AMD CDNA 5

async_copy
cluster
mbarrier
tdm

buffer_load	AMD buffer load from global memory via a scalar base pointer and a tensor of offsets instead of a tensor of pointers.
buffer_store	AMD buffer store a tensor directly to global memory via a scalar base pointer and a tensor of offsets instead of a tensor of pointers.
get_wmma_scale_layout	Get the scale layout for WMMA scaled operands.
load_shared_fp4_repacked	Load M/N-packed fp4 bytes from shared memory into a K-packed WMMA dot operand layout.
make_partitioned_dot_layouts	Create partitioned shared memory layouts and WMMA layout for a CDNA5 GEMM in order to avoid LDS partition conflicts.
scaled_downcast	Scale and convert FP16, BF16, or FP32 values to a low-precision MX format, dividing by the raw E8M0 scale payload (int8 or uint8).
scaled_upcast	Upcast an fp4 or fp8 tensor and fold raw E8M0 scale payload into the CDNA5 scaled-upcast op.
wmma	Computes matrix multiplication a * b + acc using an AMD WMMA instruction.
wmma_scaled	AMD Scaled WMMA operation.
PartitionedSharedLayout	Represents a partitioned shared memory layout that splits a tensor across multiple physical shared memory partitions.