top of page

PARALLEL AND GPU COMPUTING

Our research group has modified MOPAC so that it can use a GPU chip.

 

Over the past year, our research group has increased the speed of MOPAC2012 by a factor of about 200, for large systems. Thus for Bacteriorhodopsin, a medium-sized protein composed of 3352 atoms, the time for a single-point calculation, i.e., a 1SCF calculation, using a PC with four cores and one GPU chip, was reduced from 2,199 minutes for the legacy version of MOPAC2012 to 11.8 minutes. If a gradient calculation is included, the time rises to 13.4 minutes. 

This speed-up can be divided into three components: 

  1. By using Intel's MKL, legacy operations have been replaced by modern equivalents. This results in a speed-up of up to 42 times, for large systems. 

  2. By using Multi-Threading, a speedup of about three is achieved, when four threads are used, and about eight, when twelve threads are used. 

  3. For large systems, the use of a GPU chip approximately doubles the speed of a normal job. 

 

 

If you have a computer with a GPU chip installed, then please try the new MOPAC. Download Windows or Linux versions of MOPAC2012 that include in the label the letters "GPU". 

 

Versions of MOPAC that labeled "CPU+GPU" will detect a GPU chip automatically. GPU versions of MOPAC require the CUDA Toolkit 5.5 to be installed. If the GPU chip is not to be used, add keyword NOGPU.

 

Versions of MOPAC that can use a GPU chip are:

  • MOPAC2012 for 64 bit Windows 7

  • MOPAC2012 for 32 bit Windows 7

  • MOPAC2012 for 64 bit Linux

 

 

Versions of MOPAC that cannot take advantage of a GPU chip are:

  • MOPAC2012 with a small run-time Window, both 32-bit and 64-bit

  • MOPAC2012 for 32 bit Linux

  • MOPAC2012 for the Macintosh

 

 

Download new executables on the MOPAC's website.

 

 

 

Lastly, If you use any parallel version of MOPAC, please cite the following paper as a primary reference to MOPAC: 

Maia, J. D. C.; Urquiza Carvalho, G. A.; Mangueira, C. P.; Santana, S. R.; Cabral, L. A. F.; Rocha, G. B., Journal of Chemical Theory and Computation 2012, 8, 3072–3081. DOI: 10.1021/ct3004645

 

bottom of page