Run darknet show "CUDA Error: invalid argument: Resource temporarily unavailable". Solved

Hi,
I’m a new guy for darknet on jetson nano.
When I run darknet that will show below error, have anyone can share the experience?

CUDA-version: 10020 (10020), cuDNN: 8.0.0, GPU count: 1
OpenCV version: 4.0.0
0 : compute_capability = 530, cudnn_half = 0, GPU: NVIDIA Tegra X1
net.optimized_memory = 0
mini_batch = 1, batch = 64, time_steps = 1, train = 1
[yolo] params: iou loss: ciou (4), iou_norm: 0.07, cls_norm: 1.00, scale_x_y: 1.05
nms_kind: greedynms (1), beta = 0.600000
Total BFLOPS 12.688
avg_outputs = 104331
Loading weights from yolov4.conv.137…
seen 64, trained: 0 K-images (0 Kilo-batches_64)
Done! Loaded 137 layers from weights-file
Learning Rate: 0.001, Momentum: 0.949, Decay: 0.0005
Detection layer: 139 - type = 28
Detection layer: 150 - type = 28
Detection layer: 161 - type = 28
If error occurs - run training with flag: -dont_show
Resizing, random_coef = 1.40

288 x 288
Create 6 permanent cpu-threads
**CUDA status Error: file: ./src/dark_cuda.c : () : line: 454 : build time: Dec 3 2020 - 08:35:42 **

** CUDA Error: invalid argument**
CUDA Error: invalid argument: Resource temporarily unavailable
darknet: ./src/utils.c:326: error: Assertion `0’ failed.
Aborted (core dumped)

Here is my conditions:
JetPack 4.4.1
OpenCV 4.0.0 compiled CUDA: Yes
CUDA 10.2.89
cuDNN 8.0.0.180
Makefile
GPU=1
CUDNN=1
CUDNN_HALF=0
OPENCV=1
AVX=0
OPENMP=0
ARCH= -gencode arch=compute_53,code=[sm_53,compute_53]
NVCC=/usr/local/cuda-10.2/bin/

Any suggestion are welcome, thanks for your advice.

Hi,

It seems that you meet a similar issue of the below topic:
https://www.gitmemory.com/issue/AlexeyAB/darknet/6335/667886775

Could you try the comment to see if helps?

try random = 0 in your cfg file

Thanks.

Hi AastaLLL

It is worked.

Thank you so much.