Here is the block #10 of tasks
https://cloud.mail.ru/public/6n4r/sLZJvu1Ug

The archive contains WU 91-100.

I am preparing block #11.

Here is the block #11 of tasks
https://cloud.mail.ru/public/Mpxb/gjuF4cZZy

The archive contains WU 101-110.

I am preparing block #12.

Running now 10,11

Running now 10,11

Do you check in parallel?

yes , six blocks now , 6 7 8 9 10 11

Here is the block #12 of tasks
https://cloud.mail.ru/public/5gvc/CkuQobtmA

The archive contains WU 111-120.

I am preparing block #13.

yes , six blocks now , 6 7 8 9 10 11

Good!

run block 12 now..

Here is the block #13 of tasks
https://cloud.mail.ru/public/CjcE/TLn7mGT2M

The archive contains WU 121-130.

yes , six blocks now , 6 7 8 9 10 11

And the verification speed does not decrease when working in parallel?
Do you have a multi-core processor?

Tomas Brada
I received from you the results of blocks 6,7,8.
Thanks!

Now I will process these results.

In blocks 6,7,8 found 128 unique CF ODLK!
There is one symmetric solution with symmetry (4,31,31)

0 8 4 6 2 7 5 9 3 1
4 1 0 8 5 3 9 6 2 7
3 6 2 1 7 9 8 0 4 5
9 7 6 3 1 8 4 2 5 0
5 3 8 9 4 6 0 1 7 2
7 0 1 4 3 5 2 8 9 6
2 9 3 7 8 1 6 5 0 4
1 2 9 5 0 4 3 7 6 8
6 5 7 0 9 2 1 4 8 3
8 4 5 2 6 0 7 3 1 9

Great results!

Sorry, I am slowly generating tasks.

I work in semi-automatic mode. It takes a lot of time and a lot of attention.

Ms Makarova said that quorum is optional, so I am not going to run block 12 and less.

If boinc application is not possible now, then I suggest larger blocks. I can run 16 tasks in about a hour, therefore I would appreciate blocks of up to 128 work units. I have fast system. Manually downloading and extracting and mailing results is tedious. If there were 128 tasks in a block, I would be able to download once and let it run without attention for 8 hours. Much higher efficiency than 10 workunits.

Thank you for the source code. I compiled it without problems for linux. Let me tell you what I did and what I archived: recompiled the code for Linux (works better than under emulator), enabled compiler optimizations and I wrote a script to run workunits in parallel. I will share the script.

My script takes care of the recompiling, but it needs the sources in a "build" subdirectory. 1) Download the source archive 2) extract it 3) recode just the source files (cpp, h) to system encoding with "iconv". Downloaded files are in "Windows 1251"/"cp1251" encoding. You can use iconv like this (for all source files, adjust paths as needed):

 iconv -f cp1251 family_mar/source/kanonizator.cpp >work/build/kanonizator.cpp

My run script is as follows:

#!/bin/bash
echo "$PWD $0 $@" >log.txt
exec 4>>log.txt
uname -a >&4
wine --version >&4
date >&4

function parallel()
{
  echo "Starting WUs $@"
  for n; do
    rm -rf wu$n.d
    mkdir wu$n.d
    cd wu$n.d
    ln ../wu$n.txt input.txt
    ln -t. ../hash_tabl.bin ../build/family_mar.exe
    echo "-- family_mar.exe wu$n.txt --" >>log.txt
    md5sum hash_tabl.bin family_mar.exe input.txt >>log.txt
    yes | ./family_mar.exe >>log.txt 2>&1 &
    cd ..
  done
  wait
  echo "WUs finished ($?)"
}

function collect()
{
  >results_$1.txt
  for n in $(seq $2 $3); do
    cat wu$n.d/output.txt >>results_$1.txt
    #iconv -f cp1251 wu$n.d/log.txt >&4
    cat wu$n.d/log.txt >&4
    md5sum wu$n.d/output.txt >&4
  done
  md5sum results_$1.txt >&4
  echo "Finished $1"
}

function compile()
{
  echo "Compiling"
  cd build
  md5sum *.cpp *.h >&4
  g++ -c -O3 -o prov_blk_trans.o prov_blk_trans.cpp >&4 2>&1 || return 1
  g++ -c -O3 -o kanonizator.o kanonizator.cpp >&4 2>&1 || return 1
  g++ -fpermissive -O3 -o family_mar.exe kanonizator.o prov_blk_trans.o prov_blk_main.cpp >&4 2>&1 || return 1
  cd ..
  return 0
}

if ! compile; then
  echo "Compile error"
  exit
fi

parallel $(seq 51 66)
parallel $(seq 67 82)
parallel $(seq 83 90)
collect block6 51 60
collect block7 61 70
collect block7 71 80
collect block7 81 90

date >&4

The script needs to be in the same directory as your wuXX.txt input files (from the download) and hash_tabl.bin file and the build directory with sources. At the end, there are lines with "parallel" and "collect". You should adjust the numbers (they are inclusive range) to run the WUs you have. "parallel" runs the range of tasks in parallel, you should run the same number of tasks as your core count for optimum performance, i run 16. The "collect" line creates the "results_blockXX.txt" file, you need to set the range exactly as requested (eg wu 51-60 for block 6).

The script makes a log file, there are file checksums for verification.

---

Attention!

Blocks 1 - 8 checked.

Verified blocks are removed from the resource.

If boinc application is not possible now, then I suggest larger blocks. I can run 16 tasks in about a hour, therefore I would appreciate blocks of up to 128 work units. I have fast system. Manually downloading and extracting and mailing results is tedious. If there were 128 tasks in a block, I would be able to download once and let it run without attention for 8 hours. Much higher efficiency than 10 workunits.

Good. I will try to make a block of 128 tasks.

Thanks.

Tomas Brada
Thank you very much for your work in preparing the program for the Linux system.
Unfortunately, I do not work on a Linux system.

This can discuss ice00.

Running block 13 first
and then block 12 with low priority to double check.

Tomas Brada
Thank you very much for your work in preparing the program for the Linux system.
This can discuss ice00.

The program works well on linux system by itself, there was just on line at the end that showed error. To make it work in boinc, the boinc API needs to be included. It is not hard to do, but not trivial either. I might look into that, but I do not promise.
The script is not useful for boinc, I wrote it for myself, but I decided to share it.

To make it work in boinc, the boinc API needs to be included. It is not hard to do, but not trivial either. I might look into that, but I do not promise.
The script is not useful for boinc, I wrote it for myself, but I decided to share it.

I do not know what the boinc API is.
But ice00 seems to have written that this feature in the BOINC project ODLK1 is already included.
Please see this message
https://boinc.multi-pool.info/latinsquares/forum_thread.php?id=45&postid=309#309

That is what I wanted to get from ice00 - the launch of the new ODLK1 Application using the boinc API.
But it did not work!

Maybe ice00 needs help with this.
But I can not get an exact answer :(
ice00 just says he has no time.