Hello Dear Gaussian 09 User's
I am trying to optimize a geometry containing 72 atoms i.e. C, H, O and B and 336 electrons. I am doing this calculation on Intel Xeon server having 20 core / 40 threads and 32 GB RAM. In my input file I selected 40 threads and 16 GB RAM from Link0 option in Gaussview. But the optimization take much more time upto 24 hrs but still the optimization not completed. And I saw from the task manager the calculation only used 2.5 GB RAM for 40 threads although I selected 16 GB in my input. I performed the same computation on GAMESS US code using the same resources the optimization completed just in 37 minutes.
My question is why optimization take so much time on Gaussian 09 code using many cores and more RAM? Next why gaussian 09 not utilized the required RAM given in input file?
I am using WB97XD theory and 6-31+G(d,p) basis set.
Sorry I forget about OS. I am using windows 7 64 bit for both codes (Gaussian 09 and GAMESS US).
Not sure what is the exact issue in your case. However, the 32 bit version of Gaussian can use at max 2 GB or RAM (http://gaussian.com/g16wplat/) please check the version of your G09W
Thank you dear Sengupta for your answer. I am using 64 bit Gaussian 09 C-01 Version
My guess is that hyperthreading do not make your calculation faster, but slower.
Another question is how does calculation actually proceed. If it fall in oscillation it might lasts forever (till number of cycles exceeded). 37 minutes for such structure is rather fast, it sounds like one SCF iteration, may be few, so either your geometry was almost converged, either it made singlepoint and not optimisation. Gamess is normally slower than Gaussian, so something must be wrong in 1 or both of these calculations. Hard to guess without outputs.
Dear Sharapa
Thank you. I find out the issue actually there is problem in the computer hardware the RAM and Processors are not compatible with each other. I run the same calculation on QuadCore IntelXeon workstation the calculation terminated normally in a reasonable time. But still a little confusion, with GAMESS US its ok but with Gaussian that system is not ok. I checked the GAMESS US ouptut the optimization terminated normally with correct convergence criteria and it is not a single point calculation.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Atomic, Molecular and Optical Physics
- Atoms