About time consumption

[zhSlamer]

Thank you for your wonderful work！

I run your code on GPU Titan V , 12 GB and
set the pool_capacity: 5e6, batch_size:8192; But I don't know why my time consumption is so long. On Kitti-00, running time for 43 min. I want to ask how to adapt it to my GPU for real-time operation;

this my time details on Kitti-00:

sometime odometry time consumption over 200ms:

Thank you for your help!

[YuePanEdward]

Thank you very much for your interest in our work. The reported consuming time is similar to the operation time on my CPU. You may first check if device is accidentally set to "cpu" (default is "cuda"). If it's not the case, then you may try the following parameters:

process:
  min_range_m: 3.0
  max_range_m: 80.0
  min_z_m: -3.5
  vox_down_m: 0.12
sampler:
  surface_sample_range_m: 0.25 
  surface_sample_n: 2
  free_sample_begin_ratio: 0.5
  free_sample_end_dist_m: 1.0
  free_front_sample_n: 1
neuralpoints:
  voxel_size_m: 0.4
loss:
  sigma_sigmoid_m: 0.08
  loss_weight_on: True
  dist_weight_scale: 0.8
  ekional_loss_on: True
  weight_e: 0.5
  numerical_grad_on: True
continual:
  batch_size_new_sample: 1000
  pool_capacity: 1e7
  pool_filter_freq: 5
tracker:
  source_vox_down_m: 1.0
  iter_n: 20
  term_deg: 0.05
  term_m: 0.001
pgo:
  map_context: True
  pgo_freq_frame: 30
  context_cosdist: 0.2
  min_loop_travel_ratio: 4.0
optimizer: # mappers
  iters: 10 
  batch_size: 8192
  adaptive_iters: True

which conducts fewer iterations for mapping and registration. It takes about 80ms on single A4000 GPU.

[zhSlamer]

Thank for your help! It looks better than before. I think i might have something to do with my GPU utilisation, which is always low. I will keep looking into it.