Hi Roman,
I assume you are using Amber / AmberTools 24.
First, I suggest to use the API version, not the file based interface. It is faster and more robust. Instead of sander use the executable sander.quick.cuda to run the QM portion of a QM/MM calculation with QUICK on GPU. When using the API version you do not have to specify a quick executable name. In the input file set
&qmmm
qm_theory = ‘quick',
/
Note that QM/MM with DFT is expensive. It typically makes sense to minimize and equilibrate with MM and/or semiemprical or DFTB QM/MM before switching to ab initio QM/MM.
Regarding this error using the file based QM/MM interface:
SANDER BOMB in subroutine get_quick_forces (qm2_extern_quick_module)
Error with system call (executing QUICK)
Will quit now.
Not too much information about which system call cased the BOMB.
Any suggestions?
Execution of QUICK by system call from sander failed. In your case the executable is quick.cuda (requested in your input). There could be different reasons. Take a look at the input file QUICK_job.in that was written by sander and check the corresponding output file QUICK_job.out.
If this does not clarify what went wrong, try to execute the job manually and see if it works and if not, which error message you get.
# CUDA version
$> quick.cuda QUICK_job.in
# CPU version
$> quick QUICK_job.in
All the best,
Andy
—
Dr. Andreas W. Goetz
Associate Research Scientist
San Diego Supercomputer Center
Tel: +1-858-822-4771
Email: agoetz.sdsc.edu
Web: www.awgoetz.de
On Dec 6, 2024, at 1:14 PM, Osman, Roman via AMBER <amber.ambermd.org> wrote:
I am trying to run the QMMM quick program.
I constructed inputs for minimization, equilibration and production, but the program fails at the minimization stage.
This is my input:
Minimization input file in explicit solvent
&cntrl
! Minimization options
imin=1, ! Turn on minimization
maxcyc=5000, ! Maximum number of minimization cycles
ncyc=500, ! 100 steepest-descent steps, better for strained systems
! Potential energy function options
cut=8.0, ! Nonbonded cutoff, in Angstroms
ifqnt=1,
! Control how often information is printed to the output file
ntpr=100, ! Print energies every 100 steps
ntxo=2, ! Write NetCDF format
! Restraint options
ntr=0, ! Positional restraints for proteins, sugars, and ligands
nmropt=1,
! Set water atom/residue names for SETTLE recognition
watnam='WAT', ! Water residues are named WAT
owtnm='O', ! Water oxygens are named O
/
&wt
type='END'
/
DISANG=step5_production.rest
LISTIN=POUT
LISTOUT=POUT
&end
/
&qmmm
qmmask=':117,118,293,389,563,564',
qmcharge=0,
qm_theory='extern',
qmmm_int = 5,
/
&quick
method = 'B3LYP'
basis = 'def2-svp',
executable = 'quick.cuda',
/
This is the error message in the output:
| Program quick.cuda found!
| Executable location: /public/apps/amber/24_mpi_cuda_v2/bin/quick.cuda
--------------------------------------------------------------------------------
4. RESULTS
--------------------------------------------------------------------------------
SANDER BOMB in subroutine get_quick_forces (qm2_extern_quick_module)
Error with system call (executing QUICK)
Will quit now.
Not too much information about which system call cased the BOMB.
Any suggestions?
Roman Osman
Professor
Department of Pharmacological Sciences
Icahn School of Medicine at Mount Sinai
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!Mih3wA!H6KXPkpW5knRJSp11sxjRyRXQ4YWqbPDViPKibkEghPs3eRbzT2T61pNVB0Z1QjZi3rCBMwbcfbZxg$
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Dec 06 2024 - 15:00:20 PST
