Dear Dennis,
I would like to report multiple segmentation faults with GIBS v1.0. I tried to run a few GCMC simulations with relatively large enzymes. Input/output files, logs and APBS logs are provided in zipped format through Gdrive. All simulations were tried on two machines. One has CentOS7 (kernel 3.10.0-693.11.6.el7.x86_64) with Intel Xeon E5-2680V4 and 128 GB DDR4 RAM, the other has Ubuntu Server 16.04 (kernel 4.4.0-104-generic) with Intel Xeon E5540 and 48 GB DDR3 RAM. Both gave exactly the same errors in all cases. Also tried to play with the "MAXIMUM_NUMBER_OF_PARTICLES" keyword to modulate memory allocation, but it didn't change anything in any case.
-
A hydrolase enzyme (CaLB) with about 4,600 atoms went smoothly, without problems
(https://drive.google.com/open?id=1EWh73b5-0vbFmAwd4wb6wcIvP5FOEwwi)
-
One chain of an enzyme (PaPAM-chainA-mod, about 8,000 atoms), however, gave segmentation fault around the "CCavityGrid::setLJCellIndexMaps" step (see log.txt). Increasing the CAVITY_GRID_SPACING, and decreasing the CAVITY_GRID_SEGMENTS didn't have any effect (see log-CGSpac2.5-CGSeg-10.txt)
(https://drive.google.com/open?id=1BvKo5f-AcVSu7Shb3-6tT2r3AiitniqY)
-
Truncated model from the previous enzyme, being a homotetramer of the previously mentioned enzyme chain, gave also segmentation fault (Truncated-PaPAM). It has also ca 8,000 atoms, obviously there are breaks in the backbone as it is a truncated model, but went surprisingly farther in the simulation, but gave segmentation fault at the "[RunGCMCSimulation1.cpp(runSimulation_1()]: Updating cavity lists based on solute atoms." step (see log.txt). Changing grid spacing and segment number didn't do anything again (see log-CGSpac2.5-CGSeg-10.txt)
(https://drive.google.com/open?id=10RvK8q7lgAdrn6qTPwS_EOFhcS3uGxK-)
-
Two large enzymes (13,000 and 32,000 atoms) gave segmentation fault after opening the PQR file ([solute_type.cpp(CSoluteModel::readSoluteAllAtomCoordsFromFile)] xyz.pqr is opened.).
(https://drive.google.com/open?id=1ci9dKqS-uqeQmyJ7pEQpJDHbWi8AJXd4)
(https://drive.google.com/open?id=1drLGNlPCFg6PTt_GSx6bJjcigZeCbTbx)
Hope they can be fixed, and looking forward to see the code working with these enzymes.
Dear Dennis,
I would like to report multiple segmentation faults with GIBS v1.0. I tried to run a few GCMC simulations with relatively large enzymes. Input/output files, logs and APBS logs are provided in zipped format through Gdrive. All simulations were tried on two machines. One has CentOS7 (kernel 3.10.0-693.11.6.el7.x86_64) with Intel Xeon E5-2680V4 and 128 GB DDR4 RAM, the other has Ubuntu Server 16.04 (kernel 4.4.0-104-generic) with Intel Xeon E5540 and 48 GB DDR3 RAM. Both gave exactly the same errors in all cases. Also tried to play with the "MAXIMUM_NUMBER_OF_PARTICLES" keyword to modulate memory allocation, but it didn't change anything in any case.
A hydrolase enzyme (CaLB) with about 4,600 atoms went smoothly, without problems
(https://drive.google.com/open?id=1EWh73b5-0vbFmAwd4wb6wcIvP5FOEwwi)
One chain of an enzyme (PaPAM-chainA-mod, about 8,000 atoms), however, gave segmentation fault around the "CCavityGrid::setLJCellIndexMaps" step (see log.txt). Increasing the CAVITY_GRID_SPACING, and decreasing the CAVITY_GRID_SEGMENTS didn't have any effect (see log-CGSpac2.5-CGSeg-10.txt)
(https://drive.google.com/open?id=1BvKo5f-AcVSu7Shb3-6tT2r3AiitniqY)
Truncated model from the previous enzyme, being a homotetramer of the previously mentioned enzyme chain, gave also segmentation fault (Truncated-PaPAM). It has also ca 8,000 atoms, obviously there are breaks in the backbone as it is a truncated model, but went surprisingly farther in the simulation, but gave segmentation fault at the "[RunGCMCSimulation1.cpp(runSimulation_1()]: Updating cavity lists based on solute atoms." step (see log.txt). Changing grid spacing and segment number didn't do anything again (see log-CGSpac2.5-CGSeg-10.txt)
(https://drive.google.com/open?id=10RvK8q7lgAdrn6qTPwS_EOFhcS3uGxK-)
Two large enzymes (13,000 and 32,000 atoms) gave segmentation fault after opening the PQR file ([solute_type.cpp(CSoluteModel::readSoluteAllAtomCoordsFromFile)] xyz.pqr is opened.).
(https://drive.google.com/open?id=1ci9dKqS-uqeQmyJ7pEQpJDHbWi8AJXd4)
(https://drive.google.com/open?id=1drLGNlPCFg6PTt_GSx6bJjcigZeCbTbx)
Hope they can be fixed, and looking forward to see the code working with these enzymes.