Dear Brian, dear Dan,
I am a former PhD student in Prof. Liedl's lab, and did some work on the GIST codebase. As far as I know, there are no plans for a GPU implementation of PME-GIST, but I am sure it would be highly appreciated.
To get optimal performance, one would probably have to implement both the entropy and the PME part on the GPU. The entropy should be the easier part. For the PME energy, GIST currently uses the helPME library. If you plan on implementing PME on the GPU, you could think about adding it to helPME, to make it available to other projects as well.
Best,
Franz
Am 5. Mai 2023 19:29:54 MESZ schrieb Brian Radak via AMBER <amber.ambermd.org>:
>I'm aware GIST PME is not currently GPU accelerated, I was just curious if
>this is on anyone's to-do or wish list (Dan?).
>
>No promises, but I could potentially arrange support of some kind of that's
>what is needed besides time.
>
>Brian
>_______________________________________________
>AMBER mailing list
>AMBER.ambermd.org
>http://lists.ambermd.org/mailman/listinfo/amber
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sun May 07 2023 - 04:00:02 PDT
