Getting started

Autogen supports numerical code written in C++ or Python to be traced, compiled, and executed in parallel on the CPU and GPU.

The following example demonstrates a simple function that computes the norm of a 2D vector:

C++Python

#include <autogen/autogen.hpp>

template <typename Scalar>  // Scalar denotes the number type
struct my_function {
    void operator()(const std::vector<Scalar> &input,
                    std::vector<Scalar> &output) const {
        output[0] = sqrt(input[0] * input[0] + input[1] * input[1]);
    }
};

int main(void) {
    std::vector<double> input, output(1);
    input = {0.5, 1.5};
    autogen::Generated<my_function> gen("my_function");
    gen.set_mode(autogen::GENERATE_CUDA);
    // the function gets compiled and executed on the GPU
    gen(input, output);
    return 0;
}

import autogen as ag
import numpy as np

def my_function(xs):
    return np.sqrt(xs[0]**2 + xs[1]**2)

def __main__():
    xs = np.array([0.5, 1.5])
    gen = ag.Generated("my_function", my_function)
    gen.set_mode(ag.GENERATE_CUDA)
    y = gen(xs)
    print(y)