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 module

runtime_tuple

Provides the RuntimeTuple data structure and related utility functions for handling tuple-like data with both compile-time and runtime elements. RuntimeTuple is designed for high-performance tensor operations, supporting efficient manipulation of multi-dimensional data structures like shapes, indices, and coordinates.

Key features:

• Hybrid compile-time/runtime value handling
• Optimized for parallel execution and hardware acceleration
• Support for nested tuple structures
• Efficient conversion between linear indices and multi-dimensional coordinates
• Specialized operations for tensor shape calculations

The module includes functions for tuple manipulation (concatenation, flattening), coordinate transformations (idx2crd, crd2idx), and specialized tensor operations like shape division and prefix products.

Structs

• ​RuntimeTuple: A struct representing tuple-like data with compile-time and runtime elements. RuntimeTuple combines static (compile-time) and dynamic (runtime) handling of tuple-like data structures, typically used for tensor shapes, indices, and coordinates in high-performance computing contexts. This struct is optimized for parallel execution and hardware acceleration, allowing efficient manipulation of multi-dimensional data. It supports both known compile-time values and runtime-determined values.

Functions

• ​coalesce_nested_tuple: Coalesces a nested RuntimeTuple into a single-level RuntimeTuple, by multiplying all the values together.
• ​concat: Concatenates two IntTuple instances into a single IntTuple.
• ​crd2idx: Converts multi-dimensional coordinates to a linear index.
• ​idx2crd: Converts a linear index to multi-dimensional coordinates. This function transforms a flat index into coordinate values based on the provided shape and stride information. This is essential for mapping linear memory accesses to multi-dimensional tensor elements.
• ​is_int: Determines if a RuntimeTuple represents a scalar integer value.
• ​is_tuple: Determines if a RuntimeTuple represents a tuple rather than a scalar value.
• ​prefix_product: Computes the prefix products of elements in the RuntimeTuple.
• ​product: Computes the product of all elements in the RuntimeTuple.
• ​shape_div: Performs specialized shape division between RuntimeTuples.
• ​signum: Returns the sign of an integer value.
• ​to_index_list: Converts a RuntimeTuple to an IndexList with the same values.