##SHAPE out_proj 27
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %         8.44
    gpu__time_duration.sum                                                                   ms         1.42
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          201
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        21.89
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        48.83
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        62.47
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        16.32
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle         4.60
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         0.19
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         1.67
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         0.39
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         0.94
##SHAPE out_proj 40
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %        10.02
    gpu__time_duration.sum                                                                   ms         1.20
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          200
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        25.87
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        36.95
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        73.85
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        16.30
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle         6.38
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         0.13
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         2.58
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         0.31
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         1.71
##SHAPE out_proj 36
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %         8.39
    gpu__time_duration.sum                                                                   ms         1.47
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          224
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        29.55
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        32.13
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        60.99
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        16.35
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle         5.82
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         0.39
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         1.53
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         0.25
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         1.60
##SHAPE out_proj 41
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %         9.29
    gpu__time_duration.sum                                                                   ms         1.34
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          224
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        33.19
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        26.06
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        68.52
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        16.57
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle         6.96
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         0.34
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         1.76
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         0.18
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         2.15
##SHAPE out_proj 32
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %         8.79
    gpu__time_duration.sum                                                                   ms         1.37
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          121
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        36.98
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        53.80
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        64.76
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        32.64
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle         7.75
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         0.90
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         2.66
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         1.00
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         1.32
##SHAPE out_proj 42
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %         9.52
    gpu__time_duration.sum                                                                   ms         1.27
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          123
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        40.16
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        38.65
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        70.34
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        32.59
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle        10.84
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         1.98
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         3.37
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         0.71
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         2.10
##SHAPE out_proj 35
    dram__throughput.avg.pct_of_peak_sustained_elapsed                                        %         8.05
    gpu__time_duration.sum                                                                   ms         1.55
    launch__occupancy_limit_registers                                                     block            2
    launch__occupancy_limit_shared_mem                                                    block            2
    launch__registers_per_thread                                                register/thread          228
    sm__inst_executed_pipe_lsu.avg.pct_of_peak_sustained_active                               %        29.46
    sm__pipe_alu_cycles_active.avg.pct_of_peak_sustained_active                               %        31.46
    sm__pipe_tensor_op_imma_cycles_active.avg.pct_of_peak_sustained_active                    %        58.00
    sm__warps_active.avg.pct_of_peak_sustained_active                                         %        16.61
    smsp__average_warp_latency_per_inst_issued.ratio                                      cycle         5.74
    smsp__average_warps_issue_stalled_barrier_per_issue_active.ratio                       inst         0.48
    smsp__average_warps_issue_stalled_math_pipe_throttle_per_issue_active.ratio            inst         1.12
    smsp__average_warps_issue_stalled_not_selected_per_issue_active.ratio                  inst         0.17
    smsp__average_warps_issue_stalled_wait_per_issue_active.ratio                          inst         1.92
