I am running MD simulations with your implementation of periodic GFN-FF on GULP. Thank you so much for that, by the way! With other codes, the usage of GPUs is really nice in terms of efficiency, so we can perform longer MD simulations for big systems.
Is there any effort on making a version of GULP for using GPUs?
Thanks for your message and glad that you are finding pGFN-FF useful. We’re also increasingly using GPUs for our MD simulations, though via OpenMM (one of our in press papers even mentions GPUs in the title). I’m afraid at the moment there is no work on a GPU version of GULP - sorry. Some parts could be ported more readily to GPUs (such as the electrostatics), but some parts would take more doing & as I’m not a GPU programmer there would be a steep learning curve.