CNS executable not working

HADDOCK
1

I have downloaded and built haddock2.5 on my University’s supercomputer. I followed the installation documentation, but the cns does not seem to be compatible with my system. It gets built and i changed all the local paths etc, but when I try to run the executable I get:

[user@node067 source]$ ./cns_solve-2411121204.exe
%SETFPEPS Machine epsilon determined to be 0.494-323
%SETFPEPS error encountered: Machine epsilon value is too small
(CNS is in mode: SET ABORT=NORMal END)
WARNING: program encountered a fatal error.
However, in interactive mode, program execution
will continue. Proceed at your own risk.
Program will stop immediately.
============================================================
Maximum dynamic memory allocation: 0 bytes
Maximum dynamic memory overhead: 0 bytes
Program started at: on
Program stopped at: 13:04:54 on 12-Nov-2024
CPU time used: 0.0065 seconds
============================================================
Note: The following floating-point exceptions are signalling: IEEE_UNDERFLOW_FLAG IEEE_DENORMAL

This error propogates to the implementation of haddock as the cns is linked to this executable. I have contacted my hpc admin and they have not been able to locate the issue as well. Is there any workaround for getting a working cns executable?

2

Did you copy first all the CNS files provided with haddock (check the cns1.3 directory) to the source directory of your CNS installation?

If not, do it and recompile

3

Yes my directory looked like this before compilation:

$ls
bin collapse.inc ener.inc initia.f machvar.f README supccr.inc xangle.inc xdipo_rdc.inc
Changes collapse_rgyr.f fantacross.f inputs mac-intel-darwin rtf.inc test xccr.f xfantaccr.f
cns.f connect.f fantalin.f instlib Makefile segmnt.f utils xccr.inc xfantadipo.f
cns.inc cstran.f fantaxplor.inc intel-x86_64bit-linux modules selrpn.f vector.f xdipo_pcs.f xt1dist.f
cns_solve_env doc genic.f libraries param.inc source version.inc xdipo_pcs.inc
cnst.inc energy.f helplib License Readme sup2.inc xangle.f xdipo_rdc.f

and i am using gcc/9.3.0. When it’s compiling the following message appears:

fortran -o PSmapy -w -O PSmapy.f
gfortran -o PSmapz -w -O PSmapz.f
gcc -o to_cns -O to_cns.c -lm
to_cns.c:65:1: warning: return type defaults to ‘int’ [-Wimplicit-int]
65 | main(int argc, char *argv) {
| ^~~~
g++ -o cluster_struc -O cluster_struc.cpp -lm
lex refloat.l
gcc -O -o refloat lex.yy.c -lm -lfl
/usr/bin/ld: cannot find -lfl
collect2: error: ld returned 1 exit status
make[4]: *** [refloat] Error 1
make[3]: *** [utils] Error 2
make[2]: *** [compile-utils] Error 2
make[1]: *** [utils] Error 2

flags:
fortran → [gfortran] -fdefault-integer-8 -w -O3 -funroll-loops -ffast-math
c → [gcc] -O -DINTEGER=‘long int’ -DCNS_ARCH_TYPE_LINUX
link → [gfortran] -w

compiling: angledb.f
…

4

Make sure to have copied all the files from the haddock cns1.3 directory into the proper location (check the README file)

Did you actually try to use the pre-compiled executable provided in the haddock software email?

Also gcc9 is rather old…

5

I got it working. What ended up solving this was copying the .f files from cns1.3 to the software/cns-1.3/source rather than in the software/cns-1.3/ directory so that they are read when making the software. Additionally just running “make install” rather than “make install compiler=gfortran” kept the make from using the gfortran in my anaconda python module that was loaded and used the gcc/9.3.0 as needed.