FATAL ERROR: High global CUDA exclusion count!

From: Nima Nouri (sunny.day.programmer.2019_at_gmail.com)
Date: Wed Jun 12 2019 - 03:20:20 CDT

Dear Experts
Hi and hope you are doing well

I have a system of SBCG created by VMD.
when I simulate the system using NAMD on CPU with multithreads there is no
errors and the simulation runs perfectly.

The problem is when I try to run the system on GPU using CUDA version of
NAMD 2.13 on windows 10 I get the following error after the optimization
and when starting the MD simulation:

Charm++ fatal error:
FATAL ERROR: High global CUDA exclusion count! (2654 vs 2652) System
unstable or pairlistdist or cutoff too close to periodic cell size.

I have double checked this page for running NAMD with CUDA:
https://www.ks.uiuc.edu/Research/namd/2.13b1/ug/node97.html

using the following commandline for the run:
namd2 +idlepoll +p8 sim.conf

also tried +p1 with no change.

I should add that when running on the CPU trajectory files are fine and the
results are ok but when running on GPU the system crush inside/into itself
in matter of a few steps which is Cleary something with the force field
parameters. but I dont know what.

The required files for running the simulation are included in the zip file
shared in the drop box link below:
https://www.dropbox.com/s/feigmwjwuqu99ts/testCUDA.rar?dl=0

Any help to export the simulation to GPU will be appreciated.
Best Regards

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:20:49 CST