Hi,
On Thu, Aug 1, 2019 at 6:41 AM Karim Mahnam <karimmahnam93.gmail.com> wrote:
> I have extracted AmberTools 19 and Amber14 and put them in folder Amber18
> in /usr/local directory.
The problem here is that the pmemd CUDA install target name changed
from 'cuda' in Amber 14 to 'cuda_serial' by Amber 16. In other words,
the build system of AmberTools 19 is not compatible with Amber 14.
Your best bet to install Amber 14 alongside AmberTools 19 is probably
to maintain them separately. What I would do is download AmberTools 14
(
http://ambermd.org/AmberTools14-get.html) and use that to install
Amber 14 in e.g. ~/amber14 (or wherever you prefer), then install
AmberTools 19 in a separate directory (e.g. ~/amber18). Since
presumably all you want from Amber 14 is pmemd.cuda you can safely set
AMBERHOME to point to your AmberTools 19 install.
All this being said, I highly recommend upgrading to the latest
version of Amber for pmemd.cuda, especially if you have modern GPUs.
There have been significant performance improvements since Amber 14.
Hope this helps,
-Dan
> My graphic cart is : GF108 [GeForce GT 440] and I have installed nvidia
> driver (NVIDIA-SMI 390.116, Driver Version: 390.116) and cuda-9.0 toolkit
> software (Cuda compilation tools, release 9.0, V9.0.176) in /usr/local
> directory as superviser user and ./deviceQuery (in
> /usr/local/cuda/samples/bin/x86_64/linux/release) command works:
> deviceQuery, CUDA Driver = CUDART, CUDA Driver Version = 9.1, CUDA Runtime
> Version = 9.0, NumDevs = 1
> Result = PASS
>
> I have to install cuda-9.0 , because my graphic cart and my nvidia driver
> only work with cuda-9.0.
> my Ubnutu is 18.04 (4.15.0-55-generic) and gcc version is 6.5.0 (gcc
> (Ubuntu 6.5.0-2ubuntu1~18.04) 6.5.0 20181026) and my gfortran is 7.4.0 (GNU
> Fortran (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0).
>
> I set environmental variables as follows:
>
> test -f /usr/local/amber18/amber.sh
> source /usr/local/amber18/amber.sh
> export AMBERHOME=/usr/local/amber18
> export CUDA_HOME=/usr/local/cuda-9.0
> export PATH=/usr/local/cuda-9.0/bin${PATH:+:${PATH}}
> export LD_LIBRARY_PATH="/usr/local/cuda-9.0/lib64:${LD_LIBRARY_PATH}"
>
> in .bashrc file in root directory and source them and also inserted thses
> variables in linux terminal (echo $CUDA_HOME says: /usr/local/cuda-9.0).
>
> I have installed MPI version of amber and sander.MPI works, I want to
> install GPU version of Amber and I performed "make clean" and "./configure
> -cuda gnu" without error in Amber18 directory.
> When I perform "make install" as superviser, it says :
>
> mv cpptraj.cuda "/usr/local/amber18/bin"
> make[4]: Leaving directory '/usr/local/amber18/AmberTools/src/cpptraj/src'
> make[3]: Leaving directory '/usr/local/amber18/AmberTools/src/cpptraj'
> make[2]: Leaving directory '/usr/local/amber18/AmberTools/src'
> AmberTools serial CUDA install complete.
> make[1]: Leaving directory '/usr/local/amber18/AmberTools/src'
> make[1]: Entering directory '/usr/local/amber18/src'
> make[1]: *** No rule to make target 'cuda_serial', needed by 'install'.
> Stop.
> make[1]: Leaving directory '/usr/local/amber18/src'
> Makefile:7: recipe for target 'install' failed
> make: *** [install] Error 2
>
> and pmemd.cuda does not work.
> Would you please tell me how can I solve this problem.
> Thank you very much.
> Dr. Karim Mahnam
> Associate Professor in Shahrkord University
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Aug 01 2019 - 05:00:02 PDT
