Hi,
The GF108 variant (Fermi) is not supported by the latest versions of
pmemd.cuda; see
http://ambermd.org/GPUHardware.php for a list of
supported hardware.
Basically, NVIDIA has changed around their GPU architectures so much
over the years that the code required massive updates to properly
utilize the latest hardware.
-Dan
On Mon, Aug 19, 2019 at 1:53 AM Karim Mahnam <karimmahnam93.gmail.com> wrote:
>
> Hello Dear Dr. James Kress
> You mean, I can not run GPU of Amber by this graphic cart( GF108, GeForce
> GT 440 ) !. I have run GPU version of Gromacs package with this graphic
> cart.
> Dr. Karim Mahnam
> Associate Professor in Shahrkord University
>
>
>
> On Sat, Aug 17, 2019 at 8:46 PM James Kress <jimkress_58.kressworks.org>
> wrote:
>
> > Given NVIDIA's continuing chaos generation of its own GPU/ GPGPU market, by
> > releasing new architectures at such a rapid pace, is there a document that
> > delineates a comparison of Amber software/ NVIDIA architecture
> > compatibility? I've noticed some postings on this list that seem to
> > indicate backwards compatibility is not a goal for NVIDIA.
> >
> > Thanks.
> >
> > Jim Kress
> >
> > -----Original Message-----
> > From: Karim Mahnam <karimmahnam93.gmail.com>
> > Sent: Saturday, August 17, 2019 3:47 AM
> > To: AMBER.ambermd.org
> > Subject: [AMBER] Error searching for compatible GPU in Amber18
> >
> > Hello Dear Amber users
> > My graphic cart is NVIDIA Corporation GF108 [GeForce GT 440] (rev a1) and I
> > installed its driver (nvidia-340) and cuda toolkit 9.1 (nvcc: NVIDIA (R)
> > Cuda compiler driver, Cuda compilation tools, release 9.0, V9.0.176) and I
> > can run deviceQuary command, and Amber18 with GPU compatibility without
> > error, I also set environmental variables:
> > export AMBERHOME=/usr/local/amber18
> > export CUDA_HOME=/usr/local/cuda
> > export PATH=${PATH}:/usr/local/cuda/bin
> > export
> >
> > LD_LIBRARY_PATH=${LD_LIBRARY_PATH}:/usr/local/cuda/lib64:/usr/local/cuda/lib
> >
> > in .bashrc and source it, but when I run make test, all jobs fail.
> >
> > when I run pmemd.cuda, it says:
> > Unit 5 Error on OPEN: mdin
> >
> > But when I run pmemd.cuda for my inputs files (pmemd.cuda -O -i md.in -o
> > md1.out -p com.prmtop -c md.rst -r md1.rst -x md1.inpcrd -e mdenergy1.dat)
> > ,
> > it says:
> >
> > Error searching for compatible GPU
> >
> > I also can not run :
> > mpirun -np 8 pmemd.cuda -O -i md.in -o md1.out -p com.prmtop -c md.rst -r
> > md1.rst -x md1.inpcrd -e mdenergy1.dat 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
> >
> _______________________________________________
> 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 Mon Aug 19 2019 - 05:30:02 PDT
