL assignments. When the current client(s) next contacted the server it would be informed that the AID Assignment Key was invalid, and the assignments would be removed from the worktodo.txt file.

LaurV · 2013-03-14, 01:36

is it saying that someone is doing CPU TF to 71 for one year and half, for the same exponent, and he still has the reservation? Any progress?
If that's the case, I agree with you, there is no argument about...

2013-03-13, 01:27	#19
ixfd64 Bemusing Prompter "Danny" Dec 2002 California 2·5·239 Posts	a curious case of assignment poaching... Better yet would be the ability to control an individual client remotely, as opposed to changing the settings on PrimeNet and waiting for the client to contact the server. But implementing that would probably take a lot of work.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Poaching	Kevin	Lounge	123	2021-03-12 05:03
Poaching	blip	Data	8	2016-01-30 01:59
Poaching	davieddy	Lounge	6	2010-10-16 12:31
Poaching and v5	PrimeCruncher	PrimeNet	6	2004-04-05 19:17
New fashion poaching (???)	lycorn	Lounge	6	2003-01-31 08:33

2013-03-12, 21:04	#12
tcharron Jan 2013 2²·7 Posts	86073649

2013-03-13, 00:12	#16
ixfd64 Bemusing Prompter "Danny" Dec 2002 California 2·5·239 Posts	George could also just configure PrimeNet to accept duplicate TF results. It already accepts duplicate LL results, so I don't see any reason to reject TF double checks.

2013-03-13, 05:09	#20
LaurV Romulan Interpreter Jun 2011 Thailand 2·5·31² Posts	The server could do nothing about it. It takes like 20 minutes for a good GPU to do the 72-bit level for an 85M exponent. Sometime the poacher goes in and out after you get the assignment and the exponent is actually not assigned when you start to work on it, and as you don't contact the server in meantime, any action from the server would be futile. The server can't contact you by itself, and that, if possible, would be VERY bothersome! Some people work ranges without reserving them (which would be too complicate, this is what I did for 332M before GPU72 was able to offer them for official assignments, and before MISFIT could handle the fetching, I just worked them "unofficially", sometimes stepping on other people toes. But they knew So, if you want to work ranges which are out of the "official", you can look at this page before, eventually click on "yesterday", "a week ago" and "a month ago" and see which ranges are mostly changing, those are "hot" ranges in which you risk to step on other people toes, or they on yours (read: futile duplication of work). And pick wisely. Right now, 80M is a very hot range. Click on it and see subranges. Try to pick your own, if you want to do it manually, by yourself, and let the people know, so they can avoid you. There will always be some "non-conformists", which intentionally or by pure knowledge can bother you, but fortunately they are not many.

2013-03-14, 01:36	#22
LaurV Romulan Interpreter Jun 2011 Thailand 10010110001010₂ Posts	is it saying that someone is doing CPU TF to 71 for one year and half, for the same exponent, and he still has the reservation? Any progress? If that's the case, I agree with you, there is no argument about...