backends, api

amgcl/backend/builtin.hpp

@@ -315,7 +315,6 @@ crs<V, C, P> inverse(const crs<V, C, P> &A) {
 
 /// Builtin backend.
 /**
- * \param real Value type.
  * \ingroup backends
  */
 template <typename ValueType>

amgcl/backend/vexcl.hpp

@@ -50,7 +50,6 @@ namespace backend {
  * This is a backend that uses types defined in the VexCL GPGPU library
  * (https://github.com/ddemidov/vexcl).
  *
- * \param real Value type.
  * \ingroup backends
  */
 template <typename real>

docs/_static/custom.css

@@ -6,3 +6,10 @@ table.citation td {
     float: left;
     max-width: 85%;
 }
+
+dl.class {
+    padding-left: 5px;
+    border-left-color: #aaaaaa;
+    border-left-style: solid;
+    border-left-width: 1px;
+}

docs/api.rst

@@ -2,3 +2,11 @@ API Reference
 =============
 
 .. doxygenclass:: amgcl::make_solver
+    :members:
+
+Backends
+--------
+
+.. doxygenstruct:: amgcl::backend::builtin
+
+.. doxygenstruct:: amgcl::backend::vexcl

docs/backends.rst

@@ -1,2 +1,58 @@
 Backends
 ========
+
+A backend in AMGCL is a class that defines matrix and vector types together
+with several operations on them, such as creation, matrix-vector products,
+elementwise vector operations, inner products etc. The
+`amgcl/backend/interface.hpp`_ file defines an interface that each backend
+should extend.  The AMG hierarchy is moved to the specified backend upon
+construction. The solution phase then uses types and operations defined in the
+backend. This enables transparent acceleration of the solution phase with
+OpenMP, OpenCL, CUDA, or any other technologies.
+
+In order to use a backend, users must include its definition from the
+corresponding file inside `amgcl/backend/` folder.
+
+Builtin
+-------
+
+The builtin backend does not have any dependencies except for the Boost_
+libraries, and uses OpenMP for parallelization. Matrices are stored in the CRS
+format, and vectors are instances of ``std::vector<value_type>``. There is no
+usual overhead for moving the constructed hierarchy to the builtin backend,
+since the backend is always used internally during setup.
+
+Example:
+
+.. code-block:: cpp
+
+    #include <amgcl/backend/builtin.hpp>
+
+    ...
+
+    typedef amgcl::backend::builtin<double> Backend;
+
+    amgcl::make_solver<
+        amgcl::amg<Backend, amgcl::coarsening::smoothed_aggregation, amgcl::relaxation::spai0 >
+        amgcl::solver::cg<Backend>
+        > solve( boost::tie(n, ptr, col, val) );
+
+    std::vector<double> rhs(n), x(n);
+
+    ...
+
+    solve(rhs, x);
+
+VexCL
+-----
+
+The VexCL backend uses VexCL_ library for accelerating solution on . VexCL is a C++
+vector expression template library for OpenCL/CUDA and is able to utilize
+compute power of the modern GPUs.
+
+.. _Boost: http://www.boost.org/
+.. _VexCL: http://github.com/ddemidov/vexcl
+
+.. _`amgcl/backend/`: https://github.com/ddemidov/amgcl/blob/master/amgcl/backend/
+.. _`amgcl/backend/interface.hpp`: https://github.com/ddemidov/amgcl/blob/master/amgcl/backend/interface.hpp
+

docs/conf.py

@@ -50,7 +50,6 @@
 
 breathe_projects = {'AMGCL' : 'xml'}
 breathe_default_project = 'AMGCL'
-breathe_default_members = ['members']
 
 # Add any paths that contain templates here, relative to this directory.
 templates_path = ['_templates']