Question about Compute Workload Analysis

Question about Compute Workload Analysis on Nsight Compute

It shows Pipeline Utilization (LSU, ALU, FMA, Uniform, CBU, ADU, FP16, FP64, TEX, Tensor(DP), Tensor (FP), Tensor (INT), XU)
Is there any documentaion about describe these?

I found the description found on StackOverflow. but I want to see the description on NVIDIA document.

They are described here: Kernel Profiling Guide :: Nsight Compute Documentation

1 Like

Thank you for quick reply. I already found l1tex__data_bank_conflicts_pipe_lsu.sum for LSU. but digging down the page, I found the LSU which you pointed out. thanks.