Re: [AMBER] Cpptraj: std::bad_alloc error

From: Pratul Agarwal <pratul.agarwal-lab.org>
Date: Wed, 9 Jan 2019 17:27:41 +0000

Hi Dan,

If its not too painful, I would like to add my request for adding disk caching feature to matrix (mwcovar/correl) command in CPPTRAJ. With higher speed SSDs/NVMe becoming affordable, the feature could help analysis of larger systems (or more frames).

Pratul

Pratul K. Agarwal, Ph.D.
(Editorial Board Member: PLoS ONE, Microbial Cell Factories)
Web: http://www.agarwal-lab.org/


On 1/9/2019 9:48 AM, Daniel Roe wrote:

Hi there,

On Wed, Jan 9, 2019 at 6:34 AM Nikolay N. Kuzmich <nnkuzmich.gmail.com><mailto:nnkuzmich.gmail.com> wrote:



The amount of memory should be sufficient.
Could you please tell me what is the possible source of trouble here?



A 2D RMS plot in CPPTRAJ is stored as an upper-triangle matrix, the
size of which is (Nframes*(Nframes-1))/2. Since you have 800000
frames, this translates to 319999600000 elements, which requires about
1.2 TB to store in single-precision floating point representation (I
should probably add output of an estimate for the size to the 2drms
command).

I'm currently working on adding disk-cached data sets to CPPTRAJ, but
it's not ready for prime time. Your best bet in this case is to make
one or more 2D RMS plots, choosing different subsets of frames. Using
every 10th frame will require about 12 GB, which may be more feasible
for you. Hope this helps,

-Dan

_______________________________________________
AMBER mailing list
AMBER.ambermd.org<mailto: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 Wed Jan 09 2019 - 09:30:03 PST
Custom Search