CHARMM Development Project Forums User Discussion & Questions Installation and Testing Problem Compiling C45b1 - Free Version

Hello,

I am hoping that someone can provide direction on the following problem.

./charmm -i h2o-graphene-styr-heat.inp -o h2o-graphene-styr-heat.log&

Program received signal SIGSEGV: Segmentation fault - invalid memory reference.

Backtrace for this error:
#0 0x150079847bd0 in ???
#1 0x150079846e25 in ???
#2 0x1500794fee2f in ???
at ./signal/../sysdeps/unix/sysv/linux/x86_64/sigaction.c:0
#3 0x55a89bce2710 in ???
#4 0x55a89b794139 in ???
#5 0x55a89b4d6ff4 in ???
#6 0x55a89b4d8d72 in ???
#7 0x55a89b4d5bde in ???
#8 0x1500794e9cc9 in __libc_start_main
at ../csu/libc-start.c:308
#9 0x55a89b4d5c39 in ???

I have downloaded the free version of charrm C45b1, from the charmm website.

This is the operating system on my local laptop:
Sun Sep 06 16:33:28 [rossi@debian ~]$uname -a
Linux debian 5.7.0-3-amd64 #1 SMP Debian 5.7.17-1 (2020-08-23) x86_64 GNU/Linux

This is the gnu compiler version on my local laptop.
Sun Sep 06 16:33:34 [rossi@debian ~]$gcc -dumpversion
10 or version 10.2.0

The same error above occurs with the parallel version on my laptop, and
both exercutables were compiled with cmake.

Here is the gnu compiler version on the UConn cluster:
[anr11010@cn02 misc]$gcc -dumpversion
5.4.0

The same input above file works with a parallel version of charmm C45b1 on the UConn Linux cluster.

I understand that my laptop has a very new version of the compiler, but has charmm been tested with the latest and greatest versions of the gnu compiler?

Any help, direction, hints, and ideas would be extremely helpful.

Thanks so much for your attention to this.

Kind regards,

Angelo

gcc-10 is not working with CHARMM, unless you compile without optimization (-O2)

it would be helpful to reduce the level of optimization just for few files during the CHARMM compilation to ease the debugging. For now just use gcc-9.3, for example.

Milan

Hello,
I am using gcc 10.2. I tried passing on the -O2 argument to cmake with:

make CFLAGS="-O2" -j16

but alas, it failed to compile. Some of the warnings are:


.../charmm/build/cmake/fsshake.F90:342:77

More actual than formal arguments in procedure call at (1)
Rank mismatch between actual argument at (1) and actual argument at (2) (rank-1 and scalar)


.../charmm/build/cmake/energym.F90:588:23


Fortran 2018 deleted feature: Arithmetic IF statement at (1)
More actual than formal arguments in procedure call at (1)
type mismatch between actual argument at (1) and actual argument at (2)
Array reference at (1) out of bounds (2 > 1) in loop beginning at (2)

Best,
-Kevin

If you use VERBOSE=1 as an argument to make, you will get more information printed, esp. the argument flags actually being using by Fortran and C compilers. I do not think the CFLAGS variable has any impact on the Fortran flags, but you should be able to verify that.