Is there any gemm kernel using distributed shared memory?
I have profiled many gemm kernel of hopper using nsight compute, but not found there's no data transfer using distributed shared memory.
example: cutlass/examples/hopper_gemm

• 48_hopper_warp_specialized_gemm
• 49_hopper_gemm_with_collective_builder
• ...63_hopper_gemm_with_weight_prefetch
• ./python/CuTeDSL/hopper/*
• ./cute/tutorial/hopper/*

Thus I want to know if cutlass has implemented distributed shared memory in gemm kernel?