Wow...
That's a really really really simple GPU kernel. It's a bunch of frames
plus a spring constraint? There was no need here for this IMO. I have to
ask what the resistance to writing a simple streaming CUDA kernel here was?
But also, thanks for the answer, will ignore, doesn't seem important enough
to be a core component, and the API I am building would already support
something like this. Shuttle is dead.
On Fri, Aug 5, 2022 at 11:28 AM Roitberg,Adrian E <roitberg.ufl.edu> wrote:
> Look at the neb code, we used shuttle comms there.
>
> On Aug 5, 2022 2:07 PM, Scott Le Grand via AMBER-Developers <
> amber-developers.ambermd.org> wrote:
> [External Email]
>
> Besides EMIL, did any other methods use the Shuttle calls to mix CPU/GPU
> computation?
>
> In 2022, that's better handled by unified memory on SM 6++, and I think I
> have a better successor to it, but I just want to know if anyone else ended
> up using them.
> _______________________________________________
> AMBER-Developers mailing list
> AMBER-Developers.ambermd.org
>
> https://nam10.safelinks.protection.outlook.com/?url=http%3A%2F%2Flists.ambermd.org%2Fmailman%2Flistinfo%2Famber-developers&data=05%7C01%7Croitberg%40ufl.edu%7C8248f8905bc545e4768a08da770d60e1%7C0d4da0f84a314d76ace60a62331e1b84%7C0%7C0%7C637953196587837339%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=A5fIWQsI2fE6n03jIev%2F8e1BTmFVSYgh%2Fn1p2xgNgMU%3D&reserved=0
>
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Fri Aug 05 2022 - 12:00:02 PDT
