Skip to main content

/

DocsCode

Mojo struct

Coord

@register_passable(trivial) struct Coord[*element_types: CoordLike]

A struct representing tuple-like data with compile-time and runtime elements.

Parameters​

  • ​*element_types (CoordLike): The variadic pack of element types that implement CoordLike.

Implemented traits​

AnyType, CoordLike, Copyable, Defaultable, ImplicitlyCopyable, ImplicitlyDestructible, Movable, RegisterPassable, Sized, TrivialRegisterPassable, Writable

comptime members​

__copy_ctor_is_trivial​

comptime __copy_ctor_is_trivial = True

__del__is_trivial​

comptime __del__is_trivial = True

__move_ctor_is_trivial​

comptime __move_ctor_is_trivial = True

all_dims_known​

comptime all_dims_known = #kgen.variadic.reduce(element_types, base=True, reducer=[PrevV: Variadic[Bool], VA: Variadic[CoordLike], idx: __mlir_type.index] PrevV[0] if VA[idx].is_static_value else VA[idx].is_static_value)[0]

True if all dimensions are statically known at compile time.

DTYPE​

comptime DTYPE = DType.invalid

The data type for runtime values, or invalid for compile-time values.

flat_rank​

comptime flat_rank = Variadic.size[CoordLike](#kgen.variadic.reduce(element_types, base=, reducer=[PrevV: Variadic[CoordLike], VA: Variadic[CoordLike], idx: __mlir_type.index] #kgen.variadic.concat(PrevV, VA[idx].VariadicType if VA[idx].is_tuple else VA[idx])))

The total number of leaf elements after flattening nested Coords.

is_static_value​

comptime is_static_value = False

True if the value is known at compile time.

is_tuple​

comptime is_tuple = True

True, indicating this is a tuple type.

is_value​

comptime is_value = False

True if this is a scalar value, False for tuple types.

rank​

comptime rank = Variadic.size[CoordLike](element_types)

The number of top-level elements in this Coord.

static_product​

comptime static_product = #kgen.variadic.reduce(element_types, base=1, reducer=[PrevV: Variadic[Int], VA: Variadic[CoordLike], idx: __mlir_type.index] (VA[idx].static_value * PrevV[0]))[0]

The product of all static dimensions, or -1 if any are dynamic.

static_value​

comptime static_value = -1

Always -1 for tuple types (value not applicable).

VariadicType​

comptime VariadicType = element_types

The variadic element types of this Coord.

Methods​

__init__​

__init__() -> Self

Empty initialize a tensor with static dims.

__init__[rank: Int, dtype: DType](index_list: IndexList[index_list.rank, element_type=dtype]) -> Coord[#kgen.variadic.splat(RuntimeInt[dtype], index_list.rank._mlir_value)]

Construct a Coord from an IndexList.

Parameters:

  • ​rank (Int): The number of elements in the index list.
  • ​dtype (DType): The data type of the index list elements.

Args:

  • ​index_list (IndexList): The IndexList to convert to a Coord.

Returns:

Coord

__init__(var *args: *element_types) -> Self

Construct tuple from variadic arguments.

Args:

  • ​*args (*element_types): Values for each element.

@implicit __init__(var tuple: Tuple[element_types]) -> Self

Construct from a Tuple with matching element types.

Args:

  • ​tuple (Tuple): The Tuple to construct from.

__init__(*, var storage: VariadicPack[storage.is_owned, CoordLike, element_types]) -> Self

Construct from a low-level variadic pack.

Args:

  • ​storage (VariadicPack): The variadic pack storage to construct from.

__getitem__​

__getitem__[idx: Int](ref self) -> ref[self_is_mut._storage] element_types[idx._mlir_value]

Get a reference to an element in the tuple.

Parameters:

  • ​idx (Int): The element index to access.

Returns:

ref: A reference to the specified element.

__eq__​

__eq__[*other_types: CoordLike](self, other: Coord[other_types]) -> Bool

Check if this Coord equals another.

Parameters:

  • ​*other_types (CoordLike): The element types of the other Coord.

Args:

  • ​other (Coord): The Coord to compare with.

Returns:

Bool: True if all elements are equal, False otherwise.

__ne__​

__ne__[*other_types: CoordLike](self, other: Coord[other_types]) -> Bool

Check if this Coord is not equal to another.

Parameters:

  • ​*other_types (CoordLike): The element types of the other Coord.

Args:

  • ​other (Coord): The Coord to compare with.

