Nvc++ error: namespace "std" has no member "transform_reduce"

Accelerated Computing HPC Compilers nvc, nvc++ and nvfortran
1

Hi there, i’m trying to run a simple NVC++ parallel program but it fails to compile.

Here’s the code:

#include <iostream>
#include <execution>
#include <vector>
#include <algorithm>


int main() {
    std::vector<float> input((int)1e8, 2.0);
    std::vector<float> output(input.size());

    // std::transform(std::execution::par_unseq,input.begin(), input.end(), output.begin(), [](float f){return f*f;});
    float result = std::transform_reduce(std::execution::par_unseq, output.begin(), output.end(), 0.0f,
                                            [](float a, float b){return a+b;}, [](float f){return f*f;});



    std::cout << "Hello, World! Result is: " << result << " and it should be: " << 4.0*input.size() << std::endl;
    return 0;
}

I’m compiling with:

$ nvc++ -stdpar main.cpp -o main

On Ubuntu 20.04 LTS, with driver version 470.82.00 and Cuda Version 11.4

I’ve tried this but it doesn’t help. Thank you all for your help!

2

Hi Federico,

“transform_reduce” is defined in the “numeric” header. Adding this to your includes should fix the issue.

See: std::transform_reduce - cppreference.com

Also, the code should be using “input” instead of “output” for the arguments.

Hope this helps,
Mat