Mojo module

tensor_core_async

This module provides abstractions for using Async Tensor Cores to perform asynchronous matrix multiplication operations.

Aliases

• ​supported_mma_shape = Tuple(VariadicPack(Index(64, 8, 8), Index(64, 8, 16), Index(64, 16, 16), Index(64, 32, 16), Index(64, 64, 16), Index(64, 128, 16), Index(64, 256, 16))):
• ​WGMMA_K_BYTES = 32:

Structs

• ​TensorCoreAsync:

Functions

• ​select_k_atom:
• ​tile_layout_k_major:
• ​tile_layout_mn_major: Return the shared memory layout for mn-major input.
• ​tile_to_descriptor:
• ​wgmma_c_layout: Generates three layouts based on the WGMMA instruction dimensions and the C matrix layout.
• ​wgmma_c_thread_layout: Returns the first mode of wgmma_c_layout.
• ​wgmma_output_layout: Returns the second mode of wgmma_c_layout.

Was this page helpful?

Thank you! We'll create more content like this.

Thank you for helping us improve!

😔 What went wrong?

Some code doesn’t work

It includes inaccurate information

It's missing information I need

It was difficult to understand

Other

Submit