search

pjreddie / darknet

Convolutional Neural Networks
http://pjreddie.com/darknet/
Other
25.54k stars 21.33k forks source link

Try to set subdivisions=64 in your cfg-file. CUDA status Error: file: ..\..\src\dark_cuda.c : cuda_make_array() : line: 235 : build time: Aug 20 2019 - 12:05:40 CUDA Error: out of memory #1868

Open SouradipBh opened 4 years ago

SouradipBh commented 4 years ago

I am currently train darknet with yolov3 model.After train, when it comes to testing time, I can run only 3 instance at the parallaly.At the time of running 4th instance it gives me error (CUDA make array)

I am doing training on RTX 2080 8GB Ram.

Can anyone help me out to increase instance number of darknet?

vfbsilva commented 4 years ago

You are running out of memmory, try yolo v3 tiny or decrease the number of subdivisions, place your yolo.cfg here also. nvidia-smi can be used to check how much free memory you have avaliable.

SouradipBh commented 4 years ago

@vfbsilva Only 250 mb space is available after 3rd running instance.It consumps the total memory. I just want to give particular space for every instance.

GuessPan commented 4 years ago

Have u solve this problem？I got the same error，THX！

adavradou commented 4 years ago

Actually, as the error mentions, you should increase the subdivisions, e.g. to 64. As mentioned in the repo:

Note: if error Out of memory occurs then in .cfg-file you should increase subdivisions=16, 32 or 64: link

ghost commented 2 years ago

Actually, as the error mentions, you should increase the subdivisions, e.g. to 64. As mentioned in the repo:

Note: if error Out of memory occurs then in .cfg-file you should increase subdivisions=16, 32 or 64: link

This solution does not work anymore