Dear participants!

Ask if a new application is planned in the project?

Is there going to be a new app for this project?

See
https://boinc.progger.info/odlk/forum_thread.php?id=82&postid=2093#2093

I have many new algorithms, new experiments and, accordingly, new applications.
But the project administrators do not want to launch new experiments.
I can not start a new application in a BOINC project.
Therefore, I perform numerous experiments on my PC.

I can offer you an Individual application.
It's very simple.
For example, a batch file works

@echo off
LS10Assoc.exe < vvod.txt
copy output.txt input.txt
kanonizator_lk.exe < vvod1.txt
copy output.txt input.txt
copy assocLS10Last.txt assocLS10Last_2476.txt
family_mar.exe

All programs are ready and working.

The program LS10Assoc.exe is the generation of associative LSs. Author H. White.
The program kanonizator_lk.exe - canonization of LSs. Author A. Belyshev.
The program family_mar.exe - checking LSs on ODLS. Author A. Belyshev.

You can execute this file many times in a loop, then the calculation process will be continuous.
The experiment is very large.

At the moment I'm doing this experiment.
But there are many variants of this experiment - for different groups of associative LSs.
Everyone can get their own version of the experiment.

Does anybody want?
Please write me a personal message.

---

https://yadi.sk/i/S1HhJBwp3Pr9jX

There is another variant.

You can help the administrator ice00 run this experiment in BOINC project ODLK1.
Because ice00 complains about the lack of time for this work.

Please help us create a new application in the project.
It is very important!

The colleague informed me about the API (Application Programm Interface) in the BOINC project.

Who can tell us more about this?
Maybe links to information.

Thank you.

https://boinc.berkeley.edu/trac/wiki/BasicApi

They are the functions (in C++ language) to call from a program for communicating to the BOINC server: a programmer must use them to adapt (for example) an algorithm of solution to process a work-unit from server.

https://boinc.berkeley.edu/trac/wiki/BasicApi

They are the functions (in C++ language) to call from a program for communicating to the BOINC server: a programmer must use them to adapt (for example) an algorithm of solution to process a work-unit from server.

ice00
Thank you.

Can you use this function?

By the way, Belyshev has developed a new experiment.
Software can be downloaded from here.
http://forum.boinc.ru/default.aspx?g=posts&m=91313#post91313

This is a short experiment, totaling 52517 WU.

ice00
maybe you try to run this experiment on your server using the specified function?
I will give you the necessary advice on the experiment.
All source codes are available.

It is very interesting!

https://boinc.berkeley.edu/trac/wiki/BasicApi

They are the functions (in C++ language) to call from a program for communicating to the BOINC server: a programmer must use them to adapt (for example) an algorithm of solution to process a work-unit from server.

ice00
Thank you.

Can you use this function?

They are already used now in the odlk3 and odlkmax software. Using them in a new task is a consuming time job of programming for adapt the software to use them.

By the way, Belyshev has developed a new experiment.
Software can be downloaded from here.
http://forum.boinc.ru/default.aspx?g=posts&m=91313#post91313

This is a short experiment, totaling 52517 WU.

ice00
maybe you try to run this experiment on your server using the specified function?
I will give you the necessary advice on the experiment.
All source codes are available.

It is very interesting!

If the software is in C/C++ and the workunits are in a fixed amount I can compile it in an old server that I wake up during the holiday period for experiment and let it on for solving them.

If the software is in C/C++ and the workunits are in a fixed amount I can compile it in an old server that I wake up during the holiday period for experiment and let it on for solving them.

Yes, that's it.

Will you try?
I'll write you private message.

Thank you.

---

https://yadi.sk/i/S1HhJBwp3Pr9jX

Hello, ice00!

I write this message in public, because the question is very serious and needs to be resolved.
We are again at an impasse in dealing with this issue.

So, you can not connect the experiment, which was discussed above, to the BOINC project.
You say that you do not have time for this work.

You have done the software to perform an experiment on your server in manual mode (on a Linux system). That, of course, is good.
But I did not mean the execution of the experiment in manual mode.

Demis and I are already performing an experiment in a manual project.
In this case, you need to check those families that we do not check.
I check the family of names_44323_4696.txt
Demis checks the family of names_43423_26489.txt
You should check the family of names_44323p_21332.txt
Do you understand that you should not do the same work?

If you independently check all 52517 families, it will take about a year.
For 3 days, you checked a total of 623 families of 52517.
This speed of the experiment is unacceptable.

The experiment must be performed in the BOINC project!

My suggestions.

1. Openly contact Progger with a request to start an experiment in a BOINC project (ODLK or ODLK1).
I inform Progger that I, Demis, and you already have working software (from Belyshev) and perform the experiment in manual mode.

We are waiting for a response from the Progger.
I appeal to Progger publicly, because he does not respond to my letters. The reason for this is unknown to me.
I invite you to contact Progger also in a personal message.
Maybe he will answer you.

