ce-matrix

c++20 compile time (constexpr), header only and expression template matrix library (WIP Expect api to change)

/*
0 1 2   1 1 1
3 4 5 * 1 1 1  
6 7 8   1 1 1
*/
int main()
{
    constexpr ce::vari::matrix_t<int, 3, 3> A{ 0, 1, 2,
                                               3, 4, 5, 
                                               6, 7, 8};
    constexpr ce::vari::matrix_t<int, 3, 3> B{ 1, 1, 1,
                                               1, 1, 1, 
                                               1, 1, 1};
    constexpr ce::vari::matrix_t<int, 3, 3> C = A*B;
    return C[0];
}

The matrix mulitplication of A*B is evaluated at compile time
The above program is compiled to:(using gcc10.1 with flags: -S -std=c++2a -masm=intel)

main:
    push    rbp
    mov    rbp, rsp
    sub    rsp, 144
    mov    DWORD PTR [rbp-48], 0
    mov    DWORD PTR [rbp-44], 1
    mov    DWORD PTR [rbp-40], 2
    mov    DWORD PTR [rbp-36], 3
    mov    DWORD PTR [rbp-32], 4
    mov    DWORD PTR [rbp-28], 5
    mov    DWORD PTR [rbp-24], 6
    mov    DWORD PTR [rbp-20], 7
    mov    DWORD PTR [rbp-16], 8
    mov    DWORD PTR [rbp-96], 1
    mov    DWORD PTR [rbp-92], 1
    mov    DWORD PTR [rbp-88], 1
    mov    DWORD PTR [rbp-84], 1
    mov    DWORD PTR [rbp-80], 1
    mov    DWORD PTR [rbp-76], 1
    mov    DWORD PTR [rbp-72], 1
    mov    DWORD PTR [rbp-68], 1
    mov    DWORD PTR [rbp-64], 1
    mov    DWORD PTR [rbp-144], 3
    mov    DWORD PTR [rbp-140], 3
    mov    DWORD PTR [rbp-136], 3
    mov    DWORD PTR [rbp-132], 12
    mov    DWORD PTR [rbp-128], 12
    mov    DWORD PTR [rbp-124], 12
    mov    DWORD PTR [rbp-120], 21
    mov    DWORD PTR [rbp-116], 21
    mov    DWORD PTR [rbp-112], 21
    lea    rax, [rbp-144]
    mov    esi, 0
    mov    rdi, rax
    call    _ZNK2ce8ctmatrixIiLm3ELm3EEixEm
    mov    eax, DWORD PTR [rax]
    leave
    ret

Dependancies

• c++20 compliant compiler
  • currently working on gcc-10.1, more compilers will be supported as they get better c++20 supportDependancies
• for benchmarks and examples
  • nanobench

Examples

using namespace ce::vari;
constexpr matrix_t<float, 3, 3> A{ 1, 2, 3, 4, 5, 6, 7, 8, 9 };
constexpr matrix_t<float, 3, 3> B{ 1, 2, 3, 4, 5, 6, 7, 8, 9 };
constexpr matrix_t<float, 3, 3> C{ 1, 2, 3, 4, 5, 6, 7, 8, 9 };
//binary operators
constexpr matrix_t<float, 3, 3> D = A + B + C;
constexpr matrix_t<float, 3, 3> E = A * B - C;
constexpr matrix_t<float, 3, 3> E2 = A * 2;
//
constexpr vector_t<int, 3> v1{ 1, 2, 3 };
constexpr vector_t<int, 3> v2{ 2, 3, 4 };
constexpr auto v3 = ce::vari::dot_product(v1, v2);
static_assert(v3 == 20, "error wrong dot product value");