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

Mojo struct

ReduceMax

struct ReduceMax[dtype: DType, W: Int = simd_width_of[dtype]()]

Max reduction monoid: (self, x) -> max(self, x).

Single-field state: acc: SIMD[dtype, W]. accumulate is one expression — acc = max(acc, Self.pad[...](val)). The horizontal collapse to scalar happens once at reduce, in acc[0].

Parameters

• ​dtype (DType): The accumulator dtype.
• ​W (Int): SIMD width of the lane-wise accumulator. Defaults to the target's simd_width_of[dtype].

Fields

• ​acc (SIMD[dtype, W]): Lane-wise SIMD accumulator. reduce reduces to scalar (placed in acc[0]); bodies read acc[0].

Implemented traits

AnyType, Copyable, Deinitable, ImplicitlyCopyable, Movable, ReduceOp, RegisterPassable, TrivialRegisterPassable

comptime members

Single

comptime Single = ReduceMax[dtype, Int(1)]

width

comptime width = W

Methods

__init__

def __init__() -> Self

Identity: acc = MIN_W (False for bool: max is logical OR).

__getitem__

def __getitem__(self, j: Int) -> Self.Single

Returns lane j as a width-1 monoid.

Returns:

Self.Single

__setitem__

def __setitem__(mut self, j: Int, s: ReduceMax[dtype, Int(1)])

Writes width-1 monoid s into lane j.

accumulate

def accumulate[val_dtype: DType, w: Int](mut self, val: SIMD[val_dtype, w], idx: SIMD[DType.int64, w] = 0)

Folds a SIMD tile into acc lane-wise. Partial tiles get identity-padded (min_finite) via Self.pad.

Parameters:

• ​val_dtype (DType): The tile's source dtype.
• ​w (Int): SIMD width of the tile.

Args:

• ​val (SIMD[val_dtype, w]): The SIMD tile to fold.
• ​idx (SIMD[DType.int64, w]): Unused (Max is index-agnostic).

join

def join(mut self, other: Self)

Sequential combine: lane-wise max.

Args:

• ​other (Self): The state to combine into self.

reduce

def reduce(self) -> Self.Single

Maxes the W lane partials via the reduce_max intrinsic.

Overrides the default lane-fold so the vectorized horizontal max is emitted explicitly (a measured CPU win; GPU-neutral). bool reduces through uint8 because the FP reduce_max intrinsic rejects <N x i1> on GPU.

Returns:

Self.Single: A width-1 ReduceMax holding the maximum in acc[0].

join_parallel

def join_parallel[R: Reducer](mut self, reducer: R)

Cross-thread combine via reducer.max, splatting the scalar result across all lanes so bodies read acc.slice[w] uniformly. Runs after reduce.

Parameters:

• ​R (Reducer): The parallel scalar reducer.

Args:

• ​reducer (R): The reducer instance.