Manual results check-in is temporarily broken

GP2 · 2003-12-07, 17:52

Manual results reporting doesn't work... and the e-mail address for reporting bugs is still an entropia.com address

Quote:

PrimeNet Manual Testing Transactions

The name specified is not recognized as an
internal or external command, operable program or batch file.

Quote:

PrimeNet Manual Testing Transactions

Error 03 Results GP2.tmp not found!

Thank you for using GIMPS IPS manual testing forms.

Any transactions shown above are effective immediately.

Back to IPS Manual Tests Page

If a problem is indicated, and after verifying your account information the problem persists, please contact primenet@entropia.com,or the webmaster at the e-mail address below.

Old man PrimeNet · 2003-12-08, 01:38

It's not broken - it has been turned it off. Mining the server for off-net work or to submit garbage results is not the intended use for these forms. It will be turned on again later.

Complex33 · 2003-12-08, 02:41

Can you elaborate as to its misuse?

GP2 · 2003-12-08, 03:30

I tried to re-report the known factor for 10691533... unfortunately, someone is doing an LL test on it.

nitro · 2003-12-08, 08:17

Quote:

Originally posted by Old man PrimeNet
It's not broken - it has been turned it off. Mining the server for off-net work or to submit garbage results is not the intended use for these forms. It will be turned on again later.

How about turning it back on *now*. As a member of LMH I use this regularly to check in *large* results files, of the order of several thousands of lines.

Why? it saves George having to process these files personally.

Quite how this can be used for data mining I'm not sure.

If needs be then only accept results from certain userids.

apocalypse · 2003-12-08, 23:42

Scott -

I too have used the manual testing page to submit LMH factoring results, and would appreciate a more detailed explanation of how this is being abused, if you feel so inclined.

Pretty please?

With a dancing banana on top?

Complex33 · 2003-12-09, 00:03

Looks like we can't release exponents either. I had been using this to remotely remove P-1 factored exponents from a machine with 4 gigs of ram so it could pickup non-P-1 factored exponents... Complex miss manual forms

nitro · 2003-12-09, 08:56

Quote:

Originally posted by Old man PrimeNet
It's not broken - it has been turned it off. Mining the server for off-net work or to submit garbage results is not the intended use for these forms. It will be turned on again later.

I have a 909kb results file waiting to be checked in.

I'm afraid that until this facility is available again, then both the pc's doing LMH factoring get to be turned off.

Talk about the project shooting itself in the foot.

Prime Monster · 2003-12-09, 09:25

Oi,

The answer, as given, basically says that the function (and server) was was not designed for this type of use.

Maybe we need a feature in V5 that support this. If that is the case, then you must raise that question with the project team.

We, the participants, should not expect that we can do anything we like with the server. It was designed to support certain operations and not others, even though we can do them.

Rather than get annoyed, why don't you ask what the effect your usage has on the server and what effect it has on all other users.

PM

apocalypse · 2003-12-09, 13:27

Monster -

I do not deny that what you have stated is a valid reason to take down the Manual Testing pages.

However, since none of us believe ourselves to be 'submitting garbage results' and because no one interpreted 'Mining the server for off-net work' as 'submitting results which were not assigned by Primenet', we did not realize we were the responsible parties.

Could you perhaps explain what the server's limitations are in this regard?

Prime Monster · 2003-12-09, 14:13

I did not design this server, but I have designed a few other servers in my time. Based on experience from those, and Scotts comments, I would guess that the server has some limitations in that area. What you see as perfectly reasonable might not be.

To be more specific: With modern day cpus you can manage to do a large amount of work compared to earlier times, you are also, AFAIK, selecting to do specific types of work, again submitting large amounts of "completed" work.

I guess the system was primarily designed to send and receive results from clients, basically small data-units. Suddenly getting large data-units could potentially do harm to the system, or cause bottle-necks or other problems. Maybe the update procedure is not good enough to handle all the triggered updates that your results are generating. Maybe the scanning the system for "suitable" work-units cause problems.

Designing a system you try to take every possibility into account, but you seldom manage to get all of them. That's life. Since Scott has not been more specific about why he has shut down this function, I can only make guesses as to the cause.

Note that I did not claim you were submitting garbage results, but that I guessed that the server was not designed to handle the way you were using it.

It could also be that someone, other than yourself and the others that do specialised work, are not the "guilty" party, but someone else that is doing large scale data-mining.

You will have to wait for Scott to tells us the specific reason for turning off the function.

PM

Prime95 · 2003-12-09, 15:03

In the meantime, email LMH results here. They are a breeze to process.

P-1 results are not as easy because what you really want is the data on the server updated too. LMH exponents are not managed by the server.

Pi Rho · 2003-12-09, 15:16

Would it work to only manually report factors from P-1 tests? Or have a specific account (e.g. Mersennaries) that P-1 work could be done? This would cut down on traffic to the server as well as the callous "not assigned" errors which are hurtful.

