Hi everybody,
I’m interest about the lammps code and i have read your paper named implementing molecular dynamics on hybrid high performance computers-short range forces.
Since I’m a new user of opencl and linux. What I wan’t to know first is how the neighbor list build , I think it is a basic question for learning lammps code.
I found the code in Lal_neighbor.cpp ,the function is below:
void build_nbor_list(double **x, const int inum, const int host_inum,
const int nall, Atom<numtyp,acctyp> &atom, double *sublo,
double *subhi, tagint *tag, int **nspecial, tagint **special,
bool &success, int &max_nbors);
but i can’t understand the parameters of x,inum,host_inum……,so i can’t understand the code correctly.
Also I tried the http://users.nccs.gov/~wb8/geryon/index.htm. to find some material about geryon lib but failed it seems can’t open in china or broken. Can you help me to get some material about this lib? Thanks a lot.
PS:
I want to learn the implement of opencl accelerate of lammps in points bellow:
1.neighbor list building
2.LJ force compute
3.PPPM
Could you please give some guidance suggestion, because the Lammps software is very large.Also could you tell me the detail about your develop environment (something about how you editor the code and debug i found linux seems can’t use IDE to debug the lammps code,is that right?)
Thanks very much . Best wishes.
A letter from a china student.
Hi everybody,
I'm interest about the lammps code and i have read your paper named *implementing
molecular dynamics on hybrid high performance computers-short range forces*
.
Since I'm a new user of opencl and linux. What I wan't to know first
is how the neighbor list build , I think it is a basic question
if you are new to OpenCL *don't* start by working on a big package
program. work through some toy examples, then work on some simpler code to
hone your skills. then learn about the inner code structure and programming
conventions etc. of LAMMPS on the CPU side, and *only then* you should even
consider writing code from scratch, especially something as involved and
complex as GPU/OpenCL programming.
for learning lammps code.
I found the code in Lal_neighbor.cpp ,the function is below:
void build_nbor_list(double **x, const int inum, const int
host_inum,
const int nall, Atom<numtyp,acctyp> &atom, double
*sublo,
double *subhi, tagint *tag, int **nspecial, tagint
**special,
bool &success, int &max_nbors);
but i can't understand the parameters of x,inum,host_inum……,so i can't
understand the code correctly.
Also I tried the http://users.nccs.gov/~wb8/geryon/index.htm. to find
some material about geryon lib but failed it seems can't open in china or
broken. Can you help me to get some material about this lib? Thanks a lot.
mike brown no longer works at ORNL and thus the site has been taken down.
PS:
I want to learn the implement of opencl accelerate of lammps in
points bellow:
1.neighbor list building
2.LJ force compute
3.PPPM
Could you please give some guidance suggestion, because the Lammpssoftwareis very large.Also
could you tell me the
all three of these are already available. please read the documentation
and README of the GPU package more carefully.
detail about your develop environment (something about how you editor the
code and debug i found linux seems can't use IDE to debug the lammps
code,is that right?)
people have used IDEs, but most developers i know and myself prefer a
straight "programmer's text editor" like emacs or vi.
axel.