The first question:
CUDA Error: OUT of Memory darknet: ./src/cuda.c:36: check_error: Assertion `0 'failed The. Abandoned (core dumped)
by the Error type may know the memory required to run the model space than the memory card supplied.
Solution:
Modify cfg configuration files, reducing the batch and subdivsion.
The second question:
CUDA Error: Unknown error darknet:. /Src/cuda.c: 36 : check_error: Assertion ` 0 ' failed The Aborted.
Occurrence of this problem, the Internet gives a lot of reasons, you can try the following solutions.
The reason this piece of my record I: I This is due to the nvidia kernel driver version and the PC version does not support the cause of unity.
How to determine that this is not the reason, the input terminal may be by nvidia-smi, shown below:
Failed to initialize NVML: driver/library version mismatch
Solutions can be found in the following blog and write well:
https://blog.csdn.net/zhang970187013/article/details/81012845
In addition, I support the 384 version is available for download at:
https://www.geforce.cn/drivers/results/133208