Hello,
Recently I have installed the CUDA 6.5 for window. On the other hand I have gamess.13-64.pgi.linux.blas working on a i7 4cores and I have a NVidia GT630 (2Gb).
My intention was to include GPU to the CPU group for increase total core number but I couldn't find how can I connect gpu with gamess. I search/read the rungms file, but I did not see any option for that.
Please, could you help me with examples, references or some information that could help me.
Thank you
I am not familiar with the GAMESS toolkit,but it seems that it depends on the LIBCCHEM libraries for gpgpu computations. Perhaps your installations does not use libbchem, or you have downloaded binaries not compiled with cuda support.
In fact, The source code distribution page of GAMESS has a specific download for nvidia ( GAMESS+LIBCCHEM version May 1, 2013 R1 for NVIDIA GPUs) , so perhaps this is where you should head to.
See also this guide for compiling games with cuds support on debian, if this is of use to you.
http://www.msg.chem.iastate.edu/GAMESS/download/register/
http://verahill.blogspot.com.au/2013/05/409-failed-attempt-at-compiling-gamess_10.html
A link that could help in compilation of source files are:
http://www.chemsoft.ch/qc/SUA.htm
Thank you for your answer!
http://www.chemsoft.ch/qc/SUA.htm
Hello, Joaquim! Maybe the CUDA C Programming Guide can be useful for you. I'm not sure what you intend to, but this book is always very helpful.
Best wishes and good luck with your research!
Thank you. My intention was to use GPU for run gamess. But for the moment, it is too much difficult to implement on my system. I will wait later when this kind of systems will be easily implemented. GPU it is the future and I hope that it is a really near future.
Sure, Joaquim. I also believe that GPU is (and has been) the future. Should you have any question, please do not hesitate to contact me.
Best regards!
Thank you. I will wait for next version of gamess, and It will be fantastic that if they include an easier way for cuda and cbs methods....
You need to install visual studio 2010 or higher version. Then need to install latest version of CUDA SDK. Need to some manual setup in VS. Then you can compile some sample cuda programs.
- Badges
- Science topic