Module elementwise 

Generic elementwise kernels (binary, binary-scalar, unary, comparison, in-place accumulate). Ported from luma-core’s arith.rs, generalized over the CpuNum element trait and returning plain Vecs.

Functions

binary

out[i] = f(lhs[i], rhs[i]) over two (possibly non-contiguous) views of the same logical shape. Mirrors luma-core compute_binary_op.

binary_

In-place dst = f(dst, src) respecting both layouts. Mirrors luma-core binary_inplace_op; used by the backward pass to accumulate gradients.

binary_scalar

out[i] = f(lhs[i], scalar).

binary_scalar_

In-place dst[i] = f(dst[i], rhs).

cmp_scalar

Out-of-place comparison against scalar: out[i] = f(lhs[i], rhs).

float_unary

Dispatch a float unary op over one view.

num_binary

Dispatch a numeric binary op over two views.

num_binary_fn

num_binary_scalar

Dispatch a numeric binary-scalar op over one view.

num_cmp

Dispatch a comparison op over two views, producing a bool buffer.

num_scalar_binary

Dispatch a numeric scalar-binary op (scalar OP element) over one view.

scalar_binary

out[i] = f(scalar, rhs[i]).

unary

out[i] = f(x[i]).

unary_

In-place dst[i] = f(dst[i]).