ODLK2 Server hosting PADLS experiment

Message boards : Science : ODLK2 Server hosting PADLS experiment
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 . . . 9 · Next

AuthorMessage
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3112 - Posted: 15 Feb 2019, 21:18:50 UTC - in response to Message 3102.  

would not a quorum reset result in lost solutions?
For example, the user XAVER has the following WU 2878 3000. This WU has been processed and is awaiting checking.


Quorum of more than 1 does not cause loss of solutions. Quite the opposite. No quorum means the work will be processed two times faster, but can cause a very improbable error to go undetected. The first couple tasks were run with quorum 2, or even 3 to check that the app is working. Now all tasks are performed with no quorum requirement.

Natalia Makarova, please do not worry about "awaiting checking" status. All results that are submitted are immediately available on the server.
ID: 3112 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3113 - Posted: 15 Feb 2019, 21:33:04 UTC

Some days ago, I changed the boinc app to process split workunits. Why: I could not implement checkpoints, so every time you turn off the computer, or boinc, the progress of workunits in progress would reset to zero. Current tasks are split to 8 sub-tasks. Each running about 20 minutes instead of 3 hours. This means that if computer would shuts down, less work will be lost. Also partial results are coming in more quickly. This works quite good. If there are problems, write.

New user, "petri", registered with computer running 32-bit Windows XP. He can not contribute, because there are only 64 bit applications. Application for him could be built, but his processor is old and slow so it is not quite worth it.
ID: 3113 · Rating: 0 · rate: Rate + / Rate - Report as offensive
purplecfh

Send message
Joined: 12 Jan 18
Posts: 3
Credit: 1,202,491
RAC: 0
Message 3114 - Posted: 16 Feb 2019, 1:08:46 UTC - in response to Message 3113.  
Last modified: 16 Feb 2019, 1:12:42 UTC

I just added 92 threads to the project that should pick up some work when more is available.

I also invited some team members to come crash the party, and will help in trying to spread the word.
ID: 3114 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3115 - Posted: 16 Feb 2019, 4:22:58 UTC
Last modified: 16 Feb 2019, 4:54:24 UTC

Tomas Brada
tasks are over?

Tasks ready to send 0

https://boinc.tbrada.eu/server_status.php

I have many more tasks :)

In addition, I developed a complete version of the PADLS experiment.
ID: 3115 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3116 - Posted: 16 Feb 2019, 4:28:38 UTC - in response to Message 3114.  

I just added 92 threads to the project that should pick up some work when more is available.

I also invited some team members to come crash the party, and will help in trying to spread the word.

O! It is wonderful!
I hope the work will come soon.
ID: 3116 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3117 - Posted: 16 Feb 2019, 8:25:06 UTC

I did not expect the tasks to drain so fast. I did not load them all, because changing loaded tasks is more difficult than changing new tasks. Now all tasks are loaded.
ID: 3117 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3119 - Posted: 16 Feb 2019, 17:24:10 UTC

Tasks ready to send 0

https://boinc.tbrada.eu/server_status.php

Tomas Brada
do you want to get new tasks for the BOINC project?
Or do you want to finish processing the first batch of WUs?
ID: 3119 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3120 - Posted: 16 Feb 2019, 18:50:19 UTC - in response to Message 3119.  

Send new batch to email. They will run in parallel. With 30 users, all workunits are already sent.
ID: 3120 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3121 - Posted: 16 Feb 2019, 19:29:12 UTC - in response to Message 3120.  

Send new batch to email. They will run in parallel.

I e-mailed you new WUs.
ID: 3121 · Rating: 0 · rate: Rate + / Rate - Report as offensive
purplecfh

Send message
Joined: 12 Jan 18
Posts: 3
Credit: 1,202,491
RAC: 0
Message 3122 - Posted: 16 Feb 2019, 23:03:51 UTC - in response to Message 3120.  

Send new batch to email. They will run in parallel. With 30 users, all workunits are already sent.

Looks like the early adopter folks are starting to filter in!
ID: 3122 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Coleslaw
Avatar

Send message
Joined: 24 May 17
Posts: 3
Credit: 7,024,978
RAC: 12,734
Message 3123 - Posted: 17 Feb 2019, 3:57:45 UTC - in response to Message 3122.  

He could also try turning on BOINC-wide teams so that it auto creates many accounts automatically like many other projects do.
ID: 3123 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3124 - Posted: 17 Feb 2019, 6:29:03 UTC

An error was detected in the second portion of the WUs.
See
https://boinc.tbrada.eu/forum_thread.php?id=3005&postid=3033#3033

Tomas Brada
you must remove from the file Rows_part2_8321.txt all rows starting with the number 8.
ID: 3124 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3125 - Posted: 17 Feb 2019, 7:01:15 UTC

Tomas Brada
I deleted the wrong rows and send you a new batch of WUs.
ID: 3125 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3126 - Posted: 17 Feb 2019, 7:46:46 UTC

Tomas Brada
Excuse me.
Another error was found: the rows starting with the number 7 are also incorrect.
I made a gross mistake in the rows generator.

