triton.experimental.gluon.language.nvidia.blackwell.tma

Functions

async_gather	Asynchronously gather elements from global memory to shared memory using TMA.
async_scatter	Asynchronously scatter elements from shared memory to global memory using TMA.
async_atomic_add	Atomically add data from shared memory into global memory using TMA.
async_atomic_and	Atomically bitwise-and data from shared memory into global memory using TMA.
async_atomic_max	Atomically compute the maximum of shared memory data and global memory using TMA.
async_atomic_min	Atomically compute the minimum of shared memory data and global memory using TMA.
async_atomic_or	Atomically bitwise-or data from shared memory into global memory using TMA.
async_atomic_xor	Atomically bitwise-xor data from shared memory into global memory using TMA.
async_copy_global_to_shared	Load data from global memory to shared memory using TMA.
async_copy_shared_to_global	Store data from shared memory to global memory using TMA.
async_load	Load data from global memory to shared memory using TMA.
async_load_im2col	Load data from global memory to shared memory using TMA in im2col mode.
async_store	Store data from shared memory to global memory using TMA.
store_wait
make_tensor_descriptor

Classes

tensor_descriptor
tensor_descriptor_type	Type for tiled tensor descriptors.