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

small_reduce_kernel

def small_reduce_kernel[rank: Int, axis: Int, num_reductions: Int, BLOCK_SIZE: Int, input_fn: def[dtype: DType, width: Int, rank: Int](IndexList[rank]) capturing thin -> SIMD[dtype, width], output_fn: def[dtype: DType, width: SIMDLength, rank: Int](IndexList[rank], StaticTuple[SIMD[dtype, width], num_reductions]) capturing thin -> None, reduce_fn: def[ty: DType, width: SIMDLength, reduction_idx: Int](SIMD[ty, width], SIMD[ty, width]) capturing thin -> SIMD[ty, width], dtype: DType, simd_width: Int, accum_type: DType = get_accum_type[dtype]()](shape: IndexList[rank], init: StaticTuple[Scalar[dtype], num_reductions])

GPU kernel optimized for rows smaller than the warp size. Each warp reduces an entire row independently, allowing multiple rows to be reduced per block without shared-memory synchronization.

Parameters:

rank (Int): The tensor rank.
axis (Int): The axis along which to reduce.
num_reductions (Int): The number of fused reductions to perform.
BLOCK_SIZE (Int): The number of threads per block.
input_fn (def[dtype: DType, width: Int, rank: Int](IndexList[rank]) capturing thin -> SIMD[dtype, width]): The lambda to load input elements.
output_fn (def[dtype: DType, width: SIMDLength, rank: Int](IndexList[rank], StaticTuple[SIMD[dtype, width], num_reductions]) capturing thin -> None): The lambda to store output elements.
reduce_fn (def[ty: DType, width: SIMDLength, reduction_idx: Int](SIMD[ty, width], SIMD[ty, width]) capturing thin -> SIMD[ty, width]): The binary reduction function.
dtype (DType): The data type of the elements.
simd_width (Int): The SIMD vector width.
accum_type (DType): The accumulator data type.

Args:

shape (IndexList[rank]): The shape of the input tensor.
init (StaticTuple[Scalar[dtype], num_reductions]): The identity values for each reduction.