Mojo module
functional
Implements higher-order functions.
You can import these APIs from the algorithm package. For example:
from algorithm import mapcomptime values
BinaryTile1DTileUnitFunc
comptime BinaryTile1DTileUnitFunc = fn[width: Int](Int, Int) capturing -> None
Signature of a tiled function that performs some work with a dynamic tile size and a secondary static tile size.
Dynamic1DTileUnitFunc
comptime Dynamic1DTileUnitFunc = fn(Int, Int) capturing -> None
Signature of a 1D tiled function with dynamic tile size.
The function takes a dynamic tile size and an offset argument,
i.e. func(offset: Int, tile_size: Int).
Dynamic1DTileUnswitchUnitFunc
comptime Dynamic1DTileUnswitchUnitFunc = fn[sw: Bool](Int, Int, Int) capturing -> None
Signature of a dynamic tiled unswitch unit function.
Static1DTileUnitFunc
comptime Static1DTileUnitFunc = fn[width: Int](Int) capturing -> None
Signature of a 1D tiled function with static tile size.
The function takes a static tile size parameter and an offset argument,
i.e. func[tile_size: Int](offset: Int).
Static1DTileUnitFuncWithFlag
comptime Static1DTileUnitFuncWithFlag = fn[width: Int, flag: Bool](Int) capturing -> None
Signature of a tiled function with a static tile size, offset, and flag.
Static1DTileUnitFuncWithFlags
comptime Static1DTileUnitFuncWithFlags = fn[width: Int, left_flag: Bool, right_flag: Bool](Int) capturing -> None
Signature of a tiled function with left and right boundary flags.
Static1DTileUnswitchUnitFunc
comptime Static1DTileUnswitchUnitFunc = fn[width: Int, sw: Bool](Int, Int) capturing -> None
Signature of a tiled function with static tile size and unswitch flag.
The function takes a static tile size parameter and offset arguments,
i.e. func[tile_size: Int](offset: Int).
Static2DTileUnitFunc
comptime Static2DTileUnitFunc = fn[tile_x: Int, tile_y: Int](Int, Int) capturing -> None
Signature of a 2D tiled function with static tile size.
The function takes static tile size parameters and offset arguments, i.e.
func[tile_size_x: Int, tile_size_y: Int](offset_x: Int, offset_y: Int).
stencil
comptime stencil = _stencil_impl_cpu
CPU implementation of stencil computation.
stencil_gpu
comptime stencil_gpu = _stencil_impl_gpu
GPU implementation of stencil computation.
SwitchedFunction
comptime SwitchedFunction = fn[sw: Bool]() raises capturing -> None
Signature of a function that unswitch can take.
SwitchedFunction2
comptime SwitchedFunction2 = fn[sw0: Bool, sw1: Bool]() capturing -> None
Signature for unswitch supporting 2 predicates.
Functions
-
elementwise: Executesfunc[width, rank](indices), possibly as sub-tasks, for a suitable combination of width and indices so as to cover shape. Returns when all sub-tasks have completed. -
map: Maps a function over the integer range [0, size). This lets you apply an integer index-based operation across data captured by the mapped function (for example, an indexed buffer). -
parallelize: Executes func(0) ... func(num_work_items-1) as sub-tasks in parallel, and returns when all are complete. -
parallelize_over_rows: Parallelize func over non-axis dims of shape. -
sync_parallelize: Executes func(0) ... func(num_work_items-1) as parallel sub-tasks, and returns when all are complete. -
tile: A generator that launches work groups in specified list of tile sizes. -
tile_and_unswitch: Performs time and unswitch functional transformation. -
tile_middle_unswitch_boundaries: Divides 1d iteration space into three parts and tiles them with different steps. -
unswitch: Performs a functional unswitch transformation. -
vectorize: Simplifies SIMD optimized loops by mapping a function across a range from 0 tosize, incrementing bysimd_widthat each step. The remainder ofsize % simd_widthwill run in separate iterations.
Was this page helpful?
Thank you! We'll create more content like this.
Thank you for helping us improve!