2. If Progger does not respond to our appeal, we will turn to the community for help.
Perhaps the community can help us.

An inadmissible situation in the presence of two active BOINC projects to perform experiments in manual mode.
This is nonsense!

I remind you that I have repeatedly discussed this issue with you and with Progger.
But the issue has not yet been resolved.

What suggestions do you have?

PS. In fact, you started an individual assignment.
But this is not what is required of you. You are the administrator and developer of the BOINC project ODLK1.
You can run the experiment in ODLK1.

ice00

I am quoting your personal message because I want the discussion here to keep the community informed of the problem.

So, you need to have

GENERATOR: simple send one name: AAAABDGRU (it uses 10 bytes, so all
space required is about 500KB). This program is to code by zero.

FINDER: it takes the workunit (AAAABDGRU), then generate the data as
done by actual generator and then test it. FINDER should be one
executable that use BOINC API and that must be the union of your actual
generator and family checker: one program from the two sources you have.

This is why it is needed time in coding the program for running in BOINC
and it is no so simple like takes your actual programs and use them.

I am not an expert on software adaptation to the BOINC platform. And I did not say that it is simple.
But I know that this is possible.
This experiment was successfully (and very quickly!) performed in the Gerasim@Home BOINC project.
All results are obtained.

I did not say either that it is necessary to immediately generate all the tasks and have 160 GB to record all the WUs.
You can do in manual mode. This is exactly what Demis does, but he generated all the tasks for one group of families names_43423_26489.txt
He got about 100 GB with a WUs record. Now it handles these WUs.

I perfectly understand the scenario of generating one task described by you and checking it.
But I can't write programs for the BOINC platform.
Can you do it?

You say that you do not have time for this work.
But performing the experiment in manual mode will take much longer!
And you begin to perform a manual experiment. Do you have time for this?

PS. Please reply here.
I want to publicly discuss the problem and understand why this problem cannot be solved for six months already.
Progger seems to have refused to solve this problem.

What do you suggest for solve the problem?

to ice00 and to all

I am doing an experiment like this.

1. I take 10 families from the names_44323_4696.txt family group

AAAABTIRB
AAAABTIRH
AAAABUCLN
AAAABUGLN
AAAABUGRH
AAAABVCNK
AAAABVCQH
AAAABVENG
AAAABVHQG
AAAABVINE

2. Generating jobs for these 10 families.
3. Checking the generated jobs.
4. Processing of the results.

This is all very simple.
I do not think that writing this scenario for a BOINC project is difficult and time consuming.
We have software for each stage, except for stage 4. I perform this step separately.

Demis also sends me the results after stage 3 and I process them.

Of the 4696 families of my group, I checked on my PC only 540 families. This will take a very long time.
Checking the family group from Demis is also very long.

ice00

I am quoting your personal message because I want the discussion here to keep the community informed of the problem.

So, you need to have

GENERATOR: simple send one name: AAAABDGRU (it uses 10 bytes, so all
space required is about 500KB). This program is to code by zero.

FINDER: it takes the workunit (AAAABDGRU), then generate the data as
done by actual generator and then test it. FINDER should be one
executable that use BOINC API and that must be the union of your actual
generator and family checker: one program from the two sources you have.

This is why it is needed time in coding the program for running in BOINC
and it is no so simple like takes your actual programs and use them.

I am not an expert on software adaptation to the BOINC platform. And I did not say that it is simple.
But I know that this is possible.
This experiment was successfully (and very quickly!) performed in the Gerasim@Home BOINC project.
All results are obtained.

I did not say either that it is necessary to immediately generate all the tasks and have 160 GB to record all the WUs.
You can do in manual mode. This is exactly what Demis does, but he generated all the tasks for one group of families names_43423_26489.txt
He got about 100 GB with a WUs record. Now it handles these WUs.

I perfectly understand the scenario of generating one task described by you and checking it.
But I can't write programs for the BOINC platform.
Can you do it?

You say that you do not have time for this work.
But performing the experiment in manual mode will take much longer!
And you begin to perform a manual experiment. Do you have time for this?

PS. Please reply here.
I want to publicly discuss the problem and understand why this problem cannot be solved for six months already.
Progger seems to have refused to solve this problem.

What do you suggest for solve the problem?

It simple: it needs time to code and test if for BOINC platform.
Running a software in a PC is simple, have it running in 50.000 connected PC needs careful coding and testing, otherwise you can waste users resources or kill server resources.

I had very very few free time for coding due to my real life activities, so I cannot grant timing :(

Manual works is good only if it is one shot and then nothing more: in old server new experiment is running and it is programmed to recover itself in case of server reboot, so it not need more manual works on it.
The 52518 families I get for solutions could be resolved into 158 days in the old server at actual rate (the time will be reduced as soon as I put the incoming new server for ODLK1 and this server will be used for that task).

