dcreager

# -*- coding: utf-8 -*-
# Author: Douglas Creager <dcreager@dcreager.net>
# This file is placed into the public domain.

# Calculates the current version number.  If possible, this is the
# output of “git describe”, modified to conform to the versioning
# scheme that setuptools uses.  If “git describe” returns an error
# (most likely because we're in an unpacked copy of a release tarball,
# rather than in a git working copy), then we fall back on reading the
# contents of the RELEASE-VERSION file.

lilac

/*
The following code inverts the matrix input using LU-decomposition with backsubstitution of unit vectors. Reference: Numerical Recipies in C, 2nd ed., by Press, Teukolsky, Vetterling & Flannery.

you can solve Ax=b using three lines of ublas code:

permutation_matrix<> piv;
lu_factorize(A, piv);
lu_substitute(A, piv, x); 

*/

ddemidov

hello: hello.cpp
	g++ -std=c++0x -o hello hello.cpp -lOpenCL

ddemidov

VEXCL_ROOT=/usr/include

hello: hello.cpp
	g++ -std=c++0x -o hello hello.cpp -I$(VEXCL_ROOT) -lOpenCL -lboost_system

ddemidov

run:
	@g++ -E pp.cpp | tail -n 1 | clang-format | pygmentize -l cpp

check:
	@g++ -E pp.cpp

jnothman

ddemidov

hello: hello.cpp
	g++ -std=c++0x -O3 -o hello hello.cpp -I$(VEXCL_ROOT) \
	    -lGL -lglut -lGLEW -lOpenCL -lboost_system

ddemidov

Design

Boost.Compute provides a thin C++ wrapper around OpenCL host API at its core and builds a set of STL-like algorithms on top of that core. The user interface strongly resembles the STL and hence should be familiar to any C++ programmer.

The only minor problem I have with the design is the decision to provide another C++ wrapper for OpenCL host API instead of using the standard C++ bindings header [1] provided by the Khronos group (the body behind the OpenCL standard). This decision makes interaction with the existing OpenCL libraries

ddemidov

#include <iostream>
#include <vexcl/vexcl.hpp>

void bar() {
    vex::Context ctx(vex::Filter::Env);
    const size_t n = 1024;
    vex::vector<int> x(ctx, n), y(ctx, n);

    VEX_FUNCTION(int, baz, (int, idx),
            return incomprehensible_gibberish(idx);

ddemidov

PyAMG
----------------

multilevel_solver
Number of Levels:     5
Operator Complexity:  1.338
Grid Complexity:      1.188
Coarse Solver:        'pinv2'
  level   unknowns     nonzeros
    0      1048576      5238784 [74.73%]