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 package
amd
AMD GPU convolution kernels.
Architecture-specific implementations live under rdna/ (RDNA 3+).
Packages
-
rdna: RDNA Conv2D via implicit GEMM (fused im2col + WMMA matmul).
Modules
-
amd_4wave_conv: 4-wave FP8 implicit-GEMM convolution for AMD MI355X (CDNA4). -
amd_4wave_conv_residual: AMD 4-wave Conv2D fprop with optional residual add. -
dispatch: AMD MI355X (CDNA4, gfx950) conv2d dispatch toamd_4wave_conv. -
dispatch_3d: AMD MI355X (CDNA4, gfx950) conv3d dispatch toamd_4wave_conv.