sgemv

void sgemv(bool transA, int M, int N, float alpha, const float * A, int lda, const float * x, int incx, float beta, float * y, int incy)

source edit

Single-precision general matrix-vector multiply (GEMV).

Computes $y \leftarrow \alpha (A x) + \beta y$ in row-major layout via Accelerate's cblas_sgemv. Significantly cheaper than sgemm for vector right-hand sides because the inner loop stays in cache.

Math

$y_i \leftarrow \alpha \sum_{j=0}^{N-1} A_{ij} x_j + \beta y_i$

References

Accelerate.framework cblas_sgemv.

Parameters

transAbool

If true, compute

y \leftarrow \alpha A^T x + \beta y

instead.

M, Nint

A \in \mathbb{R}^{M \times N}

. When transA is false,

x

has

N

elements and

y

has

M

; reversed when transA is true.

alpha, betafloat

Linear-combination coefficients.

Aconst float*

Row-major matrix buffer.

xconst float*

Input vector buffer (stride incx).

yfloat*

Output vector buffer (stride incy); updated in place.

ldaint

Leading dimension (row stride) of

A

incx, incyint

Element stride within the input/output vectors.