Mojo function

grouped_matmul_amd_kernel_launcher

grouped_matmul_amd_kernel_launcher[c_type: DType, a_type: DType, b_type: DType, layout_c: Layout, layout_a: Layout, layout_b: Layout, transpose_b: Bool, config: MatmulConfig[a_type, b_type, c_type, transpose_b], elementwise_lambda_fn: OptionalReg[elementwise_epilogue_type] = None](c_tensor: LayoutTensor[c_type, layout_c, MutAnyOrigin], a_tensor: LayoutTensor[a_type, layout_a, MutAnyOrigin], b_tensor: LayoutTensor[b_type, layout_b, MutAnyOrigin], a_offsets: NDBuffer[DType.uint32, 1, MutAnyOrigin], expert_ids: NDBuffer[DType.int32, 1, MutAnyOrigin], num_active_experts: Int)