ICODE-ADC/lammps
4
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

git-svn-id: svn://svn.icms.temple.edu/lammps-ro/trunk@10206 f3b2605a-c512-4ea7-a41b-209d697bcdaa

Browse Source
This commit is contained in:
sjplimp
2013-06-29 21:40:58 +00:00
parent 64a5e8313f
commit 31f46ba045
2 changed files with 46 additions and 36 deletions
Download Patch File Download Diff File Expand all files Collapse all files

41
doc/Section_accelerate.html
View File

@ -101,28 +101,33 @@ simply trying them out.
<LI>processor command for layout <LI>processor command for layout
<LI>OMP when lots of cores <LI>OMP when lots of cores
</UL> </UL>
<P>2-FFT PPPM, also called <I>analytic differentiation</I> or <I>ad</I> PPPM, uses 2 FFTs <P>2-FFT PPPM, also called <I>analytic differentiation</I> or <I>ad</I> PPPM, uses
instead of the 4 FFTs used by the default <I>ik differentiation</I> PPPM. However, 2 FFTs instead of the 4 FFTs used by the default <I>ik differentiation</I>
2-FFT PPPM also requires a slightly larger mesh size to achieve the same accuracy PPPM. However, 2-FFT PPPM also requires a slightly larger mesh size to
as 4-FFT PPPM. For problems where the FFT cost is the performance bottleneck (typically achieve the same accuracy as 4-FFT PPPM. For problems where the FFT
large problems running on many processors), 2-FFT PPPM may be faster than 4-FFT PPPM. cost is the performance bottleneck (typically large problems running
on many processors), 2-FFT PPPM may be faster than 4-FFT PPPM.
</P> </P>
<P>Staggered PPPM performs calculations using two different meshes, one shifted slightly with <P>Staggered PPPM performs calculations using two different meshes, one
respect to the other. This can reduce force aliasing errors and increase the accuracy of the shifted slightly with respect to the other. This can reduce force
method, but also doubles the amount of work required. For high relative accuracy, using staggered aliasing errors and increase the accuracy of the method, but also
PPPM allows one to half the mesh size in each dimension as compared to regular PPPM, doubles the amount of work required. For high relative accuracy, using
which can give around a 4x speedup in the kspace time. However, for low relative staggered PPPM allows one to half the mesh size in each dimension as
accuracy, using staggered PPPM gives little benefit and can be up to 2x slower in the compared to regular PPPM, which can give around a 4x speedup in the
kspace time. For example, the rhodopsin benchmark was run on a single processor, kspace time. However, for low relative accuracy, using staggered PPPM
and results for kspace time vs. relative accuracy for the different methods are shown gives little benefit and can be up to 2x slower in the kspace
in the figure below. For this system, staggered PPPM (using ik differentiation) time. For example, the rhodopsin benchmark was run on a single
becomes useful when using a relative accuracy of slightly greater than 1e-5 and above. processor, and results for kspace time vs. relative accuracy for the
different methods are shown in the figure below. For this system,
staggered PPPM (using ik differentiation) becomes useful when using a
relative accuracy of slightly greater than 1e-5 and above.
</P> </P>
<CENTER><IMG SRC = "JPG/rhodo_staggered.jpg"> <CENTER><IMG SRC = "JPG/rhodo_staggered.jpg">
</CENTER> </CENTER>
<P>IMPORTANT NOTE: Using staggered PPPM may not give the same increase in accuracy of energy and pressure <P>IMPORTANT NOTE: Using staggered PPPM may not give the same increase in
as it does in forces, so some caution must be used if energy and/or pressure are quantities of interest, such accuracy of energy and pressure as it does in forces, so some caution
as when using a barostat. must be used if energy and/or pressure are quantities of interest,
such as when using a barostat.
</P> </P>
<HR> <HR>

41
doc/Section_accelerate.txt
View File

@ -97,28 +97,33 @@ load-balancing: balance and fix balance
processor command for layout processor command for layout
OMP when lots of cores :ul OMP when lots of cores :ul
2-FFT PPPM, also called {analytic differentiation} or {ad} PPPM, uses 2 FFTs 2-FFT PPPM, also called {analytic differentiation} or {ad} PPPM, uses
instead of the 4 FFTs used by the default {ik differentiation} PPPM. However, 2 FFTs instead of the 4 FFTs used by the default {ik differentiation}
2-FFT PPPM also requires a slightly larger mesh size to achieve the same accuracy PPPM. However, 2-FFT PPPM also requires a slightly larger mesh size to
as 4-FFT PPPM. For problems where the FFT cost is the performance bottleneck (typically achieve the same accuracy as 4-FFT PPPM. For problems where the FFT
large problems running on many processors), 2-FFT PPPM may be faster than 4-FFT PPPM. cost is the performance bottleneck (typically large problems running
on many processors), 2-FFT PPPM may be faster than 4-FFT PPPM.
Staggered PPPM performs calculations using two different meshes, one shifted slightly with Staggered PPPM performs calculations using two different meshes, one
respect to the other. This can reduce force aliasing errors and increase the accuracy of the shifted slightly with respect to the other. This can reduce force
method, but also doubles the amount of work required. For high relative accuracy, using staggered aliasing errors and increase the accuracy of the method, but also
PPPM allows one to half the mesh size in each dimension as compared to regular PPPM, doubles the amount of work required. For high relative accuracy, using
which can give around a 4x speedup in the kspace time. However, for low relative staggered PPPM allows one to half the mesh size in each dimension as
accuracy, using staggered PPPM gives little benefit and can be up to 2x slower in the compared to regular PPPM, which can give around a 4x speedup in the
kspace time. For example, the rhodopsin benchmark was run on a single processor, kspace time. However, for low relative accuracy, using staggered PPPM
and results for kspace time vs. relative accuracy for the different methods are shown gives little benefit and can be up to 2x slower in the kspace
in the figure below. For this system, staggered PPPM (using ik differentiation) time. For example, the rhodopsin benchmark was run on a single
becomes useful when using a relative accuracy of slightly greater than 1e-5 and above. processor, and results for kspace time vs. relative accuracy for the
different methods are shown in the figure below. For this system,
staggered PPPM (using ik differentiation) becomes useful when using a
relative accuracy of slightly greater than 1e-5 and above.
:c,image(JPG/rhodo_staggered.jpg) :c,image(JPG/rhodo_staggered.jpg)
IMPORTANT NOTE: Using staggered PPPM may not give the same increase in accuracy of energy and pressure IMPORTANT NOTE: Using staggered PPPM may not give the same increase in
as it does in forces, so some caution must be used if energy and/or pressure are quantities of interest, such accuracy of energy and pressure as it does in forces, so some caution
as when using a barostat. must be used if energy and/or pressure are quantities of interest,
such as when using a barostat.
:line :line