Message boards :
News :
Padls Total subproject running fine
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
Tomáš Brada Project administrator Volunteer developer ![]() Send message Joined: 3 Feb 19 Deprecated: Creation of dynamic property BoincUser::$nposts is deprecated in /var/boincadm/prj/html/inc/forum.inc on line 613 Posts: 667 Credit: 432,784 RAC: 0 ![]() ![]() |
In all, just 9 CF ODLS has been found. It is right? Are all new solutions already in the file? It appears that the workunits are now being assigned from the unproductive segments (segments not containing any/very little CF). |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Deprecated: Creation of dynamic property BoincUser::$nposts is deprecated in /var/boincadm/prj/html/inc/forum.inc on line 613 Posts: 420 Credit: 0 RAC: 0 |
In all, just 9 CF ODLS has been found. It is right? Are all new solutions already in the file? This is bad. I told you about anchor points. Anchor points exclude huge empty segments. I did a search for such points, see https://boinc.progger.info/odlk/forum_thread.php?id=115 I offered you the found anchor points. You answered that you can find all the anchor points very quickly. So find it! Project participants should not count billions of SN DLS, among which there is not a single CF! This will require a lot of time and will not give a single solution. This is not very rational. You need to throw away all the huge empty segments. |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
Have you checked the start segment in rule 51? start 0 2 3 4 5 6 7 8 9 1 3 1 0 6 7 4 5 9 2 8 1 0 2 7 6 8 9 3 4 5 2 5 1 3 8 9 4 0 6 7 5 3 7 9 4 0 8 2 1 6 6 4 8 0 9 5 2 1 7 3 7 8 9 5 0 1 6 4 3 2 4 9 6 8 1 2 3 7 5 0 9 7 5 1 2 3 0 6 8 4 8 6 4 2 3 7 1 5 0 9 end 0 2 3 4 5 6 7 8 9 1 3 1 0 6 7 4 5 9 2 8 1 0 2 9 6 7 8 3 4 5 7 6 9 3 8 1 4 5 0 2 9 5 8 1 4 0 3 2 6 7 6 3 1 8 9 5 2 4 7 0 2 9 7 5 1 8 6 0 3 4 4 8 6 0 2 9 1 7 5 3 5 7 4 2 0 3 9 1 8 6 8 4 5 7 3 2 0 6 1 9 I see in this segment only one of the solutions I found. I found 240 solutions in this segment, but I did not check the entire segment, approximately 10 million CF. This segment contains 93562426 CF. Here all SN DLS are CF. Why do not you have solutions from this segment? This is the very first segment! See my solutions in this segment (240 solutions) https://cloud.mail.ru/public/3Bnv/5FYgynNeC |
Tomáš Brada Project administrator Volunteer developer ![]() Send message Joined: 3 Feb 19 Posts: 667 Credit: 432,784 RAC: 0 ![]() ![]() |
Ok. I will add prioritization to tasks. We will run tasks from the first segments with priority. Later, I will look into the topic of anchor points. Is it known in advance which segments have no CF in them? |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
Ok. I will add prioritization to tasks. We will run tasks from the first segments with priority. Priority should only be given to segments containing a lot of CF. Do not include segments without CF in tasks! Later, I will look into the topic of anchor points. Is it known in advance which segments have no CF in them? No, it is not known in advance which segments do not contain CF. Therefore, I conducted research in the subject. See https://boinc.progger.info/odlk/forum_thread.php?id=115 1. Divide the ruler into groups. In rule 51 there are 6204 groups. In each group, the first and last SN DLS are known. 2. Each group is an interval having a start and end SN DLS. Examine each group for CFs. You can apply, for example, the multiple division method in half. 3. Anchor points are CFs that start segments containing a lot of CF. 4. Oriente on CFs ODLS from rule 51 found in the project. Many such CFs ODLS are found in the ODLK project. I oriented on these CFs ODLS. |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
Try contacting Belyshev. Maybe he knows how to quickly find segments without CF. I wrote to you in PM where you can find Belyshev. If you still have the database that you found here https://boinc.progger.info/odlk/forum_thread.phpid=102&postid=3034#3034 6088213 cf odlk total in database select all CF in rule 51 from it. It will be useful for you. |
Tomáš Brada Project administrator Volunteer developer ![]() Send message Joined: 3 Feb 19 Posts: 667 Credit: 432,784 RAC: 0 ![]() ![]() |
2. Each group is an interval having a start and end SN DLS. That is essentially what the current application is doing. However it still takes time to examine the group (in my case segment) for CFs. |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
Use the recommendation If you still have the database that you found here All CFs ODLS are in good segments. Consider these segments first. |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
Only 54 solutions were found per day. The project is idling! I suggest stopping the generator by Belyshev for now. This is a generation problem of CF that you need to solve. This problem should not be resolved on client computers! You said that you can define anchor points in a few minutes. And you were mistaken! I propose now to run my generator 3 for rule 51. It works similarly to the generators 1 and 2 that you know. But generator 3 produces a lot more pseudo-associative DLS. Let me remind you: the generator 3 produces DLS, not CF. To get CF, you need to use the program kanonizator_dlk_1.03 You did this in previous experiments. Do you agree? |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
Generator 3 testing here https://boinc.progger.info/odlk/forum_thread.php?id=105 Pay attention to it I have a problem with generation. I can not generate all the DLSs for a given row, because there are very many of them. |
Tomáš Brada Project administrator Volunteer developer ![]() Send message Joined: 3 Feb 19 Posts: 667 Credit: 432,784 RAC: 0 ![]() ![]() |
|
Tomáš Brada Project administrator Volunteer developer ![]() Send message Joined: 3 Feb 19 Posts: 667 Credit: 432,784 RAC: 0 ![]() ![]() |
Workunit tot5_51c_SzXF8oCz1GnqMzpC96MUcemFg was stuck due to too many errors (actually, weird project aborts) and the segment was therefore stuck too. I resurrected the workunit. Next, workunit tot5_51a_T72KJVr2ZHr1b8gfi6roQQfmq is also marked with too many failures (caused by the server outage and missing file), now this workuint is not associated to any segment in database, but it appears to belong to a segment that is disabled. So I will delete this workunit entirely and arrange to be regenerated from segment. |
Natalia Makarova Project scientist ![]() Send message Joined: 8 Feb 19 Posts: 420 Credit: 0 RAC: 0 |
What is this batch? 21 tot5-i 20101 17 0 21 1 It seems to me that these tasks are not calculate at all. I was wrong? |
Tomáš Brada Project administrator Volunteer developer ![]() Send message Joined: 3 Feb 19 Posts: 667 Credit: 432,784 RAC: 0 ![]() ![]() |
Batch 22 "tot5-i": Those are PADLS Total tasks from segments that have not been started yet. They are not sent, because priority is too low. Maybe I could stir up the priority. |
©2024 Tomáš Brada