In mean time ODLK and ODLK1 performs the actual experiment so user resources are used for a good result :)

The 52518 families I get for solutions could be resolved into 158 days in the old server at actual rate (the time will be reduced as soon as I put the incoming new server for ODLK1 and this server will be used for that task).

This experiment can be performed in a BOINC project in a week!

You don't seem to understand me.
I do not ask you to perform experiments in manual mode.
I need new experiments in the BOINC project!

This is not the only experiment that is proposed.
This algorithm was developed by Belyshev.

I am also developing new algorithms.
Two months ago, I suggested that you run my experiment in a BOINC project.
Do you remember? There was my Basic program.
You did not cope with this program to rewrite it in C++
And the experiment was postponed.
I have been performing this experiment on my PC since December 2017.
Excellent results were obtained in this experiment. One unique result is marked in OEIS.
https://oeis.org/A287695

Do you understand that I can perform this experiment on a PC for another 10 years???
Do you understand that this experiment must be run in a BOINC project???
In my opinion, you do not understand this!

I have a few more experiments, I am developing new algorithms.
I am not able to perform all these experiments on a single PC.

The algorithm that operates in the ODKL and ODLK1 projects is morally obsolete.
This is a good algorithm (from Belyshev), but there is much better.
It is necessary to use new developments, and not to stand still.

I do not need BOINC projects that routinely work on the same algorithm and do not use new algorithms.
You can stop the project ODLK1.
Progger can stop the ODLK project.
This is no longer interesting!

I have been persuading you and Progger to launch new experiments for six months.
Now I see that it is useless.
You and Progger ignore all my requests for a new experiment.

I will continue to perform experiments in manual mode.

You can open the results in the ODLK1 project. Let them be available to everyone in the original form.
I will not waste time processing these results.
By the way, I asked you and Progger a long time ago to do automatic processing of the results.
This is not done!
I wrote in detail lgorithm for processing the results to Progger.

If you are interested in continuing the ODLK1 project in its current form, continue.
I'm not interested.
I need new experiments.

Good luck!

PS. I offered to ask for help from the community if you are unable to independently launch an experiment in a BOINC project.
You have refused this offer.
I did my best. I no longer have the ability to convince you and Progger.

Dear participants!

The ODLK and ODLK1 BOINC projects no longer require my services as a researcher.
In these projects do not want to run new experiments.
Therefore, I am looking for the opportunity to launch a new BOINC project ODLK2.

See
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=50

Please write me PM if you are interested in this offer.

PS. If the ODLK and ODLK1 projects are stopped, write me to e-mail
natalimak1@yandex.ru

Hello, ice00!

I answer the most important here.

It seems that actually you are saying to all client people that are
running the ODLK and ODLK1 experiment, that having donating 1 year of
their resources for computation was a waste of time as experiment is old
even if 400000 unique CF ODLS come from ODLK1 and 73000 from ODLK each
month. Maybe this is what could understand a people that read the forum
actually.

The results obtained in the ODLK and ODLK1 projects are good. They will always be relevant.
But the algorithm that works in these projects is morally obsolete.
Science goes forward!
We must use the new achievements of science, and not stand still.
Do you know that my algorithm gives 5 times more results.
So you do not get 400,000 GF ODLS per month in the ODLK1 project, but 2,000,000 CF ODLS!
Do you see the difference?

Thus, customers waste time, working on the old algorithm.
They could work much more efficiently.
Do you understand this?

For ODLK2 you have my suggestion of what I will implement for adapting
your software for BOINC without crashing the server resources, maybe who
you will find that work for it can get inspiration from that for
starting up.

Else it is very important that in ODLK2 you educate the programmer to
have workunit with big works in it as BOINC is very inefficient in
saving resources and it will fill the disk with lot of files and
directories that can saturate the INODES of a disk and slow down every
backup operation.

Are you offering me for ODLK2 your new server?
Are you suggesting that I train a programmer to make the right software for the ODLK2 BOINC project?
Unfortunately, I cannot train a programmer, because I myself am not an expert in this field.
I cannot hire an experienced programmer for money, because I have no money, I live on a modest retirement.
Thus, I can not use your suggestions.

ice00

If you really want to run the ODLK2 project on your server, I suggest you announce in the community a search of an experienced programmer for this work.

I announced this a long time ago
https://boinc.berkeley.edu/dev/forum_thread.php?id=12622

If you find a programmer, I will provide the software for my algorithms for the ODLK2 project.

However, why can not start a new Application in the project ODLK1?
Do you think this is impossible? Just because you don't have time for this?
Or is it impossible for other reasons?
I do not understand this yet.
If you do not have time, you may be helped.

Thus, you decide this question yourself. I have already put a lot of effort into solving this problem, and I haven’t received a decision.

Let me know when you need my software.

ice00

Could you please tell, are you using BOINC Wrapper technology for existing application can be run under BOINC?

hi, it is uses source code adapted to run using Boinc Api.

hi, it is uses source code adapted to run using Boinc Api.

Is Boinc Api and BOINC Wrapper similar technologys?