numerics/api/cuda__stubs_8cpp_source.html

#include "core/parallel/cuda_ops.hpp"

#include <stdexcept>


namespace num::cuda {


[[noreturn]] static void no_cuda() {

    throw std::runtime_error("CUDA not available");

}


real* alloc(idx) { no_cuda(); }

void free(real*) { no_cuda(); }

void to_device(real*, const real*, idx) { no_cuda(); }

void to_host(real*, const real*, idx) { no_cuda(); }

void scale(real*, idx, real) { no_cuda(); }

void add(const real*, const real*, real*, idx) { no_cuda(); }

void axpy(real, const real*, real*, idx) { no_cuda(); }

real dot(const real*, const real*, idx) { no_cuda(); }

void matvec(const real*, const real*, real*, idx, idx) { no_cuda(); }

void matmul(const real*, const real*, real*, idx, idx, idx) { no_cuda(); }

void thomas_batched(const real*, const real*, const real*, const real*, real*, idx, idx) { no_cuda(); }


} // namespace num::cuda

cuda_ops.hpp
CUDA kernel wrappers.

num::cuda
Definition cuda_ops.hpp:7

num::cuda::to_device
void to_device(real *dst, const real *src, idx n)
Copy host to device.
Definition cuda_stubs.cpp:12

num::cuda::scale
void scale(real *v, idx n, real alpha)
v = alpha * v
Definition cuda_stubs.cpp:14

num::cuda::thomas_batched
void thomas_batched(const real *a, const real *b, const real *c, const real *d, real *x, idx n, idx batch_size)
Batched Thomas algorithm for tridiagonal systems.
Definition cuda_stubs.cpp:20

num::cuda::matmul
void matmul(const real *A, const real *B, real *C, idx m, idx k, idx n)
C = A * B.
Definition cuda_stubs.cpp:19

num::cuda::free
void free(real *ptr)
Free device memory.
Definition cuda_stubs.cpp:11

num::cuda::alloc
real * alloc(idx n)
Allocate device memory.
Definition cuda_stubs.cpp:10

num::cuda::to_host
void to_host(real *dst, const real *src, idx n)
Copy device to host.
Definition cuda_stubs.cpp:13

num::cuda::add
void add(const real *x, const real *y, real *z, idx n)
z = x + y
Definition cuda_stubs.cpp:15

num::cuda::axpy
void axpy(real alpha, const real *x, real *y, idx n)
y = alpha*x + y
Definition cuda_stubs.cpp:16

num::cuda::dot
real dot(const real *x, const real *y, idx n)
dot product
Definition cuda_stubs.cpp:17

num::cuda::matvec
void matvec(const real *A, const real *x, real *y, idx rows, idx cols)
y = A * x (row-major A)
Definition cuda_stubs.cpp:18

num::real
double real
Definition types.hpp:10

num::ipow
constexpr T ipow(T x) noexcept
Compute x^N at compile time via repeated squaring.
Definition integer_pow.hpp:34

num::idx
std::size_t idx
Definition types.hpp:11