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

async_copy_wait_group

def async_copy_wait_group(n: Int32)

Waits for the completion of n most recently committed cp.async-groups.

This function blocks execution until the specified number of previously committed cp.async-groups have completed their memory transfers.

Notes:

• Only supported on NVIDIA GPUs.
• Maps to the cp.async.wait.group PTX instruction.
• Provides fine-grained control over asynchronous transfer synchronization.
• Can be used to implement a pipeline of asynchronous transfers.

Args:

• ​n (Int32): The number of pending cp.async-groups to wait for. Must be > 0.