Can't run on cpu and ubuntu 18.04

[AdelAyman3]

I wanted replicate the results of this project so i wanted to ask if i can run it on cpu or i have to run it using gpu? And if i do run on cpu is there any changes i do on the code?

[traveller59]

I have added some code to evaluate function in train.py to support cpu inference. very simple. The problem is:

1. inference can run on cpu but eval code only support gpu. you can use pickle_result=False and use official evaluation program.
2. cpu inference is very slow. the inference speed of i7-8750H is 45x slower than 1060 in my notebook.

[lonlonago]

@traveller59 I only have CPU , I can not run inference on cpu because "python create_data.py kitti_data_prep " need spconv module , and install spconv need GPU , it's a dead loop. can you help solve the problem?

[lonlonago]

I run the code in win10

[Ganymedes-wang]

I run the code in win10

hello，have you successfully run the second/piontpillars code with CPU? thank!

[Ganymedes-wang]

@traveller59

I have added some code to evaluate function in train.py to support cpu inference. very simple. The problem is:

1. inference can run on cpu but eval code only support gpu. you can use pickle_result=False and use official evaluation program.
2. cpu inference is very slow. the inference speed of i7-8750H is 45x slower than 1060 in my notebook.

hello, Can I train the model on GPU and inference it on CPU?thank!