Mojo module

pingpong_kernel

Structs

• ​AMDPingPongMatmul: High-level ping-pong matmul implementation for AMD GPUs.
• ​KernelConfig:
• ​MmaOp: Encapsulates MMA register tiles and operations for matrix multiplication.
• ​TileBuffers: Double-buffered LDS tiles and TileLoaders for ping-pong matmul.
• ​TileLoaderLDS: Encapsulates load_to_lds with pre-computed thread positions and swizzle.

Functions

• ​chiplet_transform_chunked: Transform work group ID for better chiplet locality.
• ​load_lds_fragment: Load LDS → registers with MMA access pattern.
• ​ping_pong_matmul:

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