I need a specific answer to my query. i tried searching but could not find a answer.
1 what is the difference between intel compiler version 18 and 17 updates 6/7. the document does not mention anything what all is included/ modified/ removed in ifort 18
2 i tried to compile my codes using both version 18. i get segmentation fault when i run input file. i tried exploring further.
3 i compiled my codes using version 17 update 5/6. the update 6 or earlier versions successfully compile codes and i can run input files. the input make file had paths/ libraries where ifort 17 is installed.
4 i recompiled codes using both version 18. i observed my run input files successfully. Pl note i did not switch off/restart the system and comment the environment settings of ifort 17. i could recompile and run codes using ifort 18 even after commenting ifort 17 in bash file
5 then i supressed environment settings of ifort 17 and rebooted the system. I recompiled codes but when i ran input file. I got the old segmentation fault.
i feel issue is related to environment settings if i run using ifort 18.
i need to know what should be done so that ifort 18 runs successfully without the help of ifort 17
The main changes according to the Release Note Overview are `
For more details, look here:
Segmentation faults can have many different origins, e.g. bad reads from your input files etc. Or clashes in consistency when you compiled with the ifort 18 compiler binary, but because of your environment variables you link with the ifort17 libraries or the other way round. This you could check with ldd operating on your final binary from your code. Normally this has a dependence only on system libraries, unless you use some intrinsic modules like iso_fortran_env or iso_c_binding or OpenMP etc. In order to be sure not to mix up anything, you could deinstall ifort 17 from your system such that only ifort 18 is there. Besides this, we probably need some code and sample of input file together with the exact message of the segmentation fault and a possible backtrace to diagnose what goes on. Did you run your program with idb?
the environment settings are as
#source /opt/intel/Cluster-2018/compilers_and_libraries_2018/linux/bin/compilervars.sh intel64
#source /opt/intel/Cluster-2018/mkl/bin/mklvars.sh intel64
#source /opt/intel/Cluster-2018/bin/compilervars.sh intel64
ulimit -s unlimited
#source /opt/intel/Cluster-2018/bin/compilervars.sh ia32
source /opt/intel/PXE-u6-2017/compilers_and_libraries_2017/linux/bin/compilervars.sh intel64
source /opt/intel/PXE-u6-2017/mkl/bin/mklvars.sh intel64
source /opt/intel/PXE-u6-2017/bin/compilervars.sh intel64
#export PATH=$PATH:/opt/intel/PXE-u6-2017/compilers_and_libraries_2018/linux/bin/intel64 ? replace 2018
forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
sword 0000000000418BAD Unknown Unknown Unknown
libpthread-2.19.s 00007F2595D68330 Unknown Unknown Unknown
sword 00000000004D4FA9 Unknown Unknown Unknown
sword 0000000000435774 Unknown Unknown Unknown
sword 00000000004037B7 orddrv_ 221 sword.f
sword 00000000004032D8 MAIN__ 66 sword.f
sword 000000000040325E Unknown Unknown Unknown
libc.so.6 00007F25959B1F45 __libc_start_main Unknown Unknown
sword 0000000000403169 Unknown Unknown Unknown
the earlier discussions are on https://software.intel.com/en-us/comment/1920333#comment-1920333
i have attached source codes in zipped folder UK....and binaries in sih2-d zipped folder. the sih2-d zipped folder also contains job file. when u run ./jobfile, the scatt1g-12a0 gives segemntation fault.
i need to understand the entire issue. for sure no issues with source codes...if anything is left in setting the environment correctly, pl let me know
Are you still trying to build and run the UKRMOL-IN codes, as you described in your earlier posts? If so, the following comments apply.
The UKRMOL codes contain a small number of bugs, but may work and produce correct results even though the bugs are present. They consist of a number of programs that interact with one another through unformatted files. The programs have to be run in a proper sequence and they have to be given correct inputs. Errors in doing so, as well as the bugs in the programs, can cause segmentation faults. Many of the bugs have been reported, but there appears to be little interest in fixing them. That is acceptable, provided the users have the expertise and other means of checking and verifying the results. This state of affairs is often a fact of life with codes that were written decades ago and continue to be in use today.
You, as well as most of the people posting in this forum, are probably not an expert on UKRMOL or its subject domain. In my opinion, it is a waste of time to attempt to overcome bugs in such a third party package by switching compilers or trying different compiler options.
durgapal, shubham wrote:
i need to understand the entire issue. for sure no issues with source codes.
There is a list of bug-reports at https://ccpforge.cse.rl.ac.uk/gf/project/ukrmol-in/tracker/?action=TrackerItemBrowse&tracker_id=282 . Some of the bugs are still marked "open".
sir what you are referring is an old version of codes (0.1). The one that I am taking is latest version (1.0) released in 2016. all these issues were taken into account in its release.
2 i am not interested in bugs but what i said it to understand the issue. why ifort 17 works and not ifort 18. are there environment issues that i am missing in my bash file ?
No, I am referring to bugs that are present in both the old releases and the current release. These bugs can be demonstrated to be present, but they are by no means obvious, and their effects are unpredictable. Specifically, a number of uninitialized variables are used in the various programs that make up UKRMOL-in.
Whether or not you are interested in bugs, they have an effect on the running of the programs and the results. Unless you are also not interested in the results being correct, you should be concerned about the bugs.
While it is not recommended, f77 codes which violate the standard only by requiring initialization to 0 to be substituted for uninitialized undefined data could be accommodated by -zero option. This is not entirely reliable.
Sir, Currently I am using Intel Compiler 18.0 Update 5 Intel(R) 64 Visual Studio 2017 for running a hydrological model but when I run it's .bat file then error occurs. Please help me to get rid off error "cannot open file libucrt.lib'" and "Not all components required for linking are present in the command line."
Diksha, you've asked a Windows question in the Linux/Mac forum. Please post your question again at https://software.intel.com/en-us/forums/intel-visual-fortran-compiler-for-windows in a new thread, not a reply to an old one.