Function cudarc::cublas::result::sgemv

source ·
pub unsafe fn sgemv(
    handle: cublasHandle_t,
    trans: cublasOperation_t,
    m: c_int,
    n: c_int,
    alpha: *const f32,
    a: *const f32,
    lda: c_int,
    x: *const f32,
    incx: c_int,
    beta: *const f32,
    y: *mut f32,
    incy: c_int
) -> Result<(), CublasError>
Expand description

Single precision matrix vector multiplication. See nvidia docs

Safety

  • a, x, and y must be valid device pointers that have not been freed.
  • alpha and beta can be pointers to host memory, but must be not null
  • the strides and sizes must be sized correctly