For the complete documentation index, see llms.txt. Markdown versions of all pages are available by appending .md to any URL (e.g. /max/get-started.md).

Mojo function

reduce_add_simd

def reduce_add_simd[simd_width: SIMDLength, step_simd_width: SIMDLength, dtype: DType](mut scalar: Scalar[dtype], mut vector: SIMD[dtype, simd_width], val: SIMD[dtype, step_simd_width])

This functions adds val to either the scalar value or the vector value depending on the step_simd_width. This is useful when the simd_width varies between iterations as in vectorize.

Parameters:

simd_width (SIMDLength): The full SIMD width of the vector accumulator.
step_simd_width (SIMDLength): The width of the current step's val; when 1, val is accumulated into scalar, otherwise into vector.
dtype (DType): The element type of the accumulators and val.

Args:

scalar (Scalar[dtype]): Scalar accumulator for single-element steps; updated in place when step_simd_width is 1.
vector (SIMD[dtype, simd_width]): SIMD accumulator for full-width steps; updated in place when step_simd_width matches simd_width.
val (SIMD[dtype, step_simd_width]): The partial reduction value to accumulate into either scalar or vector.