Re: [AMBER-Developers] Cryo em refinement

From: David A Case <david.case.rutgers.edu>
Date: Mon, 7 Oct 2019 12:57:40 -0400

On Mon, Oct 07, 2019, Xiongwu Wu wrote:

>I am afraid that flexible fitting would require to calculate FFT every
>time step, which could be costly.

We have generated GPU code that can compute structure factors pretty
quickly -- for some "intersting" cases this is 3-8 times slower than a
GPU run with no structure factors. And this doesn't even use FFT's, which
should provide significant speedup for many problems.

But generally, you are right: this is not an inexpensive approach.

....dac


_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Mon Oct 07 2019 - 10:00:02 PDT
Custom Search