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 function

foreach

def foreach[dtype: DType, rank: Int, //, func: def[width: Int](Coord[*?]) capturing thin -> SIMD[dtype, width], *, target: StringSpan[ImmStaticOrigin] = StringSpan("cpu"), simd_width: Int = get_kernel_simd_width[dtype, target](), _trace_name: StringSpan[ImmStaticOrigin] = StringSpan("mogg.for_each")](tensor: ManagedTensorSlice[static_spec=tensor.static_spec], ctx: DeviceContext)

Apply the function func to each element of the tensor slice.

The func body receives the element index as a Coord. Use coord_to_index_list to convert it to an IndexList if integer index arithmetic is needed.

Parameters:

• ​dtype (DType): The data type of the elements in the tensor slice.
• ​rank (Int): The rank of the tensor slice.
• ​func (def[width: Int](Coord[*?]) capturing thin -> SIMD[dtype, width]): The function to apply to each element of the tensor slice.
• ​target (StringSpan[ImmStaticOrigin]): Indicates the type of the target device (e.g. "cpu", "gpu").
• ​simd_width (Int): The SIMD width for the target (usually leave this as its default value).
• ​_trace_name (StringSpan[ImmStaticOrigin]): Name of the executed operation displayed in the trace_description.

Args:

• ​tensor (ManagedTensorSlice[static_spec=tensor.static_spec]): The output tensor slice which receives the return values from func.
• ​ctx (DeviceContext): The call context (forward this from the custom operation).

def foreach[dtype: DType, rank: Int, //, FuncType: def[width: Int](Coord[*?]) -> SIMD[dtype, width] & RegisterPassable & ImplicitlyCopyable, *, target: StringSpan[ImmStaticOrigin] = StringSpan("cpu"), simd_width: Int = get_kernel_simd_width[dtype, target](), _trace_name: StringSpan[ImmStaticOrigin] = StringSpan("mogg.for_each")](var func: FuncType, tensor: ManagedTensorSlice[static_spec=tensor.static_spec], ctx: DeviceContext)

Apply a RegisterPassable body to each element of the tensor slice.

Value-argument twin of the foreach overload above: the body is a runtime closure passed by value rather than a comptime parameter, so callers write a unified closure instead of a capturing one. The body receives the element index as a Coord; use coord_to_index_list to convert it to an IndexList if integer index arithmetic is needed.

The wrapper captures both func and tensor and routes the store through tensor._fused_store, which preserves the fused store path that FusedOutputTensor depends on.

Parameters:

• ​dtype (DType): The data type of the elements in the tensor slice.
• ​rank (Int): The rank of the tensor slice.
• ​FuncType (def[width: Int](Coord[*?]) -> SIMD[dtype, width] & RegisterPassable & ImplicitlyCopyable): The type of the per-element body closure.
• ​target (StringSpan[ImmStaticOrigin]): Indicates the type of the target device (e.g. "cpu", "gpu").
• ​simd_width (Int): The SIMD width for the target (usually leave this as its default value).
• ​_trace_name (StringSpan[ImmStaticOrigin]): Name of the executed operation displayed in the trace_description.

Args:

• ​func (FuncType): The function to apply to each element of the tensor slice.
• ​tensor (ManagedTensorSlice[static_spec=tensor.static_spec]): The output tensor slice which receives the return values from func.
• ​ctx (DeviceContext): The call context (forward this from the custom operation).