tilelang.autotuner.capture module#

class tilelang.autotuner.capture.AutotuneInputsCapture(tensors: List[Any])#

Bases: object

tensors#
class tilelang.autotuner.capture.CaptureStack#

Bases: object

A simple stack implementation for capturing items in a thread-local context. Used to manage a stack of items (e.g., input tensors) for auto-tuning capture.

pop()#

Pop and return the top item from the stack.

Returns:

The item at the top of the stack.

Raises:

IndexError – If the stack is empty.

push(item)#

Push an item onto the top of the stack.

Parameters:

item – The item to be pushed onto the stack.

size()#

Return the number of items in the stack.

Returns:

The size of the stack.

Return type:

int

top()#

Return the item at the top of the stack without removing it.

Returns:

The item at the top of the stack.

Raises:

IndexError – If the stack is empty.

tilelang.autotuner.capture.get_autotune_inputs() Optional[List[Any]]#

Get the current autotune inputs from the stack.

tilelang.autotuner.capture.set_autotune_inputs(*args) AutotuneInputsCapture#

Set input tensors for auto-tuning.

This function creates a context manager for capturing input tensors during the auto-tuning process. It supports both:

set_autotune_inputs(a, b, c) set_autotune_inputs([a, b, c])

Parameters:

*args – Either a single list/tuple of tensors, or multiple tensor arguments.

Returns:

A context manager for auto-tuning inputs.

Return type:

AutotuneInputsCapture