Discussion related to cado-nfs

[meng <qsmeng@126.com>]

Dear Paul,

>> /home/greatnet/meng/cado-nfs/build/master/sieve/las -I 11 -rpowlim 2047 -apowlim 2047 -poly /tmp/c100.poly -fb /tmp/c100.roots -q0 2040000 -q1 2060000 -mt 2 -out /tmp/c100.rels.2040000-2060000.gz
>> the gdb result is given below :
>> #0  0x00000039f5a7a050 in memset () from /lib64/libc.so.6
>> #1  0x0000000000413c0b in init_rat_norms_bucket_region (S=0x6a0000 "", j=<value optimized out>, si=<value optimized out>)
>>     at /home/greatnet/meng/cado-nfs/sieve/las-norms.c:332
>> #2  0x000000000040980e in process_bucket_region (th=0x65fac0) at /home/greatnet/meng/cado-nfs/sieve/las.c:2858
>> #3  0x00000039f660683d in start_thread () from /lib64/libpthread.so.0
>> #4  0x00000039f5ad4fad in clone () from /lib64/libc.so.6
>> The version is 8efeb35.
 I am thinking that this should be related to  dynamic library as compiling is successful.
>$ makefb -I 11 -poly c100.poly -alim 2000000 > c100.roots
  I can reproduce this step and md5sum of c100.roots is the same as yours.

>$ las -I 11 -rpowlim 2047 -apowlim 2047 -poly /tmp/c100.poly -fb /tmp/c100.roots -q0 2040000 -q1 2060000 -mt 2 -out /tmp/c100.rels.2040000-2060000.gz -rlim 1000000 -alim 2000000 -lpbr 25 -lpba 25 -mfbr 50 -mfba 50 -rlambda 2.1 -alambda 2.2

  I cant reproduce this step. 
>First please checkout revision d7da39a (or later), recompile the "las" binary
>(maybe after removing the old one to be sure), and try again.
  I failed in these latest  version. But it is strange that I can succeed in the old version f784e49c.  I think if I can replace the sieve directory in latest version with that  old version.
 I will solve this later.
  Now when I try the mpi with  version f78e49c, there exists problem.
 After successful installation of openmpi-1.6.5 on two computers(I can run the example correctly), I do the following jobs on two computers:
cp local.sh.example local.sh
vi local.sh and add MPI=1 and MPI=/usr/local/openmpi-1.6.5/ and save it.
and rm -rf build/
make
vi mach_desc on the local host and add mpi=1 after the name of the two computers.
and run the cadofactor.pl. Finally the screen displays:

Info:Calling Block-Wiedemann...
    Error:Command `/home/greatnet/cado-nfs/build/master/linalg/bwc/bwc.pl :complete seed=1 thr=2x2 mpi=2x1 hosts=node2,localhost mpi_extra_args='--mca btl_tcp_if_exclude lo,virbr0' matrix=/home/greatnet/c59/c59.small.bin nullspace=left mm_impl=bucket interleaving=0 interval=100 mn=64 wdir=/home/greatnet/c59/c59.bwc shuffled_product=1 bwc_bindir=/home/greatnet/cado-nfs/build/master/linalg/bwc ' terminated unexpectedly with exit status 1.
    Error:STDOUT           'thr=2x2',
    Error:STDOUT           'interval=100',
    Error:STDOUT           'mpi=2x1',
    Error:STDOUT           'seed=1',
    Error:STDOUT           'mn=64',
    Error:STDOUT           'interleaving=0',
    Error:STDOUT           'splits=0,64',
    Error:STDOUT           'ys=0..64',
    Error:STDOUT           'matrix=/home/greatnet/c59/c59.small.bin'
    Error:STDOUT         ];
    Error:STDERR --------------------------------------------------------------------------
    Error:STDERR mpiexec was unable to launch the specified application as it could not access
    Error:STDERR or execute an executable:
    Error:STDERR 
    Error:STDERR Executable: /home/greatnet/cado-nfs/build/master/linalg/bwc/mf_bal
    Error:STDERR Node: node2
    Error:STDERR 
    Error:STDERR while attempting to start process rank 0.
    Error:STDERR --------------------------------------------------------------------------
    Error:STDERR /usr/local/openmpi-1.6.5/bin//mpiexec: exited with status 131

   Is there something i did wrongly? 
  Thank you.
  Regards,
Meng