Architecture-specific MMA Programming Guides# This section contains architecture-specific MMA programming guides. Warp-Level MMA Instructions Programming Guide Global Memory (GMEM) to MMA data flow overview Setting up the TiledMMA, MMA Ops Partitioning Tensors Pre and Post-Conditions for Partitioning Making Fragments Executing the GEMM (Main Loop) Complete Workflow Beyond Simple Dense MMAs Warpgroup MMA Programming Guide Global Memory (GMEM) to MMA data flow overview Setting up the TiledMMA, MMA Ops Partitioning Tensors Pre and Post-Conditions for Partitioning Making Fragments Creating SMEM layouts for A and B Executing the GEMM (Main Loop) Complete Workflow tcgen05 MMA Programming Guide Global Memory (GMEM) to MMA data flow overview Setting up the TiledMMA, MMA Ops Partitioning Tensors Making Fragments Executing the GEMM (Main Loop) Reading the accumulator from TMEM Complete Workflow Beyond Simple Dense MMAs
