CUDA compile error on Linux, but Windows Visual Studio succeed

Question

This program can build successfully on Windows - Visual Studio 12, but I got some problem when I compile it on Linux.

Compile command : nvcc main.cu

int main(int argc, char **argv) {

    ... (Skip some code)

    // Read file.
    MemoryBuffer<char> text(size + 1); // size type is size_t
    auto text_smem = text.CreateSync(size);
    fread(text_smem.get_cpu_wo(), 1, size, fp);

    ... (Skip some code)

}

Compiler error message:

1. auto text_smem = text.CreateSync(size);

   occured about error: explicit type is missing ("int" assumed) and error: no suitable conversion function from "SyncedMemory<char>" to "int" exists

2. fread(text_smem.get_cpu_wo(), 1, size, fp);

   occured about error: expression must have class type

How to solve the compiler error in Linux OS?

---

MemoryBuffer is a class, detail code can see from here .

Answer 1

This depends on version of CUDA Toolkit, that you have installed, but I guess you could try with --std=c++11 option to the nvcc , in order to explicitely enable modern C++ features.

Reffering to https://devblogs.nvidia.com/parallelforall/power-cpp11-cuda-7/

To enable C++11 support, pass the flag --std=c++11 to nvcc (this option is not required for Microsoft Visual Studio).