forked from miaoyipu/quick
-
Notifications
You must be signed in to change notification settings - Fork 0
QUICK, a GPU-enabled ab intio quantum chemistry software
agoetz/quick-orig
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
!************************************************************************ ! QUICK ** ! ** ! Copyright (c) 2014 ** ! Regents of the University of Florida ** ! All Rights Reserved. ** ! ** ! This software provided pursuant to a license agreement containing ** ! restrictions on its disclosure, duplication, and use. This software ** ! contains confidential and proprietary information, and may not be ** ! extracted or distributed, in whole or in part, for any purpose ** ! whatsoever, without the express written permission of the authors. ** ! This notice, and the associated author list, must be attached to ** ! all copies, or extracts, of this software. Any additional ** ! restrictions set forth in the license agreement also apply to this ** ! software. ** !************************************************************************ ! ! Cite this work as: ! Miao,Y.: He, X.: Ayers,K; Brothers, E.: Merz,K. M. QUICK ! University of Florida, Gainesville, FL, 2014 and ! Michigan State University, East Lansing, MI, 2014 ! ! ! If you want to use GPU version, please cite: ! Miao, Y., Merz, K. M. (2013) Acceleration of Electron Repulsion Integral Evaluation on Graphics Processing Units ! via Use of Recurrence Relations Journal of Chemical Theory and Computation 9, 965–976. ! ! Tech support: [email protected] !************************************************************************ 1. INSTALLATION To install QUICK, first you may need to configure make.in file. Intel Fortran compiler is recommanded. ! Copy corresponding make.in file from makein folder and rename it to make.in. Then type make quick 1.1 CUDA VERSION INSTALLATION If you want to install GPU QUICK, NVIDIA CUDA COMPILER is required, which uses make.in.gnu.cuda. You may test 'nvcc --version'. Copy make.in.gnu.cuda to from makein folder to your directory and remane it to make.in cp ./makein/make.in.gnu.cuda ./make.in Before compiler, you must set CUDA_HOME in make.in file if not set in bashfile, otherwise, CUBLAS and other libraries can not been compiled or linked. Notice, if you want to support f orbital, set CUDA_SPDF=y in make.in file you copied from make.in.gpu.cuda, which will take a little bit long time and memory to compile, otherwise, set CUDA_SPDF=n for s, p and d orbital. (Default is not) ! Modify CUDA_HOME=(your cuda home) in make.in file ! Modify CUDA_SPDF=y or n in make.in file if you want or do not want to support gpu f function Then type make quick.cuda in ./bin directory, you can find executable files. 2. USAGE We suggust you to put this direct into PATH. Edit ~/.bashrc, then add this line export PATH=(YOUR QUICK DIR)/bin:$PATH and add basis set path export QUICK_BASIS=(YOUR QUICK DIR)/basis where (YOUR QUICK DIR) is your directory. the basic principal of QUICK is that the simpler, the better. so in the input file, the first line is calculation card. For example HF ncyc=3 energy BASIS=6-31gs denserms=1.0e-7 This means HF calculation is wanted, and after 3 cycle, We only calculate difference of Fock matrix, basis set is 6-31g**, and the convergence criteria is 1.0E-7. If you don't know anything about quantum chemistry, this card is recommended. After first line, you input your molecule geometry C 4.5916 2.6127 4.3145 H 6.8049 4.5138 -2.5775 H 8.3134 7.7325 -1.0688 H 4.3547 2.9911 3.3201 H 4.5176 3.5187 3.7129 which is the element and x,y,z coordinates respectively. Next, type quick mol.in or GPU version quick.cuda mol.in if you save the input file as mol.in. It will generate mol.out as output file. 3. About GPU VERSION Currently, QUICK support CUDA version 2.0 and up GPU. We tested on GTX580, M2090 and K20. We can't gureentee our program is flawless, and it is experimental now. If you have any question feel free to email the author.
About
QUICK, a GPU-enabled ab intio quantum chemistry software
Resources
Stars
Watchers
Forks
Packages 0
No packages published
Languages
- C 42.5%
- Fortran 24.5%
- C++ 15.5%
- Visual Basic .NET 10.9%
- Cuda 6.4%
- Makefile 0.2%