VMD-L Mailing List

From: Oleksii Zdorevskyi (alzdorev_at_gmail.com)
Date: Mon May 04 2020 - 09:11:25 CDT

Dear John,

Thanks for your message.

When compiling vmd 1.9.4 I obtained the error:

/usr/bin/ld:
/usr/local/cuda-10.2/lib64/libcudart_static.a(libcudart_static.a.o):
undefined reference to symbol 'pthread_rwlock_trywrlock@@GLIBC_2.2.5'
//lib/x86_64-linux-gnu/libpthread.so.0: error adding symbols: DSO missing
from command line
collect2: error: ld returned 1 exit status

I have dealt with this problem by adding "-lpthread" command to the
Makefile in the end of of the "LIBS = " line.

Best,
Oleksii

пн, 13 апр. 2020 г. в 17:55, John Stone <johns_at_ks.uiuc.edu>:

> Oleksii,
> Is there a reason you're unable to run the VMD 1.9.4 test versions?
> The error you're getting below is a known bug in recent versions of CUDA
> (> 9.1 IIRC) that the new versions of VMD have workarounds for.
>
> Best,
> John Stone
> vmd_at_ks.uiuc.edu
>
> On Mon, Apr 13, 2020 at 05:45:11PM +0300, Oleksii Zdorevskyi wrote:
> > Hi, Josh,
> > Thanks for your answer.
> > In my Ubuntu 18.04 I have gcc version 8.4.0. I firstly I upgraded it
> to
> > the version 9.2.1 but during vmd compilation I got an error that my
> cuda
> > libraries are incompatible with gcc versions 9* and older, so I
> returned
> > back to the previous version.
> > I tried to do 'make veryclean' but it didn't help.
> > I even made an attemptĀ to install older CUDA libraries (8.0) but
> they
> > require an older nvidia driver version that has a conflict with my
> GPU.
> > When compiling VMD, in the log-file above there is a following error:
> > echo "Compiling " CUDAMarchingCubes.cu " --> " CUDAMarchingCubes.o "
> ...";
> > \
> > /usr/local/cuda-10.2/bin/nvcc --ptxas-options=-v -gencode
> > arch=compute_30,code=compute_30 -gencode arch=compute_30,code=sm_35
> > -gencode arch=compute_30,code=sm_37 -gencode
> > arch=compute_50,code=compute_50 -gencode arch=compute_50,code=sm_50
> > -gencode arch=compute_60,code=compute_60 -gencode
> > arch=compute_60,code=sm_60 --ftz=true Ā --machine 64 -O3
> > Ā -DARCH_LINUXAMD64 -DVMDOPENGL -DVMDOPENGLPBUFFER -DVMDGLXPBUFFER
> > Ā -DVMDACTC Ā -DVMDCOLVARS -DVMDCUDA -DMSMPOT_CUDA -DVMDMPI Ā -DVMDIMD
> > -DVMDXINPUT -DVMDLIBTACHYON -DVMDCONTRIB -DVMDPYTHON -DVMDTHREADS
> > -DWKFTHREADS -DUSEPOSIXTHREADS -D_REENTRANT -DVMDNUMPY -DVMDQUICKSURF
> > -DVMDWITHCARBS -DVMDPOLYHEDRA -DVMDSURF -DVMDMSMS -DVMDNANOSHAPER
> -DVMDTCL
> > -DVMDTK Ā -DVMDSTATICPLUGINS Ā -DVMDGUI -DVMDFLTK Ā
> -I../lib/actc/include
> > Ā -I/usr/include/mpich Ā -I../lib/tachyon/include
> > Ā -I/usr/include/python2.7 -I/usr/include/python2.7/numpy
> > -I/usr/lib/python2.7/dist-packages/numpy/core/lib Ā
> -I/usr/include/tcl
> > -I/usr/include/tk -I/usr/local/lib/vmd/plugins/include
> > -I/usr/local/lib/vmd/plugins/LINUXAMD64/molfile -I./lib/netcdf/include
> > -I../lib/fltk/include -I. -c CUDAMarchingCubes.cu -o
> > ./LINUXAMD64/CUDAMarchingCubes.o
> > Compiling Ā CUDAMarchingCubes.cu Ā --> Ā CUDAMarchingCubes.o Ā ...
> >
> /usr/local/cuda-10.2/bin/../targets/x86_64-linux/include/thrust/system/cuda/detail/cub/block/block_scan.cuh(578):
> > error: no suitable constructor exists to convert from "int" to "uint2"
> > Are there any options to deal with it?
> > Thanks.
> > Best,
> > Oleksii
> > ŠæŃ*, 10 Š°ŠæŃ*. 2020 Š³. Š² 19:21, Josh Vermaas
> > <[1]joshua.vermaas_at_gmail.com>:
> >
> > Hi Oleksii,
> > Were there any other errors higher up in the log? I've also seen
> these
> > if I got halfway through a compile and then changed something.
> Sometimes
> > CUDA can get finicky about that. Just "make clean" to remove the
> object
> > files then "make" has solved this problem for me in the past. The
> other
> > possibility I can think of is that your gcc version is too old for
> CUDA
> > 10.2 (depends on your operating system.
> > SeeĀ [2]
> https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements
> ).
> > -Josh
> > On Fri, Apr 10, 2020 at 10:03 AM Oleksii Zdorevskyi
> > <[3]alzdorev_at_gmail.com> wrote:
> >
> > Hi all,
> > I am trying to compile vmd 1.9.3 on my PC with the following
> options:
> > LINUXAMD64 OPENGL OPENGLPBUFFER FLTK TK ACTC CUDA MPI IMD XINPUT
> > LIBTACHYON NETCDF COLVARS CONTRIB TCL PYTHON PTHREADS NUMPY GCC
> > But when running 'make' command I get the following error:
> > 1 error detected in the compilation of
> >
> "/tmp/tmpxft_000021ca_00000000-8_CUDAMarchingCubes.compute_60.cpp1.ii".
> > Makefile:586: recipe for target 'CUDAMarchingCubes.o' failed
> > make: *** [CUDAMarchingCubes.o] Error 1
> > I have GPU GTX 1660 TI, the version 10.2 of cuda libraries and
> Nvidia
> > driver 440.82.
> > Could you give me an idea what's wrong?
> > Thanks in advance.
> > Best,
> > Oleksii
> >
> > References
> >
> > Visible links
> > 1. mailto:joshua.vermaas_at_gmail.com
> > 2.
> https://docs.nvidia.com/cuda/cuda-installation-guide-linux/index.html#system-requirements
> > 3. mailto:alzdorev_at_gmail.com
>
> --
> NIH Center for Macromolecular Modeling and Bioinformatics
> Beckman Institute for Advanced Science and Technology
> University of Illinois, 405 N. Mathews Ave, Urbana, IL 61801
> http://www.ks.uiuc.edu/~johns/ Phone: 217-244-3349
> http://www.ks.uiuc.edu/Research/vmd/
>