Returns:

Bool: True if any elements differ, False if all are equal.

size​

static size() -> Int

Get the total number of elements including nested ones.

Returns:

Int: The total count of all elements.

__len__​

static __len__() -> Int

Get the length of the tuple.

Returns:

Int: The number of elements in the tuple.

__len__(self) -> Int

Get the length of the tuple.

Returns:

Int: The number of elements in the tuple.

__repr__​

__repr__(self) -> String

Get the string representation of this Coord.

Returns:

String: A string in the format "Coord(elem1, elem2, ...)".

write_repr_to​

write_repr_to(self, mut writer: T)

Write the repr of this Coord to a writer.

Args:

  • ​writer (T): The writer to write the representation to.

product​

product(self) -> Int

Calculate the product of all elements recursively.

Returns:

Int: The product of all leaf values in the Coord.

sum​

sum(self) -> Int

Calculate the sum of all elements recursively.

Returns:

Int: The sum of all leaf values in the Coord.

value​

value(self) -> Int

Get the value (not valid for Coord tuples).

Returns:

Int: Never returns; aborts at compile time.

inner_product​

inner_product(self, t: IntTuple) -> Int

Calculate the inner product with an IntTuple.

Args:

  • ​t (IntTuple): The other value to compute inner product with.

Returns:

Int: The inner product of the two values.

inner_product[*other_types: CoordLike](self, other: Coord[other_types]) -> Int

Calculate the inner product with another CoordLike.

Parameters:

  • ​*other_types (CoordLike): The types of the other value.

Args:

  • ​other (Coord): The other value to compute inner product with.

Returns:

Int: The inner product of the two values.

tuple​

tuple(var self) -> Self

Get this Coord as a tuple.

Returns:

Self: This Coord (identity operation for tuple types).

reverse​

reverse(var self) -> Coord[#kgen.variadic.reduce(element_types, base=, reducer=[PrevV: Variadic[CoordLike], VA: Variadic[CoordLike], idx: __mlir_type.index] #kgen.variadic.concat(PrevV, VA[(add (mul idx, -1), len(VA), -1)]))]

Reverse the order of elements in this Coord.

Returns:

Coord: A new Coord with elements in reverse order.

concat​

concat[*other_element_types: CoordLike](var self, var other: Coord[other_element_types]) -> Coord[#kgen.variadic.concat(element_types, other_element_types)]

Concatenate this Coord with another.

Parameters:

  • ​*other_element_types (CoordLike): The element types of the other Coord.

Args:

  • ​other (Coord): The Coord to append.

Returns:

Coord: A new Coord containing elements from both Coords.

flatten​

flatten(var self) -> Coord[#kgen.variadic.reduce(element_types, base=, reducer=[PrevV: Variadic[CoordLike], VA: Variadic[CoordLike], idx: __mlir_type.index] #kgen.variadic.concat(PrevV, VA[idx].VariadicType if VA[idx].is_tuple else VA[idx]))]

Convert a nested Coord to a flattened Coord.

Examples:

from layout import Coord, Idx
var nested = Coord(
    Idx[5](),
    Coord(Idx[3](), Idx[2]()),
    Idx(7)
)
var flat = nested.flatten()
# flat is Coord(Idx[5](), Idx[3](), Idx[2](), Idx(7))

Returns:

Coord: A flattened Coord containing all leaf values in order.

make_dynamic​

make_dynamic[dtype: DType](self) -> Coord[#kgen.variadic.reduce(element_types, base=, reducer=[PrevV: Variadic[CoordLike], VA: Variadic[CoordLike], idx: __mlir_type.index] #kgen.variadic.concat(PrevV, RuntimeInt[dtype]))]

Convert all elements to RuntimeInt[dtype].

Examples:

from layout.coord import Coord, ComptimeInt, RuntimeInt
var c = Coord(ComptimeInt[3](), RuntimeInt[DType.int32](5), ComptimeInt[7]())
var dynamic = c.make_dynamic[DType.int64]()
# dynamic is Coord(RuntimeInt[DType.int64](3), RuntimeInt[DType.int64](5), RuntimeInt[DType.int64](7))

Parameters:

  • ​dtype (DType): The data type for the resulting RuntimeInt values.

Returns:

Coord: A new Coord where all elements are converted to RuntimeInt[dtype].

write_to​

write_to(self, mut w: T)

Write this Coord to a Writer.

Args:

  • ​w (T): The writer to output to.

View source

View source

Was this page helpful?

Models

Cookbook

Puzzles

Blog

Community

Contact