Keep track of the P-1 progress on the forum so work isn't duplicated and leave it at that. The factoring would still helps out the project without hindering the server.

One other thing, did the reported P-1 test results get updated in the appropriate place?

GP2 · 2003-12-09, 16:37

Quote:

Originally posted by Pi Rho
One other thing, did the reported P-1 test results get updated in the appropriate place?

When a new set of "weekly" data files is released, I verify that the P-1 factors reported in the Mersenne-aries subforum have made it into the data files. These get moved to the "Completed Missions" sub-subforum.

tom11784 · 2003-12-09, 16:38

Quote:

Originally posted by Prime95
In the meantime, email LMH results here. They are a breeze to process.

I have 2 files from two comps which are relatively small (should be between 500 and 1000 results each).
Where exactly do we email them?
I didn't think that these forums supported email accounts...

-tom11784

GP2 · 2003-12-09, 16:53

I don't know how abusing the manual reporting feature would be useful for datamining. If someone's looking for exponents to test manually, George usually assigns those. Whatever they're doing could probably be accomplished in a less brute-force way just by looking at the data files and status and cleared.

Since a user id and password must be entered to use the manual forms, presumably associated with a valid e-mail address, maybe we could just ask them what they're trying to do.

Prime95 · 2003-12-10, 02:00

My email addr is woltman@alum.mit.edu

garo · 2003-12-10, 04:51

Yeah folks! Just email the results to George as he says. I never use the manual pages and always email 'em. You may want to zip the results file if you want to conserve bandwidth. I never asked Groge if unzipping is a hassle for him and if the results files should all the called results.txt or have unique names to avoid confusion.

But he has never asked me to do otherwise so I've just emailed stuff to him.

S00113 · 2003-12-10, 07:33

When will you turn the manual forms back on again?

I'm testing a SuperDome with 64 Itanum 2's, and small cluster of dual G5's, and need more assignments and relief of results. And I need to extend the time for five exponents beeing tested on "old" G4's within 15 days.

TauCeti · 2003-12-10, 16:45

Quote:

Originally posted by S00113
I'm testing a SuperDome with 64 Itanum 2's

Wow! Awesome! How about trying a multithreaded glucas on the SuperDome?

To quote gbvalor:

Quote:

On the verifier M40 machine SPE174, a 4-way multithreaded ran at 0.024 sec/iter. The two-way verifier run did it at 0.046 sec/iter

It would be most interesting, what iteration times M40 would produce on the Dome.

I wish i could afford one, but cannot spare the 8 millon $

gbvalor · 2003-12-10, 22:06

Quote:

It would be most interesting, what iteration times M40 would produce on the Dome.

Indeed. Have you tried Glucas on that monster?.

On the other hand. Is Dual G5 really a two processors like chip?. Have you tried a two-way threaded Glucas on it?

Guillermo

gbvalor · 2003-12-10, 22:15

Quote:

It's not broken - it has been turned it off. Mining the server for off-net work or to submit garbage results is not the intended use for these forms. It will be turned on again later.

That shows the reason of why we need a server for nonprime95 clients as soon as possible.

We must to begin the work now!

Guillermo.

Old man PrimeNet · 2003-12-10, 22:33

The form was back in service this morning about 8am PDT. The new v5 server has a design plan to handle sneakernetting more carefully.

Complex33 · 2003-12-11, 00:05

Thanks OMPNet! Lot's 'O LMH results just sent

ColdFury · 2003-12-11, 02:25

Quote:

The form was back in service this morning about 8am PDT. The new v5 server has a design plan to handle sneakernetting more carefully.

Was the afore-mentioned trouble something we caused, or just some trouble maker fooling around?

2003-12-08, 01:38	#2
Old man PrimeNet Jan 2003 Altitude>12,500 MSL 101 Posts	It's not broken - it has been turned it off. Mining the server for off-net work or to submit garbage results is not the intended use for these forms. It will be turned on again later. Last fiddled with by Old man PrimeNet on 2003-12-08 at 01:47

2003-12-09, 16:53	#16
GP2 Sep 2003 5×11×47 Posts	I don't know how abusing the manual reporting feature would be useful for datamining. If someone's looking for exponents to test manually, George usually assigns those. Whatever they're doing could probably be accomplished in a less brute-force way just by looking at the data files and status and cleared. Since a user id and password must be entered to use the manual forms, presumably associated with a valid e-mail address, maybe we could just ask them what they're trying to do. Last fiddled with by GP2 on 2003-12-09 at 16:53

2003-12-10, 07:33	#19
S00113 Dec 2003 D8₁₆ Posts	Time estimate When will you turn the manual forms back on again? I'm testing a SuperDome with 64 Itanum 2's, and small cluster of dual G5's, and need more assignments and relief of results. And I need to extend the time for five exponents beeing tested on "old" G4's within 15 days.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
GPU id/name for manual results	preda	GPU Computing	15	2017-08-16 17:34
mersenne.ca and manual results	Gordon	mersenne.ca	3	2015-08-31 03:08
manual results	ramgeis	PrimeNet	8	2013-05-30 06:33
Manual Check-in glitch	bayanne	Software	2	2003-12-10 13:38
what is the url for the manual check in	crash893	Lounge	5	2003-03-23 06:34

2003-12-08, 02:41	#3
Complex33 Aug 2002 Texas 5×31 Posts	Can you elaborate as to its misuse?

2003-12-08, 03:30	#4
GP2 Sep 2003 2585₁₀ Posts	I tried to re-report the known factor for 10691533... unfortunately, someone is doing an LL test on it.

2003-12-08, 23:42	#6
apocalypse Feb 2003 AE₁₆ Posts	Scott - I too have used the manual testing page to submit LMH factoring results, and would appreciate a more detailed explanation of how this is being abused, if you feel so inclined. Pretty please? With a dancing banana on top?

2003-12-09, 00:03	#7
Complex33 Aug 2002 Texas 10011011₂ Posts	Looks like we can't release exponents either. I had been using this to remotely remove P-1 factored exponents from a machine with 4 gigs of ram so it could pickup non-P-1 factored exponents... Complex miss manual forms

2003-12-09, 09:25	#9
Prime Monster Aug 2002 2²×5×13 Posts	Oi, The answer, as given, basically says that the function (and server) was was not designed for this type of use. Maybe we need a feature in V5 that support this. If that is the case, then you must raise that question with the project team. We, the participants, should not expect that we can do anything we like with the server. It was designed to support certain operations and not others, even though we can do them. Rather than get annoyed, why don't you ask what the effect your usage has on the server and what effect it has on all other users. PM

2003-12-09, 13:27	#10
apocalypse Feb 2003 2×3×29 Posts	Monster - I do not deny that what you have stated is a valid reason to take down the Manual Testing pages. However, since none of us believe ourselves to be 'submitting garbage results' and because no one interpreted 'Mining the server for off-net work' as 'submitting results which were not assigned by Primenet', we did not realize we were the responsible parties. Could you perhaps explain what the server's limitations are in this regard?

2003-12-09, 14:13	#11
Prime Monster Aug 2002 2²·5·13 Posts	I did not design this server, but I have designed a few other servers in my time. Based on experience from those, and Scotts comments, I would guess that the server has some limitations in that area. What you see as perfectly reasonable might not be. To be more specific: With modern day cpus you can manage to do a large amount of work compared to earlier times, you are also, AFAIK, selecting to do specific types of work, again submitting large amounts of "completed" work. I guess the system was primarily designed to send and receive results from clients, basically small data-units. Suddenly getting large data-units could potentially do harm to the system, or cause bottle-necks or other problems. Maybe the update procedure is not good enough to handle all the triggered updates that your results are generating. Maybe the scanning the system for "suitable" work-units cause problems. Designing a system you try to take every possibility into account, but you seldom manage to get all of them. That's life. Since Scott has not been more specific about why he has shut down this function, I can only make guesses as to the cause. Note that I did not claim you were submitting garbage results, but that I guessed that the server was not designed to handle the way you were using it. It could also be that someone, other than yourself and the others that do specialised work, are not the "guilty" party, but someone else that is doing large scale data-mining. You will have to wait for Scott to tells us the specific reason for turning off the function. PM

2003-12-09, 15:03	#12
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 19·397 Posts	In the meantime, email LMH results here. They are a breeze to process. P-1 results are not as easy because what you really want is the data on the server updated too. LMH exponents are not managed by the server.

2003-12-09, 15:16	#13
Pi Rho Sep 2002 44₁₀ Posts	Would it work to only manually report factors from P-1 tests? Or have a specific account (e.g. Mersennaries) that P-1 work could be done? This would cut down on traffic to the server as well as the callous "not assigned" errors which are hurtful. Keep track of the P-1 progress on the forum so work isn't duplicated and leave it at that. The factoring would still helps out the project without hindering the server. One other thing, did the reported P-1 test results get updated in the appropriate place?

2003-12-10, 02:00	#17
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 19·397 Posts	My email addr is woltman@alum.mit.edu

2003-12-10, 04:51	#18
garo Aug 2002 Termonfeckin, IE 5314₈ Posts	Yeah folks! Just email the results to George as he says. I never use the manual pages and always email 'em. You may want to zip the results file if you want to conserve bandwidth. I never asked Groge if unzipping is a hassle for him and if the results files should all the called results.txt or have unique names to avoid confusion. But he has never asked me to do otherwise so I've just emailed stuff to him.

2003-12-10, 22:33	#23
Old man PrimeNet Jan 2003 Altitude>12,500 MSL 101 Posts	The form was back in service this morning about 8am PDT. The new v5 server has a design plan to handle sneakernetting more carefully.

2003-12-11, 00:05	#24
Complex33 Aug 2002 Texas 5·31 Posts	Thanks OMPNet! Lot's 'O LMH results just sent