Please delete all rows starting with the number 7.
ID: 3126 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3127 - Posted: 17 Feb 2019, 9:16:18 UTC

Tasks for rows starting with 7 or 8 were aborted. It is interesting that the program crashed! The crash was in poisk_blk_trans library (part of family_mar subprogram).
ID: 3127 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3128 - Posted: 17 Feb 2019, 9:26:52 UTC
Last modified: 17 Feb 2019, 9:32:43 UTC

I am wondering how the kanonizator_dlk_1.03 program missed wrong DLSs???

The failure was to occur at the stage of canonizing DLSs.

[code]7.8.3

(unknown error) - exit code -1073741819 (0xc0000005)

PADLS Experiment. 8 1 4 6 9 7 2 5
ASS_DLK10A: 12496 in 0.625 s
PSEVDOASS_DLK_NEW: 0 in 3.405 s
KF DLK: 10304
Split 1/9 Offset 0 size 1144 t 0
ID: 3128 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3129 - Posted: 17 Feb 2019, 9:43:59 UTC
Last modified: 17 Feb 2019, 9:48:23 UTC

Tomas Brada
WUs with this generation option has ended.

Next, you need to slightly change both generators.
Is it hard to do?
I can send the source code of the modified generators.

I am preparing the third generator, writing a program.

But first I would like to see the test results.
ID: 3129 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3130 - Posted: 17 Feb 2019, 10:59:36 UTC - in response to Message 3128.  
Last modified: 17 Feb 2019, 10:59:59 UTC

I am wondering how the kanonizator_dlk_1.03 program missed wrong DLSs???

The failure was to occur at the stage of canonizing DLSs.
 PADLS Experiment. 8 1 4 6 9 7 2 5
 ASS_DLK10A: 12496 in 0.625 s
 PSEVDOASS_DLK_NEW: 0 in 3.405 s
 [color=red]KF DLK: 10304[/color]
 Split 1/9 Offset 0 size 1144 t 0


I run this row in debug environment: Second generator produces no squares, but that is obvious. The first generator produces squares, but all of them evaluate to false in is_dlk() function and kanonizator_dlk says "Нет ДЛК в файле tmp.txt". removing that check from the program results:
Канонизатор ДЛК10

Введено ДЛК:            12496
Время загрузки:         0.02855 сек
Найдено КФ:             10304
Время поиска:           0.215923 сек
КФ записаны в файл:     tmp2.txt
Время записи:           0.012689 сек
Общее время работы:     0.257326 сек


It seems the generator produces squares that some of them have the same canonical form. Also The "is_dlk()" check is not present in boinc application, so it has same number of squares as this modified kanonizator.
ID: 3130 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Tomas Brada

Send message
Joined: 14 Jan 19
Posts: 119
Credit: 574
RAC: 0
Message 3131 - Posted: 17 Feb 2019, 11:19:13 UTC - in response to Message 3129.  

WUs with this generation option has ended.

Next, you need to slightly change both generators.
Is it hard to do?
I can send the source code of the modified generators.

I am preparing the third generator, writing a program.

But first I would like to see the test results.


I have trouble resubmitting the rows from the new part 2 file. So I will submit them as batch 3 (13) when I have more time. The problematic ones were cancelled, but the corrected ones were not loaded yet. Changing the generators is not hard, if you send me the basic source code. However it will have to wait, because I have other things to do. I will sure send you the results, but batch 1 (10) is not finished yet and it is not practical to find which results are done and which are still in progress. About 165 still in progress.
ID: 3131 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Profile Natalia Makarova
Project scientist
Avatar

Send message
Joined: 6 Apr 17
Posts: 13606
Credit: 0
RAC: 0
Message 3132 - Posted: 17 Feb 2019, 11:24:34 UTC - in response to Message 3130.  

The generator produces incorrect DLS, for example

 0  8  1  4  6  9  7  2  5  8 
 3  1  0  6  5  8  9  4  2  7 
 4  5  2  9  7  1  8  3  6  0 
 5  6  8  3  2  6  4  0  9  2 
 6  9  3  8  4  0  2  5  7  1 
 8  2  4  7  9  5  1  6  0  3 
 7  0  9  5  3  7  6  1  3  4 
 9  3  6  1  8  2  0  7  4  5 
 2  7  5  0  1  4  3  9  8  6 
 1  4  7  2  0  3  5  8  1  9
ID: 3132 · Rating: 0 · rate: Rate + / Rate - Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 . . . 9 · Next

Message boards : Science : ODLK2 Server hosting PADLS experiment


©2024 (C) Progger