error failled to reallocate

[cardmaker]

hello, im having this error ``failed to reallocate 122880000 bytes´´ whit msieve 1.52 afther the script factmsieve.084 reach the 100%.

anyine have an ideia what is this error?

read 10M relations
read 20M relations
read 30M relations
read 40M relations
read 50M relations
failed to reallocate 122880000 bytes
Return value -1. terminating...

[xilman]

Originally Posted by cardmaker

hello, im having this error ``failed to reallocate 122880000 bytes´´ whit msieve 1.52 afther the script factmsieve.084 reach the 100%.

anyine have an ideia what is this error?

read 10M relations
read 20M relations
read 30M relations
read 40M relations
read 50M relations
failed to reallocate 122880000 bytes
Return value -1. terminating...

Yes.

You don't have enough memory, physical or virtual, to proceed. The process is asking for ~120MB.

[jasonp]

If this is a 32-bit msieve binary and you need more than 3GB of virtual memory, you will get this failure even if you have more RAM than that and even if you are using a 64-bit operating system.

[cardmaker]

Yes, is a 32bit version and i have 4 mg ram. do i have to change for another pc 64bit ?

[TemPiQ]

I also had exact the same error on both a 32-Bit system with 4GB RAM and a 64 Bit system with 8GB RAM.
Therefore the size of the RAM is not the cause of this error, I would say.

How do I continue the factorization after this error without invoking having to "factmsieve example.poly" again?

[jasonp]

Originally Posted by TemPiQ

I also had exact the same error on both a 32-Bit system with 4GB RAM and a 64 Bit system with 8GB RAM.
Therefore the size of the RAM is not the cause of this error, I would say.

How do I continue the factorization after this error without invoking having to "factmsieve example.poly" again?

If you are still using the 32-bit binary then you still have a 4GB limit on the virtual memory a process can allocate, even on a 64-bit OS. You would need to recompile Msieve as a 64-bit binary to get past this point. Precompiled 64-bit binaries of the latest SVN are available here; read through the whole thread to get the binary and dependencies that you need.

[TemPiQ]

Originally Posted by jasonp

If you are still using the 32-bit binary then you still have a 4GB limit on the virtual memory a process can allocate, even on a 64-bit OS.

Yes it looks like I had mixed some x86 and x64 binaries on the x64 computer.

Originally Posted by jasonp

You would need to recompile Msieve as a 64-bit binary to get past this point. Precompiled 64-bit binaries of the latest SVN are available here; read through the whole thread to get the binary and dependencies that you need.

Thanks for the hint to the dependencies for the precompiled binaries. I had restarted the factoring with the x64 binary before seeing your message.
Shortly afterwards the process was terminated because msieve could not find "vcomp110.dll".
I downloaded and installed the corresponding Visual Studio 2012 Redistribution package and was able to continue.

Now I have another error "integrator failed 1. #QNAN0e+000 1.#INF00e+000" after which msieve.exe is terminated.

Any hint to the cause of this failure?

Thanks a bunch

PS: I was using the binary from the package "msieve152_svn939_win64_i7.zip"

[jasonp]

You usually get that failure when the NFS polynomial is not read in at all, or some piece of it is missing. Check readme.nfs to make sure all your input files are in the correct format.

[TemPiQ]

Originally Posted by jasonp

You usually get that failure when the NFS polynomial is not read in at all, or some piece of it is missing. Check readme.nfs to make sure all your input files are in the correct format.

After checking the readme.nfs, I could confirm that all the input files were correct.
The order of the polynomial coefficients were in descending order i.e. A5 ... A0, which is ok according to the readme.nfs.

I believe the problem was caused by the parameters passed to msieve by the python script factmsieve.86.py and if I am not mistaken on line #1881 viz.
ret=run_msieve('-t {0:d} -nc1'.format(LA_THREADS))
In my case the failure and the subsequent error message "error generating or reading NFS polynomials" occurred while the following command was been invoked:
msieve -s examples\rb84.dat -l examples\rb84.log -i examples\rb84.ini examples\rb84.fb -t 8 -nc1

The sieving process started after deleting the parameter -nc1.

[Dubslow]

The "error generating or reading NFS polynomials" message is benign and quite common. It may generally be ignored.

[jasonp]

Complaints about the integrator failing usually stem from the polynomial not being there in the first place.

Removing the '-nc1' switch means you are now running the quadratic sieve and not the number field sieve, which is likely